CONTENTS

OVERVIEW Joseph J. Martocchio and Gerald R. Ferris THE IMPACT OF TEAM FLUIDITY AND ITS IMPLICATIONS FOR HUMAN RESOURCE MANAGEMENT RESEARCH AND PRACTICE Brian R. Dineen and Raymond A. Noe MENTORING RESEARCH: A REVIEW AND DYNAMIC PROCESS MODEL Connie R. Wanberg, Elizabeth T. Welsh and Sarah A. Hezlett THE IMPACT OF TELECOMMUTING DESIGN ON SOCIAL SYSTEMS, SELF-REGULATION, AND ROLE BOUNDARIES David G. Allen, Robert W. Renn and Rodger W. Griffeth RESEARCH ON EMPLOYEE CREATIVITY: A CRITICAL REVIEW AND DIRECTIONS FOR FUTURE RESEARCH Jing Zhou and Christina E. Shalley THE RELATIONSHIP BETWEEN WORK-FAMILY HUMAN RESOURCE PRACTICES AND FIRM PROFITABILITY: A MULTI-THEORETICAL PERSPECTIVE Michelle M. Arthur and Alison Cook TOWARD UNDERSTANDING AND MANAGING STEREOTYPICAL BELIEFS ABOUT OLDER WORKERS’ ABILITY AND DESIRE FOR LEARNING AND DEVELOPMENT Todd J. Maurer, Kimberly A. Wrenn and Elizabeth M. Weiss
v

vii

1

39

125

165

219

253

vi

CURRENT DIRECTIONS AND ISSUES IN PERSONNEL SELECTION AND CLASSIFICATION Walter C. Borman, Jerry W. Hedge, Kerri L. Ferstl, Jennifer D. Kaufman, William L. Farmer and Ronald M. Bearden APPLYING SELF-DETERMINATION THEORY TO ORGANIZATIONAL RESEARCH Kennon M. Sheldon, Daniel B. Turban, Kenneth G. Brown, Murray R. Barrick and Timothy A. Judge BEYOND SOCIAL EXCHANGE: NEW DIRECTIONS FOR ORGANIZATIONAL CITIZENSHIP BEHAVIOR THEORY AND RESEARCH Kelly L. Zellars and Bennett J. Tepper ABOUT THE AUTHORS

287

357

395 425

OVERVIEW

We are pleased to present nine papers on long-standing foundation and emerging topics in the human resources management field. All papers are uniquely innovative contributions to our knowledge base. The papers are also interesting from a social science standpoint because a number of commonly shared assumptions have been challenged somewhat. In other words, the set of papers provides conceptual insights that will leave us thinking about each topic a little bit differently. For the purposes of this overview, we have categorized the papers into two broad categories: model building and literature reviews that chart future directions for researchers. Model building blends knowledge of appropriate literatures with researchers’ insights about specific topics. The first paper focuses on team fluidity and its implications for HRM. Dineen and Noe point out that past research involving turnover in work teams has largely focused on turnover as a dependent variable. With the growing trend towards more fluid, project-based teams, the effects of team membership changes on team processes and outcomes are in need of theoretical development and systematic study. Building on previous work by others (e.g. Arrow & McGrath, 1995; Marks, Mathieu & Zacarro, 2001), Dineen and Noe develop a framework for understanding the effects of the rate of membership change, or team fluidity, on emergent states and processes in teams. They: (a) discuss the theoretical underpinnings of team fluidity; (b) review past team research involving turnover; (c) make theoretically-grounded propositions about the effects of team fluidity on emergent states and process variables, as well as additional propositions about boundary conditions; (d) discuss implications for human resource management practices; and (e) identify methodological challenges, including measurement issues, in studying team fluidity. Organizations have become increasingly interested in developing their human resources. One tool that has been explored in this quest is mentoring. As a result, there has been a surge in mentoring research, and an increase in the number of formal mentoring programs implemented in organizations. Wanberg, Welsh and Hezlett provide a survey of the empirical work on mentoring that is organized around the major questions that have been investigated. Then they present a conceptual model, focused on formal mentoring relationships, to help understand the mentoring process. The model draws upon research from a diverse body of literature, including interpersonal relationships, career success, training and
vii

viii

OVERVIEW

development, and informal mentoring. These authors also discussed various critical next steps for research in the mentoring domain. As more companies and employees become involved in telecommuting, researchers and managers will need to understand the effects of this relatively new working arrangement on the work perceptions and behaviors of the individual telecommuter. Allen, Renn and Griffeth point out that the extant empirical literature provides mixed results and is limited by a lack of theory. Consequently, neither researchers nor managers can rely on this literature for clear direction on how telecommuting will likely affect individual telecommuters. The authors present a multi-dimensional framework of telecommuting design, and focus on how telecommuting design might affect the telecommuter’s work environment and outcomes through its effects on the social system of the telecommuter, autonomy and self-management opportunities and requirements, and role boundaries, particularly in terms of the work and non-work interface. Their goal is to provide a framework to assist managers and researchers in systematically addressing questions of how to design telecommuting arrangements to maximize their potential benefits while minimizing their potential drawbacks. The examination of contextual factors that enhance or stifle employees’ creative performance is a new but rapidly growing research area. Theory and research in this area has focused on antecedents of employee creativity. Zhou and Shalley review and discuss the major theoretical frameworks that have served as conceptual foundations for empirical studies. These authors also provide a review and critical appraisal of pertinent empirical studies. They proposed interesting possibilities for future research directions, and addressed implications of this body of work for human resource management. Few studies have investigated the relationship between work-family human resource practices and firm-level outcomes. Several organizational studies have addressed the antecedents to firm adoption of work-family initiatives; however, the majority of work-family research investigates the relationship between workfamily practices and individual-level outcomes. The paper by Arthur and Cook offers a critical analysis and synthesis of the extant work-family literature. They also integrated some of the organizational learning research on firm commitment to work-family policies and the human resource model. They conjecture that the level of firm commitment moderates the relationship between work-family policies, the human resource model, and firm performance. Several propositions for future work-family research are presented. Maurer, Wrenn and Weiss present a model of stereotypical beliefs that older workers have difficulty learning and developing and are not motivated to learn. Three categories of antecedents of the stereotypical beliefs are addressed: Experience with stereotype-consistent behaviors and promulgation of the stereotype

Overview

ix

by others, perceived learning and development inhibitors internal to the older worker, and perceived learning and development inhibitors external to the older worker. Potential consequences of the stereotypical beliefs for older workers and employing organizations are also explored. The proposed model provides a framework to help guide future research on this topic and also some suggestions for managing a work place where these beliefs might exist. Switching focus somewhat, literature reviews help researchers to take stock of our extant knowledge base on particular topics. Well written literature reviews also give us pause to recognize the gaps in our knowledge and plot out directions for further research. Borman and his colleagues (Hedge, Ferstl, Kaufman, Farmer & Bearden) offer a contemporary view of state-of-the science research and thinking done in the areas of selection and classification. They take as a starting point the observation that the world of work is undergoing important changes that are likely to result in different occupational and organizational structures. In this context, they review recent research on criteria, especially models of job performance, followed by sections on predictors, including ability, personality, vocational interests, biodata, and situational judgment tests. Borman and his colleagues also discuss person-organization fit models, as alternatives or complements to the traditional person-job fit paradigm. Sheldon, Turban, Brown, Barrick and Judge argue that self-determination theory (SDT; Deci & Ryan, 2000) provides a useful conceptual tool for organizational researchers, one that complements traditional work motivation theories. Toward this end, the authors review SDT, showing that it has gone far beyond the “intrinsic versus extrinsic motivation” dichotomy with which it began. Then they show how the theory might be applied to better understand a variety of organizational phenomena, including the positive effects of transformational leadership, the nature of “true” goal-commitment, the determinants of employees’ motivation to learn, and the positive impact of certain human resource practices. The authors note that SDT might yield significant new understanding of work motivation, and suggest opportunities to refine the theory for research on work-related phenomena. Virtually all research on organizational citizenship behavior (OCB) is rooted in a social exchange based view of citizenship performance. Zellars and Tepper recognize the significant role exchange motives play in citizenship performance. However, they perceive what amounts to a preoccupation with and over-reliance on social exchange processes in extant OCB theory. Zellars and Tepper endorse the goals of improving the prediction of citizenship performance and advancing human resource management. Toward this end, they outline several new directions for OCB theory and research. In the closing section of this Overview, I wish to recognize Jerry Ferris. In the Volume 21 Overview, Jerry announced that he would retire from his role as

x

OVERVIEW

editor, and pass along the torch to me with the completion of this volume. I wish to congratulate Jerry on his outstanding contributions to RPHRM. Jerry Ferris and Ken Rowland founded this series more than 20 years ago; and after Ken retired in 1992, Jerry served as sole editor for the past ten years. Both Jerry and Ken possessed keen insight when they anticipated a need to offer HRM researchers an outlet for high-quality, monograph-length literature reviews and conceptual models. Now, more than twenty-two years later, the series continues to thrive. I have had the privilege to witness Jerry take pride in RPHRM and his exercise of exemplary standards. Thanks to Jerry, the series is on solid ground. I look forward to continuing his fine tradition. Best wishes, Jerry! Joseph J. Martocchio Gerald R. Ferris Series Editors

THE IMPACT OF TEAM FLUIDITY AND ITS IMPLICATIONS FOR HUMAN RESOURCE MANAGEMENT RESEARCH AND PRACTICE
Brian R. Dineen and Raymond A. Noe
ABSTRACT
Past research involving turnover in work teams has largely focused on turnover as a dependent variable. With the growing trend towards more fluid, project-based teams, the effects of team membership changes on team processes and outcomes are in need of theoretical development and systematic study. Building on previous work by others (e.g. Arrow & McGrath, 1995; Marks, Mathieu & Zacarro, 2001), we develop a framework for understanding the effects of the rate of membership change, or team fluidity, on emergent states and processes in teams. Specifically, we: (a) discuss the theoretical underpinnings of team fluidity; (b) review past team research involving turnover; (c) make theoretically-grounded propositions about the effects of team fluidity on emergent states and process variables as well as additional propositions about boundary conditions; (d) discuss implications for human resource management practices; and (e) identify methodological challenges, including measurement issues, in studying team fluidity.

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 1–37 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22001-6

1

2

BRIAN R. DINEEN AND RAYMOND A. NOE

INTRODUCTION
The growth of teams in organizations is an established and continuing characteristic of contemporary business enterprise (Guzzo & Shea, 1992; Lawler, Mohrman & Ledford, 1995). The academic literature on teams in organizations has been expansive, with notable reviews outlining the progress in understanding teams (e.g. Bettenhausen, 1991; Guzzo & Dickson, 1996; Guzzo & Shea, 1992; Milliken & Martins, 1996). As teams and organizations evolve, changes in the makeup of teams and team tasks continue. For example, teams must now manage a wider range of interdependencies, constituents, and social linkages (Ancona & Caldwell, 1998; Guzzo & Dickson, 1996). Also, the fluidity of teams has increased, with members rotating in and out on a “project” or “as-needed” basis (Arrow & McGrath, 1995; Campion, Papper & Medsker, 1996; Townsend, DeMarie & Hendrickson, 1998). Two developments that have led to increased fluidity are the rise in contingent work arrangements (e.g. contract work) and a labor market that allows skilled employees to shop their services among organizations to obtain more desirable working arrangements (Muoio, 2000). Virtual teams are especially likely to be fluid in the sense of rotating membership and participation (Guzzo & Dickson, 1996; Kristof, Brown, Sims & Smith, 1995; Saunders & Ahuja, 2000). In fact, some researchers assert that a lack of fluidity in teams can be detrimental to team outcomes (e.g. Guzzo & Dickson, 1996). The present paper attempts to address some of these issues by examining the influence of team fluidity on team processes, emergent states and outcomes. Team fluidity is defined as the rate of change in team membership over time. Although progress in research on teams has been unmistakable, this research has largely overlooked the potential effects of fluidity on team processes and performance. For example, most work has relied on teams composed of the same members throughout the period of study, with no changes in team membership (e.g. Gersick, 1988; Harrison, Price & Bell, 1998; Watson, Kumar & Michaelsen, 1993). As Arrow and McGrath (1993) note, other researchers have controlled for or otherwise eliminated “participant mortality,” treating it as a problematic source of variation in team studies. The result of treating participant mortality as error variance in teams research is that most of what we know about teams is based on a static model of team membership. We lack an understanding of how changing membership affects team processes, emergent states, and outcomes. A number of researchers, led by Arrow, McGrath and associates (1991, 1993, 1995, 2000) have made strides towards addressing the issue of membership dynamics in teams. Specifically, these researchers have begun to study membership history and change, answering calls by others who have previously raised the issue (e.g. Guzzo & Shea, 1992; Smith, Smith, Olian, Sims, O’Bannon &

The Impact of Team Fluidity and Its Implications

3

Scully, 1994; Ziller, 1965). In doing so, they have called into question wellaccepted models of team stage development which rely on stable membership over time (e.g. Tuckman, 1965). Also, they have begun to differentiate between effects of voluntary versus involuntary membership change (Arrow & McGrath, 1995; Gruenfeld, Martorana & Fan, 2000). Most of these studies, however, have only looked at outcome variables such as team performance, while excluding emergent states such as team flexibility or process variables such as communication or conflict (e.g. Argote, Insko, Yovetich & Romero, 1995; Goodman & Leyden, 1991). Others (e.g. O’Connor, Gruenfeld & McGrath, 1993) have examined process variables by manipulating team membership change only at a single point in time instead of measuring change over time, or rate of member change. This paper contributes to the team literature by proposing how team fluidity affects team processes as well as team emergent states. Specifically, we focus broadly on: (a) the theoretical underpinnings of team fluidity; (b) past team research involving turnover; (c) theoretically-grounded propositions about the effects of team fluidity on certain process variables and emergent states as well as additional propositions about boundary condition effects; (d) implications of team fluidity for human resource management (HRM) practices; and (e) a discussion of some methodological challenges and future research directions. While recognizing the distinction often drawn by researchers, we follow Guzzo and Shea (1992) in treating the terms “group” and “team” interchangeably for purposes of this work.

THEORETICAL BACKGROUND
Several theoretical perspectives are relevant to understanding how changing team membership affects team processes and emergent states. First, Arrow and McGrath’s (1995) membership dynamics framework provides several general predictions about the nature and effects of group membership changes. Although not addressing relationships between membership change and specific process variables, they do differentiate between standing and acting groups (i.e. the entire group versus the part of the group actually present at a given point in time), types of work groups (i.e. task forces, teams, or crews), and outcome variables such as the well-being, support, and production of members and groups. They theorize that direction and magnitude of member change, locus of initiation of change, and the temporal patterning of change will differentially affect outcomes. We focus specifically on ongoing teams that experience periodic changes in team member makeup. Team process variables are of interest to us, in line with traditional input-process-output models of team performance (e.g. Guzzo & Shea, 1992; Hackman, 1987). In addition, consistent with recent work by Marks, Mathieu and

4

BRIAN R. DINEEN AND RAYMOND A. NOE

Zaccaro (2001), we examine team emergent states and how they might be affected by team fluidity. Ancona and Caldwell (1998) adopted a theoretical stance that is similar to that of Arrow and McGrath (1995) in their conceptual article pertaining to team composition. Specifically, they pose three questions about team composition pertaining to different potential outcomes. These questions address whether members are assigned to a team for its entire life or for only part of a project, on a part- or full-time basis, and whether the team is comprised of everyone necessary to make decisions or whether the team relies on outside “experts” to facilitate these processes. Their basic recommendation, although not empirically substantiated, is to have a team comprised of a “core” set of unchanging members, with a “periphery” of rotating team members. Both the similarity/attraction (Byrne, 1971) and social identity (Tajfel & Turner, 1986) theories are also relevant to discussions of team fluidity. Potential benefits as well as drawbacks of similarity in terms of team membership are likely to manifest themselves in more stable teams, whereas more fluid team members may have a difficult time feeling a sense of identity or attraction based on similarity. That is, ingroup/outgroup boundaries (Tajfel, 1982) may be harder to ascertain in more fluid teams. This becomes even more difficult when trying to become familiar with new team members who come from outside the organization compared to those who migrate from within another part of the organization. Sociotechnical theory (e.g. Rousseau, 1977; Trist, 1981) also lends itself to the present discussion. Sociotechnical theory posits that optimization of both the social and technical sides of the work environment leads to high performance and positive social experiences at work (Guzzo & Shea, 1992). That is, an organization must account for both the technical and social sides of work when designing work interventions, and attempts to optimize one at the expense of the other will result in a decrement to the whole (Trist, 1981). It is plausible that the stability of work groups may influence the social side of work (e.g. cohesiveness), while having direct effects on the technical side of work as well (e.g. flexibility, creativity). Another important theoretical perspective that is germane to our discussion of team fluidity is a social networks perspective. Because of its vast scope, we do not intend to conduct a thorough review of the social networks literature, but rather we will touch on how a social networks perspective might introduce important boundary conditions in developing propositions about the effects of team fluidity on team process variables. At its core, a social networks perspective suggests that process and performance in groups is at least partially associated with the patterns of relationships that exist in the group (Brass, 1995; Granovetter, 1973). For example, some groups are more tightly linked and exhibit a high degree of closeness centrality (Freeman, 1979), whereas others experience looser ties among

The Impact of Team Fluidity and Its Implications

5

members. Other groups have a greater mix of central members with several links to other members, as well as members who are more peripheral and have fewer and/or weaker ties to fellow group members. Importantly, as Brass (1995) notes, “If any aspect of the network changes, the actor’s relationship within the network also changes.” (p. 44). Thus, there are clear implications of team fluidity on the overall social network that exists in the team. We specifically address the expected differences in effects when a more centrally close versus peripheral member of the team leaves or is replaced. Finally, regarding the discussion of team process variables and how team fluidity may affect such variables, Marks et al. (2001) have developed a temporally based framework and taxonomy of team processes. Essentially, these scholars discriminate between team processes and emergent states. Emergent states are “properties of the team that are typically dynamic in nature and vary as a function of team context, inputs, processes, and outcomes” (p. 357). Team cohesiveness is a good example of an emergent state that evolves, or emerges over time, but is not an action the team engages in. Alternatively, team processes are defined by Marks et al. (2001) as “. . . acts that convert inputs to outcomes through cognitive, verbal, and behavioral activities directed toward organizing task work to achieve collective goals” (p. 357). For example, team communication is a team process because it represents an action that a team engages in. Marks et al. (2001) further differentiate between various types of team process variables. Specifically, they discuss transition, action, and interpersonal processes. Transition processes are posited to occur when a team is in transition between tasks. For example, goal specification and planning represent processes engaged in at this stage. In more fluid teams, it might be the case that transition processes occur more frequently than in more stable teams as teams realign themselves around new members and need to reorient themselves to new or ongoing tasks. Action processes occur as a task is accomplished and include such activities as monitoring and coordination. Finally, interpersonal processes occur across both transition and action phases of team cycles and include actions such as conflict management, motivation and confidence building, and affect management.

APPROACHES TO THE STUDY OF TURNOVER IN TEAMS
Turnover in teams is a frequently studied topic in the organizational literature (e.g. Milliken & Martins, 1996; O’Reilly, Caldwell & Barnett, 1989; Wiersema & Bird, 1993) with myriad antecedents identified. The effects of turnover as an independent variable, however, are studied less frequently. It is important to differentiate fluidity

6

BRIAN R. DINEEN AND RAYMOND A. NOE

from more traditional indices of turnover. Whereas turnover traditionally accounts only for those members who permanently leave the team, fluidity subsumes not only these individuals, but also individuals who may be absent from an acting team for a period of time, only to return to the team at a later date. This point is relevant to the distinction drawn by Arrow and McGrath (1995) between standing and acting groups. The standing group is composed of all members “who share an explicit and ongoing set of relations both with the group and with other members” whereas the acting group “consists of all persons involved in a particular work session or other group interaction” (p. 377). For example, if a member is still a part of the team, but only attends weekly meetings every other week, there is fluidity in the team. However, this situation would not be reflected in traditional turnover indices. Research on the effects of turnover as it has traditionally been assessed in teams is mixed. Some research suggests that continually bringing new members into a group can be disruptive and lower performance (e.g. Ancona & Caldwell, 1998; Argote et al., 1995). Less familiarity of members, a by-product of turnover, has been shown to be related to decreased productivity (see Goodman & Leyden, 1991). Goodman and Leyden (1991) also posit that this effect would likely be enhanced for more complex work. Finally, Hollenbeck, Ilgen, Sego, Hedlund, Major and Phillips (1995) concluded that instability in decision-making team membership disrupts a leader’s ability to develop effective weighing schemes (i.e. dyadic sensitivity) for each member’s input to the decision. This effect is stronger when attrition takes place in teams that are already highly familiar, supporting the proposition made by Arrow and McGrath (1995) that effects of membership change would be stronger in groups with greater prior membership continuity. However, other researchers believe that turnover has a negligible or even a positive influence on team process and outcomes. Guzzo and Dickson (1996) state that, “turnover is usually thought of as dysfunctional for team effectiveness, though it is possible that the consequences of losing and replacing members could work to the advantage of teams in some circumstances” (p. 312). Similarly, Arrow and McGrath (1995) note that neither continuity nor changes in membership can necessarily be considered desirable or undesirable. Campion et al. (1996) found that team member permanence was not significantly related to process characteristics or effectiveness. In her study of turnover’s effects on organizational learning, Carley (1992) also found inconclusive results, and suggested differences in organizational structure as a possible explanation. For example, she suggested that turnover would exhibit different effects in a hierarchical versus team-based organizational structure. Hinsz, Tindale and Vollrath (1997) suggest that group learning may not be dependent on specific group members. That is, group learning may still occur in the presence of member turnover insofar as it is likely that turnover brings more perspectives to bear on a task or problem-solving situation.

The Impact of Team Fluidity and Its Implications

7

There is little continuity in findings that attempt to relate turnover in teams to team outcomes such as productivity, effectiveness, or overall performance. To systematically study the effects of team fluidity, researchers need to open the “black box” and look more closely at process issues and emergent states instead of focusing directly on performance outcomes. The propositions presented below provide the basis for researchers to empirically study the influence of team fluidity on team processes and emergent states.

A DYNAMIC APPROACH: TEAM FLUIDITY
A framework for understanding team fluidity is presented in Fig. 1. As shown in the model, we propose that team fluidity has effects on emergent team states, as well as transition, action, and interpersonal processes in teams. In line with the plethora of prior research linking process variables to performance (for reviews, see Guzzo & Dickson, 1996; Guzzo & Shea, 1992), we include performance outcomes that reflect production, well-being, and support (Arrow & McGrath, 1995). However, we make no specific predictions regarding the cumulative effects of process variables and emergent states on performance due to the differential effects those variables are likely to have. Furthermore, we do not include specific production outcomes, but instead recognize that the importance of various outcome indices (e.g. quality, efficiency, net profits) will vary by organization or team. An important part of the model is the feedback loop from processes, emergent states, and performance outcomes to team fluidity. This illustrates the dynamic nature of the model and recognizes previous research that has linked a number of these outcomes to subsequent turnover (for a review, see Griffeth & Hom, 1995). Temporal aspects of team dynamics have received increased attention from researchers (e.g. McGrath, 1991; Saunders & Ahuja, 2000). However, most prior studies that have looked at turnover’s effects on teams have done so by focusing on researcher-induced member change at a single point in time (e.g. Arrow & McGrath, 1993; Gruenfeld et al., 2000; Schopler & Galinsky, 1990). These studies, although providing important contributions, only examine turnover dichotomously (i.e. teams either have or do not have an instance of turnover). Ziller, Behringer and Jansen (1961) were among the first to suggest the importance of rate of member change in their work on open versus closed groups. More recently, Arrow and McGrath (1995) recognized the need in proposing that both magnitude and direction of membership change matter and that the impact of change will depend on frequency of change. Further, as Arrow, McGrath and Berdahl (2000) write, “a group’s ability to adapt to change will be affected by the rate and frequency of change . . .” (p. 197).

8 BRIAN R. DINEEN AND RAYMOND A. NOE

Fig. 1. Relationships Between Team Fluidity, Team Emergent States and Processes, and Performance.

The Impact of Team Fluidity and Its Implications

9

RELATIONSHIPS BETWEEN TEAM FLUIDITY AND EMERGENT STATES AND PROCESS VARIABLES
Figure 1 shows the relationship between team fluidity and emergent team states as well as transitional, action, and interpersonal processes. Although not necessarily comprehensive in terms of all possible states and/or processes, our discussion is meant to be illustrative of some of the factors thought to be essential in considering the effects of team fluidity. That is, our purpose is not to provide a comprehensive treatment of all possible emergent states and transitional, action, and interpersonal process variables and how they are affected by team fluidity. Rather, we introduce the reader to the concept of team fluidity and provide a sampling of proposed effects from these various categories.

Team Emergent States Emergent states in teams are thought to evolve as a function of team context, inputs, processes, and outcomes. Thus, they rely on some type of prior interaction among team members. Social networks (Brass, 1995; Granovetter, 1973) and symbolic interactionist approaches (Joyce & Slocum, 1984) might be used to discuss how various states in teams emerge. As shown in Fig. 1, the fluidity of a team is likely to affect the development and maintenance of various states in those teams. Among the states that might be affected, Marks et al. (2001) mention collective efficacy, situational awareness, and cohesiveness. We discuss the probable effects of team fluidity on two of these suggested states, as well as a third emergent state, team flexibility. Collective Efficacy Collective efficacy is defined as the group’s shared belief in its ability to carry out courses of action required to attain given levels of performance (e.g. Bandura, 1997; Chen & Bliese, 2002). A primary antecedent of collective efficacy is team history (Guzzo & Shea, 1992; Peterson, Mitchell, Thompson & Burr, 2000). Knowledge of past success is likely to increase a team’s sense of efficacy for future events. A key element of this knowledge is familiarity among team members. Moreover, success is framed in terms of who was present during that success. Also, collective efficacy is framed in terms of current team makeup. In other words, a team feels efficacious or non-efficacious based on its current member makeup. There is inherent uncertainty surrounding introduction of new team members in terms of their personalities, performance potential, and fit with teammates, and the team’s

10

BRIAN R. DINEEN AND RAYMOND A. NOE

collective sense of identity is likely to be lessened. Furthermore, the makeup of the team and longevity of teammates is uncertain. For example, a team may have a “star player,” but not know how long she will stay given a high rate of turnover in the team. This uncertainty, characterized by high team fluidity, is therefore likely to reduce a team’s collective efficacy. Ziller (1965) proposed an important concept related to collective efficacy, which he termed the New Year’s Eve phenomenon. The concept refers to people’s tendency to view New Year’s as a “fresh start,” and experience reinvigoration towards goals, especially given a less-than-successful year gone by. Basically, if one has had a successful past, then change will lower collective efficacy and stability will increase it. On the other hand, change might increase efficacy for future success if the past has been less than successful, whereas stability might decrease it. For example, a baseball team may have knowledge of a prior World Championship, but may not experience collective efficacy for future championships since most of the players have switched teams. Or a baseball team that is coming off a disappointing season may experience increased collective efficacy if a number of new players have been brought onboard for the new season. This phenomenon, while compelling, was explained in the context of an isolated change in membership, with a clear past and future frame (e.g. last baseball season, the next season). It is unclear whether or not it would manifest in more fluid teams, and in fact, Campion et al. (1996) found a positive but non-significant (0.14) correlation between team member permanence and team potency (a construct similar to collective efficacy). However, on the whole, it seems as if the lack of knowledge of newer teammates would stunt any increase in collective efficacy, and most likely reduce it. Proposition 1. Team fluidity is negatively related to collective efficacy in a team. Team Flexibility Team flexibility refers to the ability of team members to perform tasks interchangeably. In more flexible teams, team members can more easily substitute for each other (Campion et al., 1996), and better adapt to a changing environment. Many contemporary work arrangements call for greater flexibility to meet increased performance and efficiency demands (Townsend et al., 1998). Oftentimes managers cite increased flexibility as a reason for rotating team members among teams. It seems intuitive that a team more used to changes in membership will be more flexible and open to change in general. Ziller (1965) recognized this, stating, “strategic membership changes can be made with greater ease in groups which experience membership changes routinely” (p. 175). Moreover, the benefits

The Impact of Team Fluidity and Its Implications

11

of rotating members among different teams are well recognized by knowledge transfer researchers (e.g. Argote, Ingram, Levine & Moreland, 2000). However, the potential for team members to interchangeably perform tasks is likely to suffer as team fluidity increases. For example, Campion et al. (1996) found that teams with more permanent members tended to be more flexible. This makes sense when one considers that more familiar team members have a chance to cross-train and learn each other’s tasks, whereas team members who are relatively new or expect to turn over more quickly are less likely to learn teammates’ tasks. Furthermore, team members are likely to adapt more easily to changing circumstances in the environment if they know how teammates are likely to react. Such knowledge grows out of stability and familiarity among members. The concept of shared mental models has been developed to describe this familiarity amongst team members that is thought to promote predictability and coordination (e.g. Klimoski & Mohammed, 1994; Mathieu, Heffner, Goodwin, Salas & Cannon-Bowers, 2000). In addition, sociotechnical theory suggests that the technical and social sides of work must be aligned for maximum performance (Trist, 1981). However, it appears that increased fluidity is likely to diminish both technical (e.g. flexibility) and social conditions within the team. Proposition 2. Team fluidity is negatively related to team flexibility. Cohesiveness Team cohesiveness is another emergent state that is likely affected by team fluidity. Although often pitted as the counterpart of conflict, cohesiveness and conflict are distinct (Pelled, 1996). Through stable team membership, team members tend to develop similar schemata based on similar past events and experiences. Such similarity should enhance cohesiveness among team members (Michel & Hambrick, 1992). This follows directly from the similarity/attraction paradigm (Byrne, 1971). In their discussion of team composition, Ancona and Caldwell (1998) note that underbounded teams (i.e. teams without stable boundaries) may have trouble developing cohesiveness. Gruenfeld, Mannix, Williams and Neale (1996) reviewed research suggesting that teammates that are more familiar with one another are more likely to exhibit higher levels of cohesiveness. Furthermore, a number of studies have examined a proxy for familiarity and stability, team tenure. For example, O’Reilly et al. (1989) found that heterogeneity in team tenure was negatively related to team cohesiveness. However, Smith et al. (1994) found no association between team tenure and social integration, and Riordan and Shore (1997) found no relationship between similarity in tenure and cohesiveness. It should be noted, though, that tenure was measured as time in present position

12

BRIAN R. DINEEN AND RAYMOND A. NOE

(Smith et al., 1994) and time in organization (Riordan & Shore, 1997) rather than time in the team in these studies, possibly explaining the discrepant results. Proposition 3. Team fluidity is negatively related to team cohesiveness.

Transitional Processes Transitional processes are posited to occur when a team is in transition between tasks. Marks et al. (2001) list mission analysis formulation and planning, goal specification, and strategy formulation as key transitional processes. It can be argued that a common thread running through these various transitional processes is a need for creativity in the team. For example, in formulating strategy, creativity may enhance the team’s chances of realizing the best possible approach. Team Creativity Research generally supports the idea that creativity in teams evolves from an availability of a variety of perspectives among team members. For example, the team diversity literature generally acclaims diversity as leading to greater creativity (e.g. Nemeth, 1986), although it also tends to lead to greater conflict (Pelled, Eisenhardt & Xin, 1999). Beyond the effects of diversity, however, team fluidity adds another important factor that may enhance creativity. New members entering a team might not only be diverse, but more importantly are likely to arrive from a former organizational or team situation, with its related ideas. This gives the team a “boundary spanning” advantage. That is, the team gains not only a new team member, but also that person’s past experiential knowledge (Ancona & Caldwell, 1998). Thus, the team benefits not only from the new member, but also from all of the prior experiences and interactions of that member as well. One of the earliest studies that linked changes in team membership to creativity grew out of Ziller and associates’ (1962, 1965) open versus closed group theory. Specifically, Ziller, Behringer and Goodchilds (1962) found that groups experiencing membership changes were more creative than stable groups. However, this study included only one membership change per group, and thus was not concerned with differences in rate of member change. Similarly, Stein (1982) reviewed literature that suggested that older more established groups were lower in creativity. However, it should be emphasized that team tenure level only serves as a proxy for team fluidity. The literature suggests that team members tend to be on “better behavior” and more inhibited in the presence of strangers (Gruenfeld et al., 2000; O’Connor et al., 1993; Shah & Jehn, 1993). That is, people may not be as comfortable expressing

The Impact of Team Fluidity and Its Implications

13

disagreement either when they are relatively new to a team, or when they relate to other members who are relatively new (Gruenfeld et al., 1996). In fact, a sense of “false cohesiveness” may prevail, whereby members go out of their way to maintain good relations while getting to know one another (Longley & Pruitt, 1980). Social facilitation or evaluation apprehension dynamics also may underlie a tendency towards inhibition in teams consisting of relatively unfamiliar members (Zajonc, 1965). However, the tendency towards greater evaluation apprehension or social facilitation effects (Gruenfeld et al., 1996; Zajonc, 1965) in the presence of strangers is more likely to result in the face of conflict rather than creativity. Although similar effects could inhibit creativity in relatively unfamiliar teams, this is unlikely because creativity carries more of a positive, acceptable meaning than conflict, and therefore should be more accepted by team members. For example, a newcomer advancing a novel pattern of ideas or thoughts that increases overall team creativity is much less likely to meet resistance than a newcomer who directly instigates conflict with currently existing ideas in the team. Therefore, evaluation apprehension effects should be minimal in the case of creative contributions. Proposition 4. Team fluidity is positively related to team creativity.

Action Processes Action processes occur as a task is accomplished and include such activities as monitoring and coordination. To effectively monitor and coordinate activities, it is likely that communication with internal and external sources is of prime importance to teams. Internal Communication Communication has been defined as “the transfer of information, ideas, understanding, or feelings” (cf. Pelled, 1996, p. 620). Communication is generally recognized as a precursor to effectiveness in teams (e.g. Katz, 1982), although some suggest that highly effective teams require less communication because members can “anticipate” each other or share mental models (Klimoski & Mohammed, 1994). Several studies have examined internal communication in the context of team stability. For example, Campion et al. (1996) failed to find a relationship between a measure of team member permanence and internal team communication. Mathieu et al. (2000) found that shared team- and task-based mental models related positively to team process, operationalized in terms of strategy formulation and coordination, cooperation, and internal communication. Katz (1982) showed

14

BRIAN R. DINEEN AND RAYMOND A. NOE

that teams with longer average tenure exhibited higher levels of communication with internal members. Finally, Zenger and Lawrence (1989) demonstrated that shared experience in working together makes communication easier among team members. These findings likely stem from either a familiarity or lack of familiarity with team member communication habits. Such habits are likely more difficult to ascertain when team membership is more fluid (Hightower, Sayeed, Warkentin & McHaney, 1998).

External Communication Researchers also are recognizing the importance of boundary spanning, or interteam relationships and communication (e.g. Ancona & Caldwell, 1998). External communication refers to communication that occurs with outside teams or constituents. Teams with a higher level of fluidity likely consist of members who bring more outside linkages to the team. According to the social networks literature (e.g. Brass, 1995; Granovetter, 1982), this might result in a situation whereby the team gains “strength through weak ties.” That is, external constituents are less likely to know one another, creating a low density, high diversity network that is rich in non-redundant information. Although few studies have specifically addressed this issue, Katz (1982) showed that teams with lower average tenure tended to communicate more with outside constituents. Based on work reviewed above in the areas of internal and external communication, we posit: Proposition 5a. Team fluidity is negatively related to internal team communication level. Proposition 5b. Team fluidity is positively related to external team communication level.

Interpersonal Processes Finally, interpersonal processes occur across both transition and action phases of team cycles and include actions such as conflict management, motivation and confidence building, and affect management. We specifically address the first of these as an example of the potential effects of team fluidity. Task Conflict Conflict in teams has generally been treated as containing task – and relationshiprelated elements (e.g. Pelled, 1996; Simons & Peterson, 2000). Recently,

The Impact of Team Fluidity and Its Implications

15

Jehn (1997) differentiated between task content conflict (i.e. what to do) and task process conflict (i.e. how to do it); however we develop propositions related to the more general two-factor taxonomy. First, task conflict arises when team members disagree about the nature and process of accomplishing tasks (Pelled, 1996). For example, a team of software designers may disagree about whether or not to add a specific feature to a software package, and if so, how to go about doing it. Generally, such conflict is linked to higher performance because members are forced to consider more alternatives and think through those alternatives more thoroughly (e.g. Pelled et al., 1999). An antecedent of task conflict that might derive from team fluidity is informational diversity, or the diversity in viewpoints and ideas that exist within a team (Jehn, Northcraft & Neale, 1999). Also, O’Connor et al. (1993) suggest that increased stress may result from membership change, leading to greater conflict. They do not differentiate between task and relationship conflict, however, in making this argument. Some studies have found that team member changes might lead to less task conflict. For example, we earlier reviewed studies showing that team members tend to be on “better behavior” and more inhibited in the presence of strangers (e.g. Gruenfeld et al., 2000; O’Connor et al., 1993). That is, people may not be as comfortable expressing disagreement either when they are relatively new to a team, or when they relate to other members who are relatively new (Gruenfeld et al., 1996). This sense of “false cohesiveness,” whereby members go out of their way to maintain good relations while getting to know one another (Longley & Pruitt, 1980), may complement social facilitation or evaluation apprehension dynamics that also may underlie a tendency towards inhibition in teams of relatively unfamiliar members (Zajonc, 1965). The degree to which this effect manifests will likely depend on when task conflict is measured, and we address measurement issues later in the paper. However, given the fact that more fluid teams will generally consist of individuals who are newer to the team, the effect is more likely to occur in this type of team than in a more stable team. Arguments linking greater informational diversity to greater task conflict are logical, but contradicted by evidence that informational diversity may not translate into conflict over use of that information in less familiar teams. That is, the tendency for newcomers holding diverse information to be on “better behavior” and a desire to “fit in” should lead to less task conflict. Recent work by Noe, Colquitt, Simmering and Alvarez (2003) might shed some further light on this apparent controversy. In discussing the creation of intellectual capital in teams, Noe et al. suggest constructive controversy (i.e. task conflict) as an important element of intellectual capital creation. Further, in discussing the likely effects of team design characteristics on intellectual capital creation, they suggest that a moderate level of instability is likely to be optimal in fostering creative

16

BRIAN R. DINEEN AND RAYMOND A. NOE

controversy. Too much stability may lead to stagnation and tendencies towards groupthink (Janis, 1972), whereas too much instability creates a constant stream of newcomers who are likely to exhibit evaluation apprehension early on and shy away from engaging in conflict with other team members regarding the task. This suggests a curvilinear relationship between fluidity and task conflict. Proposition 6. The relationship between team fluidity and task conflict is curvilinear, such that task conflict is greatest when there is a moderate level of fluidity. Relationship Conflict Relationship conflict arises as a result of disagreements over non-task related issues. Generally, such conflict is detrimental to team functioning (e.g. Jehn et al., 1999). Further, Meyerson, Weick and Kramer (1996) note that temporary teams, which are more similar to highly fluid teams than stable teams, rarely exhibit dysfunctional team dynamics (e.g. relationship conflict) since they do not have the time to do so. Arrow and her colleagues (2000) also recognize that members of less stable teams know that they probably will only be working together for a short time, and thus will avoid bothering with relationship conflict. Also, the evaluation apprehension and social facilitation arguments advanced in the preceding section may deter relationship conflict. Despite arguments that members of changing teams may not bother with relationship conflict, Arrow et al. (2000) also recognize that high rates of member turnover translate into a situation where team members are “out of sync” with each other in terms of development stage. This, in turn, might create relationship conflict as members interact using different frames of reference. For example, a more tenured member may become impatient with a newcomer who is trying to become familiar with team norms. Supporting this suggestion, Pelled et al. (1999) found that tenure diversity was positively associated with emotional conflict. Also, whereas it might be true that members will not bother with conflict if they expect their or their teammates’ tenure to be temporary, the very opposite could also be true. That is, a person may express relationship conflict knowing that they will not have to work with the others on the team for very long. Furthermore, heterogeneity in values has been linked to relationship conflict (e.g. Jehn et al., 1999). In line with ingroup/outgroup distinctions (Tajfel, 1982), it is likely that a more stable team (i.e. more familiar members) will experience more of a “melding of common values” over time, whereas a more fluid team will continue to experience changing value structures. Thus, more relationship conflict is likely to manifest. Finally, according to the similarity/attraction paradigm (Byrne, 1971), such differences may make teammates appear more unattractive, furthering the potential for relationship conflict to develop.

The Impact of Team Fluidity and Its Implications

17

Proposition 7. Team fluidity is positively related to relationship conflict in a team.

Identification of Potential Boundary Conditions Task interdependence and team size are potential boundary conditions of some of the relationships proposed in the model shown in Fig. 1. Task interdependence and team size are likely to affect the relationship between fluidity and team processes because they each relate to how much interaction takes place among individual members. In a large team, for example, interactions likely involve subsets of the team instead of the entire team. Task Interdependence Task interdependence refers to the degree of interaction and cooperation required among team members to accomplish their tasks. Ziller (1965) is among the researchers who suggest that the optimum rate of team membership change might depend on task demands. We propose that higher task interdependence will accentuate the relationship between team fluidity and three of the team process/emergent state variables described earlier. Team collective efficacy is highly dependent on knowledge of how team members are likely to perform (Guzzo & Shea, 1992). Above we described how this knowledge base is reduced in a more fluid team. Such a reduction most likely leads to a decrease in collective efficacy for more fluid teams. In more task interdependent teams, such an effect is likely to be more profound in that knowledge of how team members are likely to perform is more critical. This makes a lack of such knowledge equally critical, and is likely to lead to an accentuated decrement in collective efficacy. Thus, whereas we predict a negative relationship between fluidity and collective efficacy in general, teams that are more task interdependent are likely to exhibit a stronger negative relationship between fluidity and collective efficacy in particular. Proposition 8a. The negative relationship between team fluidity and collective efficacy will be moderated by task interdependence, such that the relationship will be stronger in more task interdependent teams. The curvilinear relationship between team fluidity and task conflict is also likely to be accentuated in a more task interdependent team. For example, the inhibition felt by new team members (Gruenfeld et al., 2000) is likely to be greater if forced to work closely with unfamiliar team members. Similarly, the “false cohesiveness” effect whereby members go out of their way to maintain good relations (Longley &

18

BRIAN R. DINEEN AND RAYMOND A. NOE

Pruitt, 1980) may be greater when members know they must work more closely on tasks. On the other hand, at higher levels of task interdependence, the informational diversity accompanying moderate levels of fluidity should enhance task conflict to an even greater degree. Proposition 8b. The curvilinear relationship between team fluidity and task conflict will be moderated by task interdependence, such that the relationship will be stronger in more task interdependent teams. The positive relationship between team fluidity and relationship conflict is also likely to be accentuated in more task interdependent teams. Relationship conflict often grows out of values differences (Jehn et al., 1999) and team members being “out of sync” developmentally with one another (Arrow et al., 2000). In a more task interdependent team, such differences are easier to uncover and more likely to manifest as members interact more frequently in performing tasks. Thus, all teams should exhibit a positive relationship between fluidity and relationship conflict. However, the fluidity-relationship conflict association should be greater in more task interdependent teams. Proposition 8c. The positive relationship between team fluidity and relationship conflict will be moderated by task interdependence, such that the relationship will be stronger in more task interdependent teams. Task interdependence is likely to attenuate the relationships between team fluidity and three other process variables/emergent states. First, task interdependence is likely to attenuate the negative relationship between team fluidity and team cohesiveness. We proposed that team fluidity would be negatively associated with cohesiveness based on the premise that changing teams would have members who were not as familiar with each other as in more stable teams. In teams performing a highly interdependent task, however, familiarity should be facilitated more quickly, easing the negative effects of changing membership. In fact, there potentially could be greater cohesiveness on a more fluid team performing a highly interdependent task than on a stable, yet relatively independent team. Thus, whereas we predict a negative relationship between fluidity and cohesiveness for all teams, we predict that the fluidity-cohesiveness relationship will be less negative for teams performing highly interdependent tasks. Proposition 8d. The negative relationship between team fluidity and team cohesiveness will be moderated by task interdependence, such that the relationship will be weaker in more task interdependent teams. Similarly, the negative relationship between team fluidity and team flexibility should be attenuated in more task interdependent teams. Although rapidly

The Impact of Team Fluidity and Its Implications

19

changing team membership precludes cross-training and skill sharing among teammates, task interdependence more readily exposes teammates to one another’s tasks and skill-sets. This likely facilitates cross-fertilization of skills and subsequent flexibility despite changing membership. Whereas changing membership is likely to diminish such cross-fertilization, increased task interdependence will counteract this effect, making the fluidity-flexibility relationship less negative. Proposition 8e. The negative relationship between team fluidity and team flexibility will be moderated by task interdependence, such that the relationship will be weaker in more task interdependent teams. Finally, the negative relationship between team fluidity and internal communication also is likely to be attenuated in more task interdependent teams. Greater interdependence is likely to lead to greater information exchange as a necessity for accomplishing tasks in close proximity. While it is certainly true that changing membership is likely to impair the formation of “mental models” or anticipatory reactions among more familiar teammates (Klimoski & Mohammed, 1994), high task interdependence should facilitate information transfer even among relative strangers, as the task necessitates increased intrateam communication. Thus, the fluidity-internal team communications relationship should be similar to the previous two relationships described above (e.g. fluidity-flexibility and fluidity-cohesiveness). Proposition 8f. The negative relationship between team fluidity and internal team communication will be moderated by task interdependence such that the relationship will be weaker in more task interdependent teams. We make no specific propositions regarding the last two process variables, external communication and team creativity. Such variables are largely orthogonal to task interdependence. Team Size When studying team fluidity’s effects on team process and emergent state variables, a logical boundary condition to consider is team size. Specifically, the relationships between team fluidity and team process and emergent state variables are likely to be attenuated in larger teams. Colquitt, Noe and Jackson (2002) suggest that this results from decreased psychological bond strength between team members. More specifically, higher levels of team fluidity likely reduce the bond strength among members, which in turn reduces the effects that fluidity exhibits on emergent states and process variables. Therefore, we propose that the effects of fluidity on given process and emergent state variables will be in a similar direction among all teams, but will be relatively stronger in smaller teams.

20

BRIAN R. DINEEN AND RAYMOND A. NOE

Proposition 9. The relationship between team fluidity and all of the proposed team process variables will be moderated by team size, such that the relationships will be stronger in smaller teams. Other Potential Boundary Conditions Arrow and McGrath (1995) have suggested several other contingent factors that may differentially influence the relationships we have proposed. The relative standing of those in a team who leave is likely to influence the proposed relationships differently. Losing and replacing the two highest-ranking people in the team should be different than losing and replacing the two lowest members. Related to these suggestions is the notion of social network position or closeness centrality. Figures 2 and 3 illustrate two scenarios that are likely to impact emergent states and process variables differently. In Fig. 2, two central team members (A and B) leave during a month period, whereas in Fig. 3, two peripheral members of the team network (I and J) leave. In Fig. 2, it is likely that the effects of fluidity on emergent states and process variables are greater compared to the scenario shown in Fig. 3. For example, the loss of two long-standing “opinion leaders” in a self-managed work team is different than the loss of two “outsiders” who have only recently joined that team. Of course, there are other aspects of the social network that should be considered, such as the overall density of the network (Brass, 1995), and the effects of fluidity in dense and loosely-connected team networks. Another point to consider is that different patterns of fluidity are likely to exhibit different effects on teams. For example, losing and replacing one member a month for four months might lead to certain process effects, whereas turning over four members at once at the end of four months might lead to others. Also, the timing

Fig. 2. Central Members Leave a Team.

The Impact of Team Fluidity and Its Implications

21

Fig. 3. Peripheral Members Leave a Team.

of fluidity is potentially important. Membership change just before a big team project deadline is likely to differ from member changes just after completion of a project. We make no specific propositions concerning the preceding possibilities, but reiterate Arrow and McGrath’s (1995) suggestion that these issues guide future team dynamics research.

IMPLICATIONS OF TEAM FLUIDITY FOR HUMAN RESOURCE MANAGEMENT PRACTICES
Few studies have directly investigated how to staff, train, manage performance, and compensate individuals in fluid teams. As noted previously, most of the HRM literature assumes that team membership is stable. Team fluidity is especially an issue in virtual organizations. Most work within a virtual organization is project-based, control and authority resides within team members, the success of the organization is dependent on collaboration and cooperation, the work environment is flexible, dynamic, and fluid, and work is conducted across time and space (Ellingson & Wiethoff, 2002). Below we discuss the implications of team fluidity for HRM practices and identify important research questions.

Staffing A wide range of individual characteristics and skills have been suggested as predictors of team effectiveness including cognitive ability, Big Five personality

22

BRIAN R. DINEEN AND RAYMOND A. NOE

factors (openness to experience, extraversion, conscientiousness, agreeableness, neuroticism), conflict resolution, risk-tolerance, and collaborative problem solving (e.g. Ellingson & Wiethoff, 2002; Stevens & Campion, 1994). In a study of how team member ability and personality related to differences in team effectiveness, Barrick, Stewart, Neubert and Mount (1998) found that teams with higher cognitive ability and conscientiousness were better performers than teams that were lower in cognitive ability and conscientiousness. Moynihan and Peterson (2001) identified three approaches that have been used to understand the team member personality-team performance relationship. Personality is believed to influence team processes either universally across all teams, contingently by task or organizational culture, and configurationally by taking into account the integration of team member personality traits (Moynihan & Peterson, 2001). Moynihan and Peterson (2001) found support for all three approaches. For example, the configuration approach to personality in teams showed that team configurations that are high and homogeneous on levels of conscientiousness, agreeableness, and openness to experience, and heterogeneous on extroversion lead to positive team processes and outcomes. The configuration approach to studying personality in teams research assumes it is either the similarity or dissimilarity of complementary traits within a team that influence performance. Whereas these approaches have received support, they do not account for the existence of team fluidity and it would be interesting to see how the personality configurations of the teams relate to team effectiveness in more fluid teams. For example, are teams with more variability in personality traits better able to absorb and utilize the skills of a new team member than teams more homogeneous in traits that are joined by a team member whose standing on a personality trait is different from the rest of the team? Klimoski and Jones (1995) make an important observation in noting that the traditional staffing model ignores team life-cycle issues. For example, a team having difficulty may seek to add members that bring new interpersonal and task specific skills to a team. These skills might not have been initially identified in an analysis of the team task. The selection process for a team that is forming likely relies more on the knowledge, skills, and abilities requirements uncovered in a job and task analysis. As interpersonal processes unfold in the team, however, the need for certain types of interpersonal skills might only then become salient. Klimoski and Jones (1995) suggest that to aid in the selection of new team members, assessment procedures that include employees as observers could be used to identify stable and emergent team staffing requirements. Assessment could be in the form of an assessment center or structured interview of team members. Also, specific KSAs could be identified that are more important in a fluid team arrangement as opposed to a more stable team. For example, there is a growing literature base that examines adaptability as an important KSA given the

The Impact of Team Fluidity and Its Implications

23

changing nature of work and the employment contract (e.g. LePine, Colquitt & Erez, 2000; Pulakos, Arad, Donovan & Plamondon, 2000). We further develop the idea of adaptability in fluid teams in our discussion of performance management.

Performance Management Team effectiveness is usually evaluated by looking at team outcomes such as customer service, productivity, quality, or innovativeness. In fluid teams, the importance of team members being able to work effectively despite membership changes is key. Underlying the ability to deal with changes in team membership is the concept of adaptability. Pulakos et al. (2000) developed a taxonomy of adaptive performance. They identified eight dimensions of adaptive performance including handling emergencies or crisis situations, work stress, solving problems creatively, dealing with uncertain and unpredictable work situations, learning work tasks, technologies, and procedures, and demonstrating interpersonal adaptability, cultural adaptability, and physically-oriented adaptability. They developed an instrument, the Job Adaptability Inventory (JAI), which can be used to evaluate team member performance, diagnose the adaptability requirements of jobs, and help in training team members in appropriate adaptive responses. To improve the effectiveness of selection of members to fluid teams, studies are needed to identify how individual differences relate to each of the dimensions of the JAI. Another important consideration in managing the performance of fluid team members is in actually assessing performance of those team members. A useful example to illustrate this point is in conducting 360-degree performance appraisals, which consist of the combined evaluations of superiors, subordinates, peers, and sometimes customers and other outsiders. It is again often assumed that in carrying out such appraisal processes, the members of the group or team in question are stable, and can thus provide accurate appraisals of a member because they have consistently been around that individual. Such is not always the case in more fluid teams; therefore, determining the best sources of appraisal information is an important area for research.

Team Training Cannon-Bowers, Tannenbaum, Salas and Volpe (1995) suggest that in teams where turnover is rapid (such as may be the case in fluid teams), task-specific competencies are critical and team-specific competencies are less important. Task-specific competencies include understanding the roles and role significance of different

24

BRIAN R. DINEEN AND RAYMOND A. NOE

positions on the team, skills in leadership or team management, feedback and performance monitoring, and coordination. They suggest that in these situations appropriate training strategies include task simulation, cross-training, guided task practice, role-playing and passive demonstration. In a lab experiment, Marks, Sabella, Burke and Zaccaro (2002) examined the role of three different types of cross-training in developing shared team interaction models, coordination, and performance in action teams. Action teams require more specialized skill sets, rely more heavily on coordination, perform in less familiar and more challenging environments, and may be temporary. Shared mental models are believed to have a direct effect on team coordination and backup processes that lead to performance. In teams, shared mental models represent knowledge and understanding of the team’s purpose and characteristics, connections among team member actions, and the roles and patterns required by team members to successfully complete collective action. These models directly relate to Cannon-Bowers et al.’s (1995) idea of task specific competencies. The types of cross-training examined included position clarification (training designed to raise awareness about team members jobs through lecture or discussion), position modeling (training involving both verbal discussion and observation of team members’ roles) and position rotation (provides team members with experience carrying out team members duties through taking on their role). The results suggest that the two more in-depth types of cross-training created more shared team interaction knowledge among team members, but positional rotation was not necessary to obtain this effect. More research on cross-training and the other strategies suggested by Cannon-Bowers et al. (1995) is needed; of particular interest are studies that identify whether or not less intensive training (such as positional clarification) can help develop shared mental models in fluid teams.

Creation of Intellectual Capital Intellectual capital is created through the combination and exchange of existing intellectual resources including tacit and explicit knowledge. Combination refers to the connection of elements previously unconnected or the development of novel ways of combining elements. For example, in crafting a new marketing campaign, one team member might provide first-hand knowledge of the general educational level of a targeted geographic area, whereas another might lend technological knowledge of the product to the prediction of whether it will be suited to individuals residing in that geographic area. Exchange refers to social interaction between individuals through teamwork, collaboration, and sharing. Nahapiet and Ghoshal (1998) model the creation of intellectual capital. They propose that structural,

The Impact of Team Fluidity and Its Implications

25

cognitive, and relational dimensions of social capital influence the combination and exchange of intellectual capital, which directly affects intellectual capital creation. Trust, norms, obligation, and identification are believed to be critical elements of relations needed for combination and exchange to occur. In teams that are charged with developing intellectual capital, the level of fluidity might work to the detriment of building trust and identification within the team. Indeed, Noe et al. (2003) posit that membership stability is an important team design characteristic when considering the development of intellectual capital, and that the relationship between membership stability and intellectual capital creation is likely to be complex. Periodic changes in the basic composition of the team can introduce new sources of individual intellectual capital, but may also detract from the team’s intellectual capital, particularly when the departing member(s) possessed tacit knowledge. New members or new configurations of existing members should prevent the group from stagnating. This should be particularly true for teams who institutionalize explicit member knowledge (i.e. create a “team memory”), which can outlast individual members. However, there is certainly a point at which membership instability will become counterproductive. Too much variation in either the standing or acting group will result in a situation where intellectual capital must repeatedly re-emerge in the new collective, as members must first decide “what they know” before concerning themselves with improving on that level of knowledge. In sum, Noe et al. (2003) suggests that there may be a threshold level of fluidity such that a lack of fluidity causes a team to stagnate in terms of task conflict (i.e. constructive controversy) and creativity but too much fluidity interferes with the development of team knowledge structures.

Compensation In teams that experience fluidity, forms of compensation that reward flexibility and cooperation should be considered (see Heneman, Tansky & Tomlinson, 2002). Team fluidity has implications for base pay, incentives, and indirect rewards. Base pay is the amount of wages or salary provided to employees for their services. Person-based pay approaches focus on the competencies of the job incumbent rather than the job. Competencies include interests, attitudes, knowledge, skills, and abilities. In a broadbanding pay system there are a small number of pay grades but large pay ranges within and between pay bands. In broadbanding systems there is more room to reward individual differences that relate to fluid team success (e.g. team process factors, contributions to team task, seniority) than in traditional pay systems that have many narrow pay grades.

26

BRIAN R. DINEEN AND RAYMOND A. NOE

Many organizations use incentive plans that include cash bonuses that are linked to employee, team, business unit, or organizational performance. In teams, and especially fluid teams, incentive distribution should include an individual component as well as a team component. The team component would be based on the degree to which the team meets objective performance outputs (such as quality, sales, reduced scrap, etc.). The individual component would be based on a subjective evaluation (ratings) of how the individual behaviors contributed to team accomplishment. That is, both objective performance measures as well as subjective performance measures should be used in team incentive systems. Subjective 360-degree evaluations based on self, manager, and other team members can provide a comprehensive evaluation of team members’ “soft skills” (e.g. communication, cooperation, information sharing), although the potential pitfalls of 360-degree appraisals in fluid teams as discussed earlier should be considered. Also, team members should receive incentives for team outputs that are commensurate with their contribution in helping the team realize the output. This should be the case both for team members who are currently on the team as well as those who have left the team. A critical task in determining incentive pay in fluid teams is determining the contribution levels of those who quickly move in and out of a team as it completes various phases of a project. For example, Team Member A might only be on the team for one week, but might make the most substantial contribution, whereas Team Member B might only contribute incrementally, but do so over a period of a year. Such nuances will need to be considered by compensation researchers and practitioners. Indirect rewards involve recognition and development. Recognition involves praise, gifts, awards, and time-off. Recognition confers status to members of fluid teams for individual contributions to team success as well as overall team performance. It may also be an important form of social capital because recipients become more visible to influential people in the organization. Being a member of a fluid team may also be rewarding to an individual if it is part of the individual’s development plan. Job experiences are one type of employee development activity. McCauley, Ruderman, Ohlott and Morrow (1994) identified five different types of job demands that require employees to stretch their skills (e.g. forced to learn a new skill, apply skills and knowledge in a new way, and master new experiences). These demands include making transitions, creating change, having a high level of responsibility, being involved in a non-authority relationship, and facing obstacles. Membership in a fluid team may provide an opportunity to deal with all of these job demands. Research has not adequately addressed issues of team compensation. For example, research is needed on the effectiveness of different combinations of individual, team, and higher-unit performance based pay. Also, researchers should examine

The Impact of Team Fluidity and Its Implications

27

how compensation systems interact with other human resource systems (Heneman, Ledford & Gresham, 2000) in the context of a fluid team-based organization.

METHODOLOGICAL ISSUES IN STUDYING TEAM FLUIDITY
Certain methodological issues should be considered when studying the effects of team fluidity on team emergent states and processes. These include the measurement of fluidity, the measurement of interrater agreement versus reliability, measuring outcome variables, and potential reverse causality issues involving fluidity. One of the most vital tasks researchers face is establishing a means of assessing and quantifying the level of fluidity in a team. For example, a team of eight members who experience the replacement of four of those members in two weeks is likely to differ from a similar team experiencing four replacements over a year. Traditional turnover indices only consider those members who permanently leave the team. Alternatively, fluidity can include shifts in membership that are temporary or only occur across the boundary between the standing and acting team. And, while our current measures of turnover seem inadequate from a conceptual standpoint, some of the existing turnover indices are inadequate from a methodological standpoint as well. For example, Carley (1992) defines the rate of turnover as: 1 , (number of periods/number of members turned over) or one divided by the mean number of periods between exits of team members. Selection of a “period” is arbitrary (i.e. the measurement is relative to the period chosen). However, Cohen, Cohen, West and Aiken (2003) review fundamental problems with usage of ratio measures such as this. Specifically, one could arrive at a correlation between this index that is driven by number of periods, number of members who left the team, or the ratio of the two. Without decomposing the ratio there is no way to tell which of the three potential drivers is acting. Although the scope of this paper does not allow us to develop a specific fluidity measure, we encourage researchers to begin addressing this task. In addition to objective measures of fluidity, recent work in the areas of relational demography and person-organization fit suggests the importance of considering perceived indices of fluidity in teams. That is, although objective measures of team membership changes are likely to relate to team processes and emergent states as we have suggested, team members’ perceptions of the amount of fluidity in the team might be an important predictor as well. Riordan (2000) distinguishes

28

BRIAN R. DINEEN AND RAYMOND A. NOE

between actual and perceived demographic similarity in her review of the relational demography literature and notes that “the use of both actual and perceived measures of demographic characteristics can dramatically increase the amounts of variance explained in outcome measures” (p. 160). For example, Cleveland and Shore (1992) found that variance explained rose from 0.01 to 0.58 when including a measure of perceived age in addition to actual age, and Riordan (1997) found only small correlations between actual and perceived measures of demographic characteristics such as age, gender, and race. Also, similar differences in variance explained between actual and perceived measures of person-organization fit have been found in recruitment research (e.g. Dineen, Ash & Noe, 2002; Judge & Cable, 1997). Interrater agreement is considered to be a necessary part of conducting teams research. Interrater agreement refers to a common perception among team members concerning a construct of interest, and is often measured through the use of indices such as r(wg) (James, Demaree & Wolf, 1984). In most traditional team studies in which members of the team stay together for the duration of the study, achieving an acceptable level of agreement among raters is common. However, studies that assess fluidity may encounter difficulties in achieving interrater agreement. Consider cohesiveness as an example. It is likely that some level of cohesiveness exists in a team. However, team members are likely to derive their “ratings” of cohesiveness from different frames of reference when their team tenures differ. For example, a longer tenured member may take into account feelings he has had over the last few months, whereas a new member may only be able to reference the five days she has been on the team. Related to this, Klein, Conn, Smith and Sorra (2001) found that a primary antecedent of within-group agreement was the degree of social interaction in the group, although they were not able to rule out reverse causation as an explanation for this result. Indeed, social interaction may tend to be less in more fluid teams without established patterns of interaction. Also, Dansereau, Yammarino and Kohles (1999) suggest that teams can move between levels of analysis depending on their current state. For example, a highly tenured, cohesive team may be studied at a team level of analysis, but transform to an individual level of analysis following a rash of member replacements. Such movements between levels may potentially affect a researcher’s ability to accurately measure team member agreement. One possible remedy may be to utilize an additive composition model to indicate that a higher-level construct is simply a summation of lower level units, regardless of the variance among those units (Chan, 1998). For example, indices of cohesiveness in teams could be operationalized as the “total amount” of felt cohesiveness on the team, regardless of the variance in cohesiveness perceptions. Other researchers have made similar use of Chan’s composition typology in team

The Impact of Team Fluidity and Its Implications

29

research (e.g. Colquitt et al., 2002). A second possible remedy may be to rely more heavily on reliability indices rather than agreement indices when making aggregation decisions. Whereas agreement requires that all actors rate an item the same (e.g. a “3” on a 5-point scale), reliability requires only consistency in responses by raters (Bliese, 2000). That is, as long as responses are consistently different, reliability still exists. For example, if Rater A only uses 1–3 on a 5-point scale, whereas Rater B only uses 3–5, but they do so consistently, reliability exists. ICC(1) and ICC(2) are common reliability indices that might be more useful in assessing the potential for aggregating fluid team constructs given the potential for differing frames of reference among team members. Clearly these issues demand greater attention, and researchers need to carefully consider aggregation issues in examining team fluidity because a lack of agreement does not necessarily mean a lack of reliability or the absence of a construct, but rather might result from the level of team fluidity and differing frames of reference brought to a team by changing members. Another methodological concern has to do with properly measuring outcomes of team fluidity. Specifically, whereas team fluidity might be conceived of as exits/entrances during a specified period of time, the proper measurement of outcomes (e.g. cohesiveness or collective efficacy) is less clear. For example, should researchers measure cohesiveness once at the end of the period in question? Or should it be measured multiple times at specified time points? Although specific measurement details are beyond the scope of this paper, one potential avenue might be to pursue the use of time series data. This strategy has a long history of use in economics, but only recently has been used in teams research (e.g. Sawyer, Latham, Pritchard & Bennett, 1999). Another possible approach is to measure team fluidity over a specified period of time, while also measuring process and emergent state variables at the beginning and end of the period, noting any changes in those variables during that time period (see Neuman, Bolin & Lonergan, 2000). As illustrated in Fig. 1, a third important consideration when examining team fluidity is that there may be reciprocal effects between fluidity and emergent states, process or performance variables. For example, as previously described, team fluidity is likely to lead to increased relationship conflict in teams. However, it has also been suggested that relationship conflict leads to voluntary turnover (i.e. increased fluidity) as well (e.g. Pelled, 1996). Similarly, a sense of decreased collective efficacy could lead to greater fluidity, just as fluidity is likely to relate to decreased collective efficacy. Also, performance is likely to have a reciprocal effect on fluidity, although the direction of this effect is uncertain. For example, Griffeth and Hom (1995) reviewed studies that attempted to link performance to subsequent turnover, noting divergent results across these studies. Despite meta-analytic results that have demonstrated an overall negative relationship

30

BRIAN R. DINEEN AND RAYMOND A. NOE

between performance and subsequent turnover (cf. Griffeth & Hom, 1995), these authors concluded that the relationship is likely mediated by a number of subprocesses. Separating these effects is a considerable challenge for researchers, but it is especially important as team membership becomes more fluid. As a start, researchers should endeavor to engage in more longitudinal studies in order to parse out reciprocal effects (e.g. Neuman et al., 2000). Finally, in addition to the reciprocal effects described above, we recognize that other antecedents to team fluidity likely exist on both a macro and micro level. For example, the labor market is likely to affect fluidity on a large scale, with increased fluidity predicted during a “tight” labor market (i.e., organizations are experiencing a shortage of skilled employees so team members have job opportunities readily available in other organizations). Organizational structure or norms may influence the extent to which teams are more fluid versus stable. Also, the extent to which an organizational workforce is diverse has been shown to relate to turnover (e.g. O’Reilly et al., 1989; Wiersema & Bird, 1993). Although a comprehensive treatment of team fluidity antecedents is beyond the scope of this paper, these examples highlight the importance of considering them in future research.

FUTURE RESEARCH
To gain a better understanding of the relationship between team fluidity and team processes, emergent states, and outcomes, research should address the propositions presented in this paper and seek to develop additional propositions. In addition, there are other areas of research on team fluidity that warrant attention. First, as noted at the beginning of the paper, the virtual environment is gaining recognition as a viable means of team collaboration. Research should specifically address any differences that might exist between a virtual and traditional team environment when team fluidity increases. As Townsend et al. (1998) note, virtual teams are generally more fluid, so studying the effects of fluidity in virtual teams may be even more vital. Also, because virtual teams are often composed of members from functionally diverse backgrounds, the effects of fluidity should be examined in parallel with functional diversity as well as other types of diversity. For example, it might be the case that the effects of fluidity are attenuated in an already-diverse team, such that the introduction of new members is not as much of a “shock” as if might be in a more homogeneous team. Second, with an increasing trend towards fluidity comes a concomitant need to effectively staff work teams on a continual basis. Following research on underand over-staffed groups (Cini, Moreland & Levin, 1993), researchers should investigate staffing issues in more fluid teams, with a focus on what, if any,

The Impact of Team Fluidity and Its Implications

31

individual difference variables might help predict success when working in a more fluid team. For example, we mentioned adaptability characteristics earlier (Pulakos et al., 2000), but urge scholars to continue to uncover additional characteristics that might be important for newcomers to fluid teams as well as current members. Next, we echo Arrow and colleagues’ (2000) call for more qualitative research that attempts to break down the complex effects of team fluidity. One possible approach is naturalistic observation (Whitley, 1996), whereby the researcher observes and records process phenomena over time. Researcher involvement could vary on a continuum from direct participation as an acting member of a team to passive observer of team behavior. In addition, diaries, behavioral checklists, and palm pilots can be used in time sampling studies designed to collect data from team members as processes occur (e.g. Williams & Alliger, 1994). Such research would likely help direct future empirical investigations of team fluidity and uncover the most important variables in need of study. Time sampling studies could also help to confirm whether transition processes occur more frequently in teams experiencing high levels of fluidity. Finally, as suggested by Arrow and McGrath (1995), the effects of temporary fluidity should be examined independent of more permanent member change. For example, changes in membership in which a team member leaves the team for good and is replaced by another are likely different from temporary absences of team members who are periodically not present but remain part of the standing team. Both situations are forms of team fluidity, but likely differ in their effects on process and emergent state variables. Gruenfeld et al. (2000) studied the influence of changing group members on the production and transfer of knowledge and experience. They measured the influence of itinerant group member’s unique knowledge and experience on the team the member temporarily visited and the team they returned to. Itinerant group members refer to individuals who span team boundaries for the purpose of importing or exporting group knowledge. They tested hypotheses related to the direct and indirect influence of itinerants on team members (e.g. convincing the team to accept ideas or advice). Also, they tested hypotheses related to social perceptions of itinerants. For instance, itinerant members’ involvement in and contribution to group activities should be greater in the group of origin than in the temporary group. The results suggested that the direct influence by iterant members was reduced after they changed groups and had unique knowledge to share. Their unique ideas were used as often as those of indigenous members. Consistent with expectations, indirect influence by itinerant members was greater after they returned to their group of origin. These findings are an example of the ongoing work that needs to occur as we continue to examine fluidity in standing and acting teams.

32

BRIAN R. DINEEN AND RAYMOND A. NOE

CONCLUSION
The use of teams has grown considerably in organizations, with a paralleled growth in research aimed at increasing their effectiveness. A growing characteristic of teams is their enhanced fluidity, or likelihood of turning over members throughout their existence. Research has not been as quick to study this phenomenon, and this paper is a step towards fostering such investigation. By delving into the “black box” of team processes and emergent states resulting from team fluidity, researchers can begin to understand some of its complex effects. In doing so, they can more readily begin to predict performance outcomes and better inform both research and practice. In closing, we recognize that Arrow and McGrath (1995) raise a compelling question in asking when exactly a group “ends” and a new group “begins.” Consider a manufacturing team. At any given organization that has used manufacturing teams for a prolonged period, membership changes are inevitable. For example, such a team in 1970 is certainly different from the same team in 2003. Yet, when did the team “change?” When the supervisor left? When the last round of layoffs hit? The boundaries are unclear. For many teams or groups of people, change is a gradual, evolving process. For other teams, however, change is more rapid, such as a virtual team with members who might come and go by the week or even day. Charting the effects of team fluidity is necessary if we are to accurately measure present-day real-life team dynamics. Doing so seems especially timely given current business trends and more fluid team arrangements.

ACKNOWLEDGMENTS
We thank Jason A. Colquitt for his helpful comments and suggestions on an earlier version of this paper.

REFERENCES
Ancona, D. G., & Caldwell, D. F. (1998). Rethinking team composition from the outside in. In: M. A. Neale, E. A. Mannix & D. H. Gruenfeld (Eds), Research in Managing Groups and Teams (Vol. 1, pp. 21–38). Stamford, CT: JAI. Argote, L., Ingram, P., Levine, J. M., & Moreland, R. L. (2000). Knowledge transfer in organizations: Learning from the experience of others. Organizational Behavior and Human Decision Processes, 82, 1–8. Argote, L., Insko, C. A., Yovetich, N., & Romero, A. A. (1995). Group learning curves: The effects of turnover and task complexity on group performance. Journal of Applied Social Psychology, 25, 512–529.

The Impact of Team Fluidity and Its Implications

33

Arrow, H., & McGrath, J. E. (1993). Membership matters: How member change and continuity affect small group structure, process, and performance. Small Group Research, 24, 334–361. Arrow, H., & McGrath, J. E. (1995). Membership dynamics in groups at work: A theoretical framework. In: B. M. Staw & L. L. Cummings (Eds), Research in Organizational Behavior (Vol. 17, pp. 373–411). Greenwich, CT: JAI. Arrow, H., McGrath, J. E., & Berdahl, J. L. (2000). Small groups as complex systems: Formation, coordination, development, and adaptation. Thousand Oaks, CA: Sage. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Barrick, M. R., Stewart, G. L., Neubert, M. J., & Mount, M. K. (1998). Relating member ability and personality to work-team processes and team effectiveness. Journal of Applied Psychology, 83, 377–391. Bettenhausen, K. L. (1991). Five years of groups research: What we have learned and what needs to be addressed. Journal of Management, 17, 345–381. Bliese, P. D. (2000). Within-group agreement, non-independence, and reliability: Implications for data aggregations and analysis. In: K. J. Klein & S. W. J. Kozlowski (Eds), Multilevel Theory, Research, and Methods in Organizations (pp. 349–381). San Francisco: Jossey-Bass. Brass, D. J. (1995). A social network perspective on Human Resource Management. In: G. Ferris (Ed.), Research in Personnel and Human Resources Management (Vol. 13, pp. 39–79). Greenwich, CT: JAI Press. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Campion, M. A., Papper, E. M., & Medsker, G. J. (1996). Relations between work team characteristics and effectiveness: A replication and extension. Personnel Psychology, 49, 429–452. Cannon-Bowers, J. A., Tannenbaum, S. I., Salas, E., & Volpe, C. E. (1995). Defining competencies and establishing team training requirements. In: R. A. Guzzo, E. Salas & Associates (Eds), Team Effectiveness and Decision-Making in Organizations (pp. 333–380). San Francisco: JosseyBass. Carley, K. (1992). Organization learning and personnel turnover. Organization Science, 3, 20–46. Chan, D. (1998). Functional relations among constructs in the same content domain at different levels of analysis: A typology of composition models. Journal of Applied Psychology, 83, 234–246. Chen, G., & Bliese, P. D. (2002). The role of different levels of leadership in predicting selfand collective efficacy: Evidence for discontinuity. Journal of Applied Psychology, 87, 549–556. Cini, M. A., Moreland, R. L., & Levine, J. M. (1993). Group staffing levels and responses to prospective and new group members. Journal of Personality and Social Psychology, 65, 723–734. Cleveland, J. N., & Shore, L. M. (1992). Self- and supervisory perspectives on age and work attitudes and performance. Journal of Applied Psychology, 77, 469–484. Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum. Colquitt, J. A., Noe, R. A., & Jackson, C. (2002). Justice in teams: Antecedents and consequences of procedural justice climate. Personnel Psychology, 55, 83–109. Dansereau, F., Yammarino, F. J., & Kohles, J. C. (1999). Multiple levels of analysis from a longitudinal perspective: Some implications for theory building. Academy of Management Review, 24, 346–357. Dineen, B. R., Ash, S. R., & Noe, R. A. (2002). A web of applicant attraction: Person-organization fit in the context of web-based recruitment. Journal of Applied Psychology, 87, 723–734. Ellingson, J. E., & Wiethoff, C. (2002). From traditional to virtual: Staffing the organization of the future today. In: R. L. Heneman & D. B. Greenberger (Eds), Human Resource Management in Virtual Organizations (pp. 141–177). Greenwich, CT: Information Age Publishing.

34

BRIAN R. DINEEN AND RAYMOND A. NOE

Freeman, L. C. (1979). Centrality in social networks: Conceptual clarification. Social Networks, 1, 215–239. Gersick, C. J. G. (1988). Time and transition in work teams: Toward a new model of group development. Academy of Management Journal, 31, 9–41. Goodman, P. S., & Leyden, D. P. (1991). Familiarity and group productivity. Journal of Applied Psychology, 76, 578–586. Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78, 1360– 1380. Granovetter, M. (1982). The strength of weak ties: A network theory revisited. In: P. V. Marsden & N. Lin (Eds), Social Structure and Network Analysis (pp. 105–130). Beverly Hills, CA: Sage. Griffeth, R. W., & Hom, P. W. (1995). The employee turnover process. In: G. R. Ferris & K. M. Rowland (Eds), Research in Personnel and Human Resources Management (Vol. 13, pp. 245–293). Greenwich, CT: JAI Press. Gruenfeld, D. H., Mannix, E. A., Williams, K. Y., & Neale, M. A. (1996). Group composition and decision making: How member familiarity and information distribution affect process and performance. Organizational Behavior and Human Decision Processes, 67, 1–15. Gruenfeld, D. H., Martorana, P. V., & Fan, E. (2000). What do groups learn from their worldliest members? Direct and indirect influence in dyadic teams. Organizational Behavior and Human Decision Processes, 82, 45–59. Guzzo, R. A., & Dickson, M. W. (1996). Teams in organizations: Recent research on performance effectiveness. Annual Review of Psychology, 47, 307–338. Guzzo, R. A., & Shea, G. P. (1992). Group performance and intergroup relations in organizations. In: M. D. Dunnette & L. M. Hough (Eds), Handbook of Industrial and Organizational Psychology (Vol. 3, 2nd ed., pp. 269–313). Palo Alto, CA: Consulting Psychologists Press. Hackman, J. R. (1987). The design of work teams. In: J. W. Lorsch (Ed.), Handbook of Organizational Behavior (pp. 315–342). Englewood Cliffs, NJ: Prentice-Hall. Harrison, D. A., Price, K. H., & Bell, M. P. (1998). Beyond relational demography: Time and the effects of surface- and deep-level diversity on work group cohesion. Academy of Management Journal, 41, 96–107. Heneman, R. L., Ledford, G. E., Jr., & Gresham, M. T. (2000). The changing nature of work and its effects on compensation design and delivery. In: S. L. Rynes & B. Gerhart (Eds), Compensation in Organizations (pp. 195–240). San Francisco: Jossey-Bass. Heneman, R. L., Tansky, J. W., & Tomlinson, E. C. (2002). Hybrid reward systems for virtual organizations: A review and recommendations. In: R. L. Heneman & D. B. Greenberger (Eds), Human Resource Management in Virtual Organizations (pp. 245–262). Greenwich, CT: Information Age Publishing. Hightower, R. T., Sayeed, L., Warkentin, M. E., & McHaney, R. (1998). Information exchange in virtual work groups. In: M. Igbaria & M. Tan (Eds), The Virtual Workplace (pp. 199–216). Hershey, PA: Idea Publishing Group. Hinsz, V. B., Tindale, R. S., & Vollrath, D. A. (1997). The emerging conceptualization of groups as information processors. Psychological Bulletin, 121, 43–64. Hollenbeck, J. R., Ilgen, D. R., Sego, D. J., Hedlund, J., Major, D. A., & Phillips, J. (1995). Multilevel theory of team decision making: Decision performance in team incorporating distributed expertise. Journal of Applied Psychology, 80, 292–316. James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating within group interrater reliability with and without response bias. Journal of Applied Psychology, 69, 85–98. Janis, I. L. (1972). Groupthink. Boston: Houghton Mifflin.

The Impact of Team Fluidity and Its Implications

35

Jehn, K. A. (1997). A qualitative analysis of conflict types and dimensions in organizational groups. Administrative Science Quarterly, 42, 530–557. Jehn, K. A., Northcraft, G. B., & Neale, M. A. (1999). Why differences make a difference: A field study of diversity, conflict, and performance in workgroups. Administrative Science Quarterly, 44, 741–763. Joyce, W. F., & Slocum, J. (1984). Collective climate: Agreement as a basis for defining aggregate climate in organizations. Academy of Management Journal, 27, 721–742. Judge, T. A., & Cable, D. M. (1997). Applicant personality, organizational culture, and organizational attraction. Personnel Psychology, 50, 359–394. Katz, R. (1982). The effects of group longevity on project communication and performance. Administrative Science Quarterly, 27, 81–104. Klein, K. J., Conn, A. B., Smith, D. B., & Sorra, J. S. (2001). Is everyone in agreement? An exploration of within-group agreement in employee perceptions of the work environment. Journal of Applied Psychology, 86, 3–16. Klimoski, R., & Jones, R. G. (1995). Defining competencies and establishing team training requirements. In: R. A. Guzzo, E. Salas & Associates (Eds), Team Effectiveness and Decision-Making in Organizations (pp. 292–332). San Francisco: Jossey-Bass. Klimoski, R., & Mohammed, S. (1994). Team mental models: Construct or metaphor? Journal of Management, 20, 403–437. Kristof, A. L., Brown, K. G., Sims, H. P., Jr., & Smith, K. A. (1995). The virtual team: A case study and inductive model. In: M. M. Beyerlein, D. A. Johnson & S. T. Beyerlein (Eds), Advances in Interdisciplinary Studies of Work Teams: Knowledge Work in Teams (Vol. 2). Greenwich, CT: JAI Press. Lawler, E. E., III, Mohrman, S. A., & Ledford, G. E., Jr. (1995). Creating high performance organizations: Practices and results of employee involvement and total quality management in Fortune 1000 companies. San Francisco: Jossey-Bass. LePine, J. A., Colquitt, J. A., & Erez, A. (2000). Adaptability to changing task contexts: Effects of general cognitive ability, conscientiousness, and openness to experience. Personnel Psychology, 53, 563–594. Longley, J., & Pruitt, D. G. (1980). A critique of Janis’ theory of groupthink. In: L. Wheeler (Ed.), Review of Personality and Social Psychology (Vol. 1). Beverly Hills, CA: Sage. Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally based framework and taxonomy of team processes. Academy of Management Review, 26, 356–376. Marks, M. A., Sabella, M. J., Burke, C. S., & Zaccaro, S. J. (2002). The impact of cross-training on team effectiveness. Journal of Applied Psychology, 87, 3–13. Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas, E., & Cannon-Bowers, J. A. (2000). The influence of shared mental models on team process and performance. Journal of Applied Psychology, 85, 273–283. McCauley, C. D., Ruderman, M. N., Ohlott, P. J., & Morrow, J. E. (1994). Assessing the developmental components of managerial jobs. Journal of Applied Psychology, 79, 544–560. McGrath, J. E. (1991). Time, interaction, and performance (TIP): A theory of groups. Small Group Research, 22, 147–174. Meyerson, D., Weick, K. E., & Kramer, R. M. (1996). Swift trust and temporary groups. In: R. M. Kramer & T. R. Tyler (Eds), Trust in Organizations: Frontiers of Theory and Research (pp. 166–195). Thousand Oaks: Sage. Michel, J. G., & Hambrick, D. C. (1992). Diversification posture and top management team characteristics. Academy of Management Journal, 35, 9–37.

36

BRIAN R. DINEEN AND RAYMOND A. NOE

Milliken, F. J., & Martins, L. L. (1996). Searching for common threads: Understanding the multiple effects of diversity in organizational groups. Academy of Management Review, 21, 402–433. Moynihan, L., & Peterson, R. (2001). A contingent configuration approach to understanding the role of personality in organizational groups. In: B. Staw & R. Sutton (Eds), Research in Organizational Behavior (Vol. 23, pp. 327–378). New York: JAI Press. Muoio, A. (2000, June). How to play the talent game. Fast Company, 362–371. Nahapiet, J., & Ghoshal, S. (1998). Social capital, intellectual capital, and the organizational advantage. Academy of Management Review, 23, 242–266. Nemeth, C. (1986). Differential contributions of majority and minority influence. Psychological Review, 93, 23–32. Neuman, G. A., Bolin, A., & Lonergan, J. M. (2000). Models of team performance: What’s the real effect of team processes on team performance? Paper presented at the 2000 annual conference for the Society for Industrial and Organizational Psychology. Noe, R. A., Colquitt, J. A., Simmering, M. J., & Alvarez, S. A. (2003). Knowledge management: Developing intellectual and social capital. In: S. E. Jackson, M. A. Hitt & A. S. DeNisi (Eds), Managing Knowledge for Sustained Competitive Advantage: Designing Strategies for Effective Human Resource Management (pp. 209–242). San Francisco: Jossey-Bass. O’Connor, K. M., Gruenfeld, D. H., & McGrath, J. E. (1993). The experience and effects of conflict in continuing work groups. Small Group Research, 24, 362–382. O’Reilly, C. A., III, Caldwell, D. F., & Barnett, W. P. (1989). Work group demography, social integration, and turnover. Administrative Science Quarterly, 34, 21–37. Pelled, L. H. (1996). Demographic diversity, conflict, and work group outcomes: An intervening process theory. Organization Science, 7, 615–631. Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. (1999). Exploring the black box: An analysis of work group diversity, conflict, and performance. Administrative Science Quarterly, 44, 1–28. Peterson, E., Mitchell, T. R., Thompson, L., & Burr, R. (2000). Collective efficacy and aspects of shared mental models as predictors of performance over time in work groups. Group Processes & Intergroup Relations, 3, 296–316. Pulakos, E. D., Arad, S., Donovan, M. A., & Plamondon, K. E. (2000). Adaptability in the workplace: Development of a taxonomy of adaptive performance. Journal of Applied Psychology, 85, 612–624. Riordan, C.M. (1997). Advancing relational demography theory: A construct validity study of three measures of demographic similarity. In: L. N. Dosier & J. B. Keys (Eds), Academy of Management Best Paper Proceedings. Riordan, C. M. (2000). Relational demography within groups: Past developments, contradictions, and new directions. In: K. Rowland & G. Ferris (Eds), Research in Personnel and Human Resource Management (pp. 131–173). Greenwich, CT: JAI Press. Riordan, C. M., & Shore, L. M. (1997). Demographic diversity and employee attitudes: An empirical examination of relational demography within work units. Journal of Applied Psychology, 82, 342–358. Rousseau, D. M. (1977). Technological differences in job characteristics, employee satisfaction, and motivation: A synthesis of job design research and sociotechnical systems theory. Organizational Behavior and Human Performance, 19, 18–42. Saunders, C. S., & Ahuja, M. K. (2000). A framework for understanding temporal issues of virtual teams. Paper presented at the 2000 annual conference of the Academy of Management.

The Impact of Team Fluidity and Its Implications

37

Sawyer, J. E., Latham, W. R., Pritchard, R. D., & Bennett, W. R., Jr (1999). Analysis of work group productivity in an applied setting: Application of a time series panel design. Personnel Psychology, 52, 927–967. Schopler, J. H., & Galinsky, M. J. (1990). Can open-ended groups move beyond beginnings? Small Group Research, 21, 435–449. Shah, P. P., & Jehn, K. A. (1993). Do friends perform better than acquaintances: The interaction of friendship, conflict, and task. Group Decision and Negotiation, 2, 149–166. Simons, T. L., & Peterson, R. S. (2000). Task conflict and relationship conflict in top management teams: The pivotal role of intragroup trust. Journal of Applied Psychology, 85, 102–111. Smith, K. G., Smith, K. A., Olian, J. D., Jr, Sims, H. P., O’Bannon, D. P., & Scully, J. A. (1994). Top management team demography and process: The role of social integration and communication. Administrative Science Quarterly, 39, 412–438. Stein, M. I. (1982). Creativity, groups, and management. In: R. A. Guzzo (Ed.), Improving Group Decision Making in Organizations (pp. 127–156). New York: Academic Press. Stevens, M. J., & Campion, M. A. (1994). The knowledge, skill, and ability requirements for teamwork: Implications for human resource management. Journal of Management, 20, 533–550. Tajfel, H. (1982). The social psychology of intergroup relations. Annual Review of Psychology, 33, 1–39. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In: S. Worchel & W. G. Austin (Eds), Psychology of Intergroup Relations (pp. 33–47). Chicago: Nelson-Hall. Townsend, A. M., DeMarie, S. M., & Hendrickson, A. R. (1998). Virtual teams: Technology and the workplace of the future. Academy of Management Executive, 12, 17–29. Trist, E. L. (1981). The evolution of sociotechnical systems: A conceptual framework and an action research program. Toronto, Ontario: Quality of Working Life Centre. Tuckman, B. W. (1965). Developmental sequence in small groups. Psychological Bulletin, 63, 384–399. Watson, W. E., Kumar, K., & Michaelsen, L. K. (1993). Cultural diversity’s impact on interaction process and performance: Comparing homogeneous and diverse task groups. Academy of Management Journal, 36, 590–602. Whitley, B. E., Jr. (1996). Principles of research in behavioral science. Mountain View, CA: Mayfield Publishing Co. Wiersema, M. F., & Bird, A. (1993). Organizational demography in Japanese firms: Group heterogeneity, individual dissimilarity, and top management team turnover. Academy of Management Journal, 36, 996–1025. Williams, K. J., & Alliger, G. M. (1994). Role stressors, mood spillover, and perceptions of work – family conflict in employed parents. Academy of Management Journal, 37, 837–868. Zajonc, R. B. (1965). Social facilitation. Science, 149, 269–274. Zenger, T. R., & Lawrence, B. S. (1989). Organizational demography: The differential effects of age and tenure distributions on technical communication. Academy of Management Journal, 32, 353–376. Ziller, R. C. (1965). Toward a theory of open and closed groups. Psychological Bulletin, 64, 164–182. Ziller, R. C., Behringer, R. D., & Goodchilds, J. D. (1962). Group creativity under conditions of success or failure and variations in group stability. Journal of Applied Psychology, 46, 43–49. Ziller, R. C., Behringer, R. D., & Jansen, M. J. (1961). The newcomer in open and closed groups. Journal of Applied Psychology, 45, 55–58.

MENTORING RESEARCH: A REVIEW AND DYNAMIC PROCESS MODEL
Connie R. Wanberg, Elizabeth T. Welsh and Sarah A. Hezlett
ABSTRACT
Organizations have become increasingly interested in developing their human resources. One tool that has been explored in this quest is mentoring. This has led to a surge in mentoring research and an increase in the number of formal mentoring programs implemented in organizations. This review provides a survey of the empirical work on mentoring that is organized around the major questions that have been investigated. Then a conceptual model, focused on formal mentoring relationships, is developed to help understand the mentoring process. The model draws upon research from a diverse body of literature, including interpersonal relationships, career success, training and development, and informal mentoring. Finally, a discussion of critical next steps for research in the mentoring domain is presented.

INTRODUCTION
Mentoring refers to a one-on-one relationship between a less experienced (i.e. prot´ g´ ) and a more experienced person (i.e. mentor), and is prototypically e e intended to advance the personal and professional growth of the less experienced individual (Mullen, 1994). Mentoring relationships can be “informal” in nature,

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 39–124 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22002-8

39

40

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

having developed naturally between two individuals, or they can be “formal” in nature, likely the consequence of an assigned pairing of a mentor and prot´ g´ within e e an organization. Due to increasing interest in leveraging human and social capital within organizations, informal and formal mentoring have gained the attention of academicians and practitioners as potentially critical developmental tools. Research on mentoring has surged in the past decade, creating the need for a thorough review of the literature. The intent of this review is to: (1) provide a survey of empirical work on mentoring that is organized around the major questions that have been investigated; and (2) extend current understanding of mentoring relationships via an integrative, conceptual model. The first portion of the review consequently addresses the question “where are we now?,” and describes research priorities for the areas discussed. The second portion of the review presents a dynamic framework for understanding and examining formal mentoring relationships. Our review complements Noe, Greenberger and Wang (2002), who described the evolution of the concept of mentoring, theoretical frameworks that have been used in studying mentoring, methodological and measurement issues pertaining to mentoring research, and attachment theory as a framework for examining mentoring.

MENTORING RESEARCH: WHERE ARE WE NOW?
Our review focuses on published, under review, and in press research on “traditional” mentoring within the workplace (Noe et al., 2002). Articles for this review were identified through searches of ABI/Inform, Econlit and PsychInfo databases (through December, 2002), along with queries of mentoring researchers. Studies focusing on the mentoring of students are not comprehensively summarized, but are discussed as needed to call attention to research questions, methods, and results that may be fruitful to pursue in work settings. The results of our literature search indicated the majority of the research on mentoring in the workplace has been published within the last 25 years. Although a diverse array of issues have been examined, empirical work has been concentrated in five major topic areas: (1) outcomes of mentoring; (2) the role of diversity (especially gender and race) in mentoring; (3) the role of other individual characteristics in mentoring; (4) dynamics of mentoring relationships; and (5) formal mentoring programs. We begin this section with a brief discussion of the concept of mentoring, then review each of these five major topic areas in turn. The Concept of Mentoring A number of authors have proposed typologies differentiating “traditional” (oneon-one, hierarchical mentoring) from other supportive workplace relationships,

Mentoring Research

41

such as peer mentoring and developmental networks (Eby, 1997; Higgins & Kram, 2001; Missirian, 1982; Shapiro, Haseltine & Rowe, 1978; Whitely & Coetsier, 1993; Whitely, Dougherty & Dreher, 1991). Looking across these typologies, dimensions on which traditional mentoring relationships have been distinguished from other developmental relationships include the power of the more senior person, the emotional intensity of the relationship, the hierarchical distance between the participants, the social origins of the relationship, and the amount and focus of assistance provided by the more senior person. Despite differences between models and typologies of developmental relationships, it is generally agreed that mentoring is the most intense and powerful one-on-one developmental relationship, entailing the most influence, identification, and emotional involvement. The characteristics of mentoring relationships that contribute to participants’ development typically have been referred to as mentoring functions. Two broad categories of mentoring functions are widely recognized: career and psychosocial. Career functions are conceptualized as those mentoring functions that aid career advancement. They may include challenging assignments, coaching, exposure, protection, and sponsorship. Psychosocial functions help build a sense of identity, competence, and effectiveness. They may include acceptance, counseling, friendship and role modeling. These mentoring functions originally were identified in qualitative research (Kram, 1985b), and have been regularly used in subsequent quantitative research. More detailed descriptions of the functions, along with sample items from the three most well-known instruments regularly used to measure multiple mentoring functions (Ragins, 1999), are presented in Table 1. Other established instruments yield an overall index of mentoring functions (Dreher & Ash, 1990) or solely assess career functions (Whitely, Dougherty & Dreher, 1992). There are two areas of uncertainty regarding how best to represent the construct space of mentoring functions. One area of ambiguity is how many distinct dimensions of mentoring functions there are. Research on the dimensionality of mentoring functions has been clouded by the use of principal components analysis, which does not permit strong inferences about underlying, latent structures. Results of exploratory factor analyses have been mixed, with some supporting a two-function model (e.g. Noe, 1988a) and others suggesting a three-function model. Several three-factor solutions suggest role-modeling should be viewed as a distinct mentoring function, rather than as an aspect of the psychosocial mentoring function (Barker, Monks & Buckley, 1999; Burke, 1984; Ochberg, Tischler & Schulberg, 1986; Scandura, 1992; Scandura & Ragins, 1993). Other studies support alternate three dimensional solutions (Steinberg & Foley, 1999; Turban & Dougherty, 1994). Two separate sets of confirmatory factor analyses, each based on data collected with different instruments, also have supported

42

Table 1. Mentoring Functions, Definitions, and Sample Scale Items.
Function Career functions Sponsor Definition Sample Scale Items

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Actively supporting an employee for lateral transfers and promotions Shielding an employee from damaging contact with key senior figures in the organization

Protect

Exposure

Giving an employee assignments that provide contact with key senior figures Helping an employee prepare for greater responsibility by providing challenging work and feedback that encourages skill development Sharing advice, information, and ideas that help an employee attain objectives and achieve recognition

Challenging assignments

Coacha

My mentor helps me attain desirable positions (MRI) Mentor gave you assignments or tasks in your work that prepare you for an administrative position (MFS) My mentor shields me from damaging contact with important people in the organization (MRI) Mentor helped you finish assignments/tasks or meet deadlines that otherwise would have been difficult to complete (MFS) My mentor helps me be more visible in the organization (MRI) Mentor helped you meet new colleagues (MFS) My mentor gives me tasks that require me to learn new skills (MRI) Mentor gave you assignments that present opportunities to learn new skills (MFS) My mentor suggests specific strategies for achieving career aspirations (MRI) Mentor has shared history of his/her career with you (MFS) My mentor is someone I can trust (MRI) My mentor guides my professional development (MRI) My mentor has demonstrated good listening skills in our conversations (MFS) My mentor sees me as being competent (MRI) My mentor has conveyed feelings of respect for me as an individual (MFS)

Psychosocial functions Friendship Counseling

Sharing informal social experiences Using active listening to enable an employee to explore personal concerns about self and career Conveying positive regard

Acceptance/Confirmation

Mentoring Research

Socialb Parentb Role modelingc

Participating in informal social activities one-on-one outside of work Construing the relationship as a parent/child relationship Serving as a model for an employee to emulate

My mentor and I frequently socialize one-on-one outside the work setting (MRI) My mentor is like a father/mother to me (MRI) My mentor represents someone who I want to be (MRI) I admire mentor’s ability to motivate others (MFQ)

Notes: MRI = Mentoring Role Instrument (Ragins & McFarlin, 1990); MFS = Mentoring Functions Scale (Noe, 1988a); MFQ = Mentoring Functions Questionnaire (Scandura, 1992; Scandura & Ragins, 1993). Unless otherwise noted, functions were identified and defined by Kram (1985b). a Items measuring coaching are part of the psychosocial scale of the MFS. b Social and parenting are mentoring functions that were proposed by Ragins and McFarlin (1990), drawing on observations made by Kram (1985b). c Some factor analytic results suggest this is a third mentoring function, distinct from career and psychosocial mentoring.

43

44

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

different models (Scandura & Williams, 2001; Tepper, Shaffer & Tepper, 1996). Overall, the bulk of the evidence indicates that there are at least two distinct mentoring functions (career and psychosocial), but is less clear on whether a third dimension is needed to adequately represent the construct space. A second area of uncertainty about the construct space of mentoring functions involves whether specific, narrow mentoring functions are facets of psychosocial or career mentoring functions. For example, Kram’s (1985b) classification of coaching as a career function and friendship as a psychosocial function has been followed in the development of some measures of mentoring functions (e.g. Ragins & McFarlin, 1990). In contrast, guided by the results of an exploratory factor analysis, Noe (1988a) included items assessing coaching in his measure of psychosocial mentoring and dropped items measuring friendship due to small loadings. Additional facets of psychosocial mentoring also have been proposed and studied (Ragins & Cotton, 1999; Ragins & McFarlin, 1990). While there is general agreement and support for the functions that a mentor may provide, these differences might be kept in mind during the review as the exact content of “psychosocial” or “career” mentoring may differ across studies. Researchers have begun to consider how mentoring relates to constructs from other streams of research (e.g. McManus & Russell, 1997). This is an important step in enhancing our understanding of the nomological network in which mentoring is embedded. Most empirical work in this area has focused on exploring how mentoring differs from supervision and leadership. Two distinct lines of research have been pursued, one comparing mentoring with “typical” supervisory relationships (Burke, McKenna & McKeen, 1991; Fagenson, 1994; Horgan & Simeon, 1990b; Tepper, 1995) and one examining the relationship between leader behaviors and mentoring functions (Godshalk & Sosik, 2000; Scandura & Schriesheim, 1994; Sosik & Godshalk, 2000a; Thibodeaux & Lowe, 1996). This work has been supportive of mentoring as distinguishable from supervision and leadership. Another question relevant to the construct of mentoring that has attracted attention concerns possible differences in the nature of formal and informal mentoring. Early researchers were skeptical about the potential of formal mentoring relationships (see, for example, Kram, 1985a). Recent evidence provides more optimistic conclusions (Ragins, Cotton & Miller, 2000). Nevertheless, formal and informal mentoring relationships do differ in significant ways (Ragins & Cotton, 1999). For this reason, as well as because of the recent interest and enthusiasm for formal mentoring programs among organizations, studies explicitly considering formal mentoring are discussed separately in our review. As a last overall observation, definitions of mentoring have been criticized for being inconsistent (Carden, 1990; Chao, 1998; Pollock, 1995). This criticism

Mentoring Research

45

is valid. It is unquestionable that it would be useful for researchers to agree on a consistent definition of mentoring to use when asking participants if they have had a mentor. However, we wish to note there is a high consistency in the literature in regard to the general concept of mentoring, the more general understanding of what is meant by a traditional mentoring relationship. Progress has been made in conceptually differentiating mentoring from other kinds of developmental relationships and in empirically distinguishing mentoring from supervision and leadership. Additional clarity about the construct of mentoring can be achieved through further research on how best to represent the construct space of mentoring functions, and, as will be seen later in this review, by more carefully distinguishing between informal and formal mentoring relationships.

Mentoring Outcomes Whether mentoring and its associated functions impact work- and career-related outcomes has been a major focus of mentoring research. Indeed, we identified one meta-analysis (Allen, Eby, Poteet, Lentz & Lima, 2002) and over 90 studies that have examined outcomes of mentoring. To characterize the individual studies, we coded each of them in terms of their focus, methodology, type of mentoring examined, source of data, and sample size (see Table 2). This analysis shows that most studies have focused on outcomes for prot´ g´ s, rather than on outcomes for e e mentors or organizations. Furthermore, most studies have used cross-sectional survey research methods and have relied almost exclusively on self-report data. While the range of sample sizes is quite broad, from 22 to 2,321, most of the studies have had sample sizes of more than 100 participants. In contrast to the precise categorizations on the dimensions of focus, methodology, source of data, and sample size, we were only able to categorize with certainty one in three studies in terms of type of mentoring examined (i.e. formal or informal). We were unable to categorize many studies because the authors did not explicitly distinguish between formal and informal relationships in the instruments or instructions provided to study participants. Without explicit instruction, respondents may have referred to either an informal or a formal mentoring relationship when answering questions about that relationship. This is a measurement issue that should be remedied in future studies. In the rest of this section, we review the current understanding of mentoring outcomes based on the studies reported in Table 2 that have either focused on informal mentoring or have not differentiated between outcomes for informal and formal mentoring. As mentioned earlier, studies that have explicitly examined

46

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Table 2. Characteristics of Empirical Studies Examining Mentoring Outcomes.
Category Focusa Prot´ g´ outcomes e e Mentor outcomes Organization outcomes Type of mentoring Formal only Informal only Both, separated Both, not separated Unspecified Research methoda Survey Interview (Qualitative) Experimental Research design Cross-sectional Longitudinal with two waves Longitudinal more than two waves Data sourcesa Self-report only Dyad Other (files, co-workers, etc.) Sample sizeb Less than 100 100 to less than 300 300 to less than 500 More than 500
a Does b Does

%

Number

96 13 3 9 11 8 5 66 93 8 2 88 9 3 81 10 10 17 49 17 17

92 12 3 9 11 8 5 63 89 8 2 84 9 3 78 10 10 16 47 16 16

not add to 100% as studies may be categorized in more than one way. not add to 100% as one study did not provide sample size. Mean sample size = 324.

formal mentoring programs are reviewed separately in a later section of this paper. A caveat to the current section is that to the extent that some research has mixed individuals with informal and formal mentoring relationships, we can only report the outcomes of mentoring in general (regardless of relationship type). It is notable, however, that research on frequencies of mentoring type is suggestive of the idea that most of the mentoring relationships reported in studies with a formal/informal mix are informal in nature (e.g. Day & Allen, 2002; Godshalk & Sosik, 2000; Ragins & Cotton, 1999).

Mentoring Research

47

Prot´ g´ Outcomes e e A fundamental, salient question in the mentoring literature has been whether mentoring relationships lead to positive outcomes for the prot´ g´ , such as higher e e salary and increased job satisfaction. The research in this area is fairly consistent in finding that there is an association between being a prot´ g´ and favorable e e outcomes. We provide a survey of this available evidence, then propose that an important next step for this literature is to attend more closely to internal validity issues in order to determine to what extent positive outcomes for the prot´ g´ can e e be uniquely attributed to mentoring as opposed to other factors. Correlational relationship results. A narrative summary of the over 90 studies that have examined prot´ g´ outcomes is difficult. Fortunately, a meta-analysis on e e prot´ g´ outcomes was recently conducted (Allen, Eby et al., 2002),1 providing e e an overall synthesis of prot´ g´ outcome results across studies. The results of e e this meta-analysis are informative. First, they demonstrate that despite the large number of studies in this area, once the studies are categorized into specific relationships (e.g. mentoring-promotion, mentoring-job satisfaction), the number of studies in each relationship cell is rather small. Second, the results are supportive of positive benefits of mentoring. More specifically, they show that aggregating across multiple studies, mentoring is positively related to both subjective and objective outcomes for prot´ g´ s. e e First, when comparing individuals who had mentors to those who did not (Allen, Eby et al., 2002), individuals with mentors had more positive subjective outcomes including higher expectations for advancement, career satisfaction, job satisfaction, career commitment, and intentions to stay at their organizations. Effect sizes for these relationships were small (Cohen, 1988), but their confidence intervals did not include zero. Individuals with mentors also had higher levels of compensation (small effect size) and promotions (medium effect size) when compared to individuals without mentors. The number of studies available in each relationship cell was small, with findings based on a range of three (promotions) to ten (job satisfaction) studies. An illustrative study that compared outcomes between those with mentors and those without is that of Prevosto (2001). In a cross-sectional study of 171 army reserve nurses, individuals with mentors reported higher job satisfaction and intention to stay than individuals without mentors. Allen, Eby et al. (2002) also examined the relationship between the amount of psychosocial and career mentoring received by individuals and both subjective and objective outcomes. Psychosocial mentoring had relationships that were small in effect size with prot´ g´ career satisfaction, job satisfaction, intention to stay e e with the organization, compensation, and promotions, and a relationship that was large in effect size with mentor satisfaction. Parallel to the results for psychosocial

48

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

mentoring, career mentoring had relationships that were small in effect size with prot´ g´ career satisfaction, job satisfaction, compensation, salary growth, and e e promotions, and a relationship that was large in effect size with mentor satisfaction. In general, relationships with the objective outcomes tended to be weaker than relationships with subjective outcomes. The number of studies available for each relationship meta-analyzed was small, ranging from 3 to 11. A recent illustrative study is that of Seibert, Kraimer and Liden (2001), who found that level of career sponsorship provided to the prot´ g´ (including sponsorship, exposure e e and visibility, challenging assignments, and protection) was positively related to prot´ g´ salary level, number of promotions, and career satisfaction in a sample of e e university alumni. Looking beyond the meta-analysis, other prot´ g´ outcomes have been studied e e in a limited number of studies. Work-life balance has been studied in two samples. In a sample of female lawyers, no relationship was found between being a prot´ g´ and level of work-nonwork conflict (Wallace, 2001). Nielson, Carlson and e e Lankau (2001) found a relationship between level of psychosocial support and role modeling received and family interference with work, but not with work interference with family. Procedural justice has also been studied within the context of mentoring. For example, Scandura (1997) found that individuals without mentors had lower levels of procedural justice (e.g. non-prot´ g´ s were e e less likely to report that “employees are allowed to challenge or appeal job decisions made by the manager” p. 63) than individuals with mentors. Wallace’s study compared prot´ g´ and non-prot´ g´ levels of procedural justice with similar e e e e findings. Other outcomes that have been studied include job burnout (Fagan & Walter, 1982) and organizational power (Fagenson, 1988). Fagan and Walter (1982) found that individuals with more than one mentor experienced more job burnout than individuals without mentors and those with only one mentor. Fagenson (1988) reported that individuals who had a mentor reported more power in their organization, including policy influence, access to important people and resource power, than individuals who did not have a mentor. Internal validity issues. Based upon positive empirical findings such as those just reviewed, it seems safe to conclude that research has been supportive of positive outcomes being correlated with both prot´ g´ status and the level of mentoring e e functions received. However, it is common to see the causal assertion that mentoring is a significant determinant of career success. Koberg, Boss and Goodman (1998), for example, stated “Mentors have been shown to affect promotion and compensation decisions and can influence career and organizational success” (p. 59). Ragins and Scandura (1997) stated “Mentoring relationships have been shown to be an important determinant in career success and advancement” (p. 945).

Mentoring Research

49

A detailed examination of studies on the effects of mentoring leads us to suggest, however, that research that more closely examines the extent to which mentoring is the unique driver of positive outcomes is needed. In order to more strongly establish the extent to which mentoring causally impacts career success, it is important for research not only to demonstrate that there is a relationship between mentoring and career success outcomes, but to show that mentoring precedes career success outcomes in time and that alternative explanations for the positive outcomes have been considered. Highly controlled experimental research designs where individuals from large random samples are allowed versus not allowed to have mentors are improbable, making this area of examination difficult. Initial research on the temporal sequencing of mentoring and outcomes is promising, although less than ten studies were found that analyzed whether mentoring received at one point in time was correlated with outcomes at a later point in time. As an illustration of this type of longitudinal research, Orpen (1995) found that the level of career coaching reported by new workers at six months of tenure was related to the number of promotions and salary growth over their first four years. Donaldson, Ensher and Grant-Vallone (2000) found that high quality mentoring relationships measured at Time 1 were correlated with organizational commitment and self-reported level of organizational citizenship behaviors (OCB) reported six months later. Silverhart (1994) found that new life insurance agents who reported having a mentor in their agency during their first few months of work were more likely still to be with the organization at the end of the first year, to have sold more policies during that year, and to have higher confidence in their ability to succeed. Finally, controlling for prot´ g´ gender, e e race, firm tenure, and firm orientation activities, Higgins and Thomas (2001) found that the organizational level of an individual’s set of developmental relationships measured at Time 1 was related to promotion to partner measured seven years later. While there is budding evidence that being a prot´ g´ precedes positive e e outcomes, alternative or spurious explanations for the outcomes have not been sufficiently examined. For example, while Dreher and Ash (1990) concluded that individuals who reported receiving higher levels of mentoring had higher levels of income (e.g. each unit of increase in mentoring was associated with an increase in pay of $3,236), it is possible that only a small component of this higher income should be attributed to the mentoring process. Although the researchers statistically controlled for a number of demographic, career history, and situational variables, it is possible that a substantial component of the income differential was due to individual characteristics that were not examined, such as achievement motivation, job performance, cognitive ability or emotional intelligence. Talented individuals who possess drive and commitment to their profession, career, and

50

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

organization, and who are good at developing social networks may be both more likely to get a mentor and to experience career success. The well-established linkages between ability, conscientiousness, motivation, and job performance (Sackett, Gruys & Ellingson, 1998; Schmidt & Hunter, 1998) and between ability, motivation, and objective career success (Tharenou, 1997) support the need for research that carefully examines the association of mentoring to career success outcomes above and beyond prot´ g´ ability, motivation, and e e other individual characteristics such as prot´ g´ personality and job performance. e e Only one study focused on mentoring outcomes was found that controlled for aspects of both prot´ g´ ability and motivation (Green & Bauer, 1995). These e e authors observed a significant zero-order correlation between career mentoring and publications, but this relationship was not significant in a multivariate context where graduate student ability (i.e. verbal and quantitative GRE scores) and commitment (i.e. toward the graduate program and to a research career) were controlled. A minority of others have controlled only for aspects of ability (e.g. Godshalk & Sosik, in press, through college GPA and GMAT scores; and Laband & Lentz, 1995, through rankings of graduation status and LSAT scores) or have included proxies for motivation such as work centrality, desire for upward mobility, and number of hours worked per week (e.g. Aryee, Wyatt & Stone, 1996; Johnson & Scandura, 1994; Wallace, 2001; Wayne, Liden, Kraimer & Graf, 1999; Whitely & Coetsier, 1993; Whitely et al., 1991, 1992). Controls for other relevant individual characteristics, such as job performance, that may be associated with both the receipt of more mentoring and with career success have also been rare. Another issue that limits the ability to understand the unique contribution of mentoring to career outcomes is the literature’s reliance on prot´ g´ self-reports. e e Single source methodology may inflate correlations. For example, it is possible that individuals with high negative affectivity (Watson & Clark, 1984) are both more likely to report: (a) “no one has helped me with my career”; and (b) “I dislike my job.” One of the few multi-source studies found that quality of mentoring relationship was related to prot´ g´ self-reported level of OCB but e e not to co-worker reported level of prot´ g´ OCB (Donaldson et al., 2000). We e e recognize the value of self-report data and the argument that the self sometimes is in the best position to report his or her own behavior or experience, especially on subjective outcomes such as job, career, and life satisfaction; organizational commitment; and intention to turnover (Howard, 1994), but additional sources of information would contribute to the literature. New areas of research. While the question of whether mentoring leads to positive outcomes is the primary focus of the prot´ g´ outcome literature, recent research has e e begun to address other questions. These include how (through what mechanisms)

Mentoring Research

51

the mentoring process impacts outcomes, how having more than one mentor might impact outcomes, and what negative, rather than positive, experiences may occur within mentoring relationships. To date, the mechanisms through which mentoring works (i.e. whether there are any mediators or moderators between mentoring and distal outcomes like job satisfaction) have not been well examined; very few studies were found looking at this issue. Day and Allen (2002) examined whether career motivation and self-efficacy mediated the relationship between mentoring provided and prot´ g´ outcomes. They found that career motivation fully mediated e e the relationship between career mentoring received and self-reported performance effectiveness, but they found only partial support for self-efficacy as a mediator of this relationship. In a second study, Lankau and Scandura (2002) found that prot´ g´ learning mediated the relationship between mentoring functions and role e e ambiguity and job satisfaction. The impact of having more than one mentor on outcomes is also being examined. While findings are mixed, it does appear that having a set of mentors is related to a number of positive outcomes for prot´ g´ s. For example, Higgins and e e Thomas (2001) found that for intention to remain and work satisfaction, the quality and character of lawyers’ multiple developmental relationships explained variance above and beyond that explained by their primary mentor. Using the same sample, Higgins (2000) found that more mentoring relationships and more mentoring received were correlated with higher work satisfaction. Baugh and Scandura (1999) found that number of mentors was related to enhanced career expectations, and lower work role ambiguity. Finally, Peluchette and Jeanquart (2000) found that early in faculty members’ careers having mentors from multiple sources was positively related to both objective and subjective career success and that in mid-career it was related to objective success. However, Riley and Wrench (1985) did not find a significant difference between female lawyers who had a group of mentors and those who had not had mentors in terms of perceived success or satisfaction. However, having a single, primary mentor may have some advantages over having multiple mentors. For example, Baugh and Scandura (1999) found that having more than one mentor was related to increased role conflict possibly due to conflicting advice given by different mentors. Similarly, Fagan and Walter (1982) found that having multiple mentors was related to higher job burnout for a sample of police, nurses and teachers when compared to individuals without a mentor and individuals with only one mentor. Higgins (2000) also found that having one person who provided high levels of psychosocial support was more highly related to work satisfaction than any other combination of number of mentors and level of support. Finally, Riley and Wrench (1985) compared female lawyers that reported receiving at least a given level of career mentoring from one mentor to those that reported receiving at least that same level of career mentoring from two

52

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

or more supportive individuals. This study showed that the lawyers who received support from two or more individuals had lower levels of general satisfaction than those with one mentor. Empirical research has also begun to attend more closely to the possibility that negative experiences occur in mentoring relationships. Two empirical studies were identified that explicitly examined negative or dysfunctional experiences of mentoring. In a study of 50 faculty, graduate and undergraduate students where 26 reported some type of conflict or hurt resulting from their relationship with their mentor (Kalbfleisch, 1997), 16 conflict events were identified and categorized using factor analysis. This factor analysis found four types of conflict events: disagreement (e.g. the mentor and prot´ g´ disagreed on ideas), embarrassment e e (e.g. mentor embarrassed or criticized the prot´ g´ ), negativity (e.g. mentor said the e e prot´ g´ made the mentor look bad), and request (e.g. the mentor asked the prot´ g´ e e e e for help on a project). Similarly, in a study of 156 prot´ g´ s where 84 reported at least e e one negative mentoring relationship, Eby, McManus, Simon and Russell (2000) developed a taxonomy of negative mentoring experiences by performing a content analysis of responses to questions about prot´ g´ s most negative mentoring experie e ences. The taxonomy consists of five broad categories including, from most to least common, problems with mentor/prot´ g´ match (i.e. dissimilar values, working e e styles and personalities), mentor distancing behavior (i.e. neglect of the prot´ g´ or e e focus on outcomes for his/herself rather than for the prot´ g´ ), mentor manipulative e e behavior (i.e. use of inappropriate power, taking inappropriate credit, or deception of the prot´ g´ ), lack of mentor expertise (i.e. lack of interpersonal or technical e e competence), and general dysfunctionality (i.e. the mentor had a negative attitude and/or personal problems). While these categorizations are helpful, since both studies found that over 50% of the participants reported having at least one negative experience or relationship at some time point (Eby et al., 2000; Kalbfleisch, 1997), they are only a starting point and research examining the antecedents and the consequences of dysfunctional experiences in mentoring relationships is still required. Mentor Outcomes Whether mentors receive positive outcomes from mentoring has also been a question of interest. In their seminal book on adult development, Levinson, Darrow, Klein, Levinson and McKee (1978) stated, “There is a measure of altruism in mentoring . . . But much more than altruism is involved: the mentor is doing something for himself. He is making productive use of his own knowledge and skill in middle age. He is learning in ways not otherwise possible” (p. 253). Similarly, in her well-known book on mentoring, Kram (1985b) suggested, “While helping a young adult establish a place in the adult world of work, an individual benefits from

Mentoring Research

53

providing support and guidance. Through helping others, a mentor gains internal satisfaction, (and) respect for his or her capabilities as teacher and advisor” (p. 3). While positive outcomes have been proposed for mentors, only a few studies have empirically examined these outcomes. From these studies it does appear that being a mentor is related to positive outcomes, although the effect size is not known. Studies on mentor outcomes have rarely focused on the types of outcomes examined among prot´ g´ s, such as to what extent mentoring leads to improved e e mentor compensation and promotions. Only two studies with a total sample of less than 500 individuals have examined these more distal types of outcomes. In the first study, faculty indicated that being a mentor positively affected their career satisfaction (Johnson, Yust & Fritchie, 2001). In the second, individuals who had been mentors had higher levels of self-reported career success and career satisfaction, as well as higher incomes (Collins, 1994). Limitations in this research are similar to those within the prot´ g´ research, as we cannot separate the impact of mene e toring on these outcomes versus the impact of other, pre-existing characteristics of the mentor. Instead, studies on mentor outcomes have typically focused on documenting “mentoring benefits,” and have found that mentors receive both intrinsic (e.g. personal satisfaction when helping others) and extrinsic (e.g. more supporters in the organization) benefits. From interviews with over 100 executives, Zey (1984) recognized four categories of benefits that mentors receive: career enhancement, intelligence/information, advisory role (where the prot´ g´ advises the mentor) e e and psychic rewards. Similarly, from her interviews with mentors, Kram (1985b) identified three categories of benefits for mentors: confirmation and support from the prot´ g´ , intrinsic satisfaction for helping a younger person develop, and e e recognition and respect from others. Expanding on these benefit categories, Allen, Poteet and Burroughs (1997) distilled the benefits that 27 mentors reported during interviews into four categories: builds a support network (e.g. develops close relationships that may result in future benefit for the mentor), self-satisfaction (e.g. a general satisfaction in helping others), job-related self-focused (e.g. helps increase mentor’s learning, visibility, or recognition) and job-related other-focused (e.g. helps to develop a competent workforce). These categories incorporate all of the benefits discussed by Zey (1984) and Kram (1985b), and introduce one category they did not identify, “job-related other-focused.” Support for Allen, Poteet and Burrough’s (1997) categorization comes from research that has looked at mentor benefits. Prot´ g´ s have cited loyalty to the mentor e e (Burke, 1984) and mentors have listed professional and personal support (Busch, 1985) as benefits for mentors engaged in mentoring, examples of Allen, Poteet and Burrough’s (1997) “builds a support network” category. Prot´ g´ s (Burke, 1984) e e and mentors (Busch, 1985) listed a sense of pride in seeing individuals develop

54

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

as a benefit for mentors, and executives agreed mentoring was a “rewarding experience” (Ragins & Scandura, 1999), examples of the “self-satisfaction” category. Examples of the “job-related self-focused” category are found relatively frequently in the mentoring literature. For instance, Burke (1984) found that prot´ g´ s mentioned “effectively performing subordinates . . . the perspective and e e energetic drive of youth . . . and recognition by others for effective mentoring” (p. 361) as benefits for mentors. In addition, Mullen and Noe (1999) found that mentors and prot´ g´ s agreed that mentors sought and received technical, referent e e and normative information, job performance and social feedback from their prot´ g´ s, and Busch (1985) found that mentors believed that mentoring enhanced e e their own careers by motivating them to keep up to date with their field. Finally, Ragins and Scandura (1999) found that executives agreed that being a mentor improved their job performance. No support for the “job-related other-focused” benefit category was found, although this may be because these benefits are often attributed to the organization rather than the mentor. Allen, Poteet and Burrough’s (1997) also identified four negative aspects of mentoring: time requirements, favoritism issues (e.g. it can feel uncomfortable to pay special attention to one particular person), abusive relationships (e.g. the prot´ g´ could use the relationship in a destructive fashion), and feelings of failure e e if the relationship did not go well. These are similar to problems previously identified by Zey (1984), although he also identified risk to the mentor’s reputation and exposure of self (i.e. mentors must share their experiences including their mistakes) as issues for mentors. Based upon the small body of research that has examined the mentor’s perspective, it appears that mentors do receive benefits from being a mentor. However, there also appear to be drawbacks to the relationship that must be considered. Whether mentoring translates into outcomes for mentors similar to those examined for prot´ g´ s is not clear from the scant evidence currently available. e e Organization Outcomes The organizational outcomes of mentoring stem from mentor and prot´ g´ e e outcomes. As such, outcomes relevant to the organization have received attention at an individual level of analysis via the research focused on prot´ g´ and mentor e e outcomes. Very little attention has been given to mentoring at an organizational, or aggregate, level of analysis, however. For example, do companies where mentoring is frequent have higher retention rates? Are these companies more successful by other measures? Zey’s (1984) interviews with over 100 executives provide an important preliminary insight into possible organizational level outcomes. Based on the interviews, Zey (1984) identified seven organizational outcomes of mentoring: employee integration, reduction in turnover, organizational communication,

Mentoring Research

55

management development, managerial succession, productivity and socialization to power. Research Priorities: Mentoring Outcomes We see five critical priorities for research on mentoring outcomes. First, and foremost, researchers must differentiate between individuals with formal and informal mentors in their research. To the extent that formal mentoring is different from informal mentoring, outcomes may be biased or unclear when individuals with both types of mentors are included in a study without distinction. Second, future research on mentoring outcomes should carefully examine the extent to which mentoring contributes to career success outcomes above and beyond pre-existing prot´ g´ characteristics. Both longitudinal research and the careful e e use of relevant control variables will be useful in this regard. Third, more research examining the mechanisms through which mentoring is related to positive outcomes is required. The mentoring literature would benefit from a clearer delineation of factors that mediate the relationship between mentoring received and more distal outcomes, such as increased compensation or promotions. We will propose a framework of proximal and distal outcomes in the second half of this paper. This progression of “digging deeper” into a criterion space is almost identical to that which has occurred in the socialization literature after Saks and Ashforth (1997) made a similar call for the distinction between proximal and distal outcomes of socialization and the understanding of the mediational paths. A fourth area where more work is needed is in the area of mentor and organizational outcomes. For example, little is currently known about what work-related outcomes, such as improved job performance, if any, that mentors receive from mentoring. In addition, research that examines exactly what organizations gain from mentoring and the size of that gain is needed, with careful attention paid to levels of analysis issues. Finally, continued research examining the impact of multiple mentors and possible negative experiences that stem from mentoring would also be highly useful.

Mentoring and Diversity Diversity and mentoring have been the focus of considerable research attention. In this section, research on the role of gender and race in mentoring is reviewed, combining findings from research focused on prot´ g´ s, mentors, and dyads to e e address critical questions raised in the literature. In a following section, we expand this discussion into a deeper understanding of underlying individual and

56

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

job/career history characteristics of the prot´ g´ and mentor that are associated e e with mentoring received. Mentoring and Gender Driven by the search to understand the glass ceiling phenomenon in organizations, mentoring and gender have received substantial attention. Indeed, our literature search identified over 50 articles and two recent, comprehensive reviews (O’Neill, 2002; Ragins, 1999) on this topic. The logic underlying the possible connection between mentoring and the glass ceiling is that women may be at a disadvantage in their careers in comparison to men if they are less likely to obtain a mentor, or if when they are mentored, the nature of that mentoring is different and leads to different outcomes. Compelling questions have been investigated in this literature, including: (1) do women report having had fewer mentors than men?; (2) when in a mentoring relationship do women receive different types or levels of mentoring than men?; and (3) do women reap fewer benefits from mentoring than men? An initial question is whether women are less likely to have mentors than men. One proposition is that women may encounter more barriers to obtaining a mentor than men and may, therefore, be less likely to have a mentor (Noe, 1988b; Ragins, 1989). However, Ragins’ (1999) conclusion, based on her recent review in this area, is that most research has found that women are as likely to report that they have had a mentor as men. Ragins’ conclusion has been supported by research that has become available since the writing of her review (e.g. see Hubbard & Robinson, 1998; McGuire, 1999; Ragins & Cotton, 1999; Witt Smith, Smith & Markham, 2000). For example, in a study of administrators at institutions of higher education, women actually were more likely to have had a mentor early in their career than men (Hubbard & Robinson, 1998). A second salient question in this literature addresses whether women who are in a mentoring relationship receive different amounts or kinds of mentoring than men. Differences may occur because women require different functions to succeed in organizations or because women are more likely to be in cross-gender relationships and, given the dynamics of these relationships, less mentoring may be provided (Ragins, 1997). Research in this area has not been definitive (Ragins, 1999). For example, some studies have found that women receive different amounts of career-related mentoring than men, with some finding that women receive less than men (Koberg, Boss, Chappell & Ringer, 1994; McGuire, 1999) and one finding that women receive more than men (Sosik & Godshalk, 2000b), while others have not found a difference (Burke, 1984; Burke, McKeen & McKenna, 1990; Noe, 1988a; Ragins & Cotton, 1999; Ragins & McFarlin, 1990; Scandura & Williams, 2001; Thomas, 1990; Turban & Dougherty, 1994; Witt Smith et al., 2000). In addition, some studies have found that women receive

Mentoring Research

57

higher levels of psychosocial mentoring than men (Burke, 1984; McGuire, 1999; Noe, 1988a), while others have not found a difference (Koberg et al., 1998; Ragins & McFarlin, 1990; Scandura & Williams, 2001; Sosik & Godshalk, 2000b; Thomas, 1990; Turban & Dougherty, 1994; Witt Smith et al., 2000). Finally, one study found that women receive more role modeling than men (Sosik & Godshalk, 2000b), while others have not found a difference (Burke, 1984; Ragins & Cotton, 1999; Ragins & McFarlin, 1990; Scandura & Williams, 2001). Because of these conflicting findings, research has examined whether there are moderators to the relationship between prot´ g´ gender and mentoring received. e e Specifically, are there certain conditions under which women receive different types or levels of mentoring when compared to men? Initial research suggests that the gender of the mentor, the duration of the relationship, and who initiated the relationship show promise for further investigation. For example, Sosik and Godshalk (2000b) found that female prot´ g´ s with male mentors reported more e e career development functions than female prot´ g´ s with female mentors, and e e Ragins and Cotton (1999) found that female prot´ g´ s with female mentors tended e e to engage in more social activities with their mentor than did female prot´ g´ s with e e male mentors. Turban, Dougherty and Lee (2002) found that the duration of the mentoring relationship moderated the level of functions received when comparing mixed gender and same gender dyads in a graduate school setting. Specifically, same gender dyads when compared to mixed gender dyads initially had higher levels of exposure, visibility and sponsorship, psychosocial mentoring, and protection. As the relationship progressed, however, the levels of these functions increased in mixed gender dyads to the point that higher levels were eventually found in mixed gender dyads than in same gender dyads. Finally, Scandura and Williams (2001) found that who initiated the relationship moderated the quantity of mentoring received when comparing male and female prot´ g´ s. Female prot´ g´ s e e e e reported more mentoring than male prot´ g´ s when either the mentor or both the e e prot´ g´ and the mentor initiated the relationship, but less when the prot´ g´ initiated e e e e the relationship. A third important question is whether women reap fewer objective or subjective work-related outcomes from mentoring (e.g. job satisfaction, promotions) than men. Most of the research to date has not found differences between outcomes for male and female prot´ g´ s (Ragins, 1999). Research is suggestive of the possibility, e e however, that mentor gender may be important to consider. For example, Ragins and Cotton (1999) found that having prior male mentors was related to higher compensation levels. Similarly, two studies found that female prot´ g´ s with male e e mentors earned significantly more (Wallace, 2001) or had higher career attainment scores, a combination of compensation and position prestige (Bahniuk, Hill & Darus, 1996), than female prot´ g´ s with female mentors. e e

58

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

The combination of mentor gender with mentor race has also found to be associated with prot´ g´ compensation. Dreher and Cox (1996) and Dreher e e and Chargois (1998) found that prot´ g´ s with Caucasian male mentors were e e compensated more than individuals without mentors, but that there was no statistical difference between prot´ g´ s with other types of mentors and individuals e e without mentors. These results are supported by a study of health care executives which found that prot´ g´ s with white male mentors had higher salaries than e e prot´ g´ s with other types of mentors (Weil & Kimball, 1996). e e A strong caveat to this discussion is that studies examining compensation did not consider other important explanatory variables in their analyses. One possible explanation for the difference between having a male and a female mentor may be that female mentors are, on average, at lower levels of the organization than male mentors given that women are under-represented at senior organizational levels (Federal Glass Ceiling Commission, 1995). The organizational level of mentors likely impacts the amount of power that mentors have and their ability to influence outcomes and provide rewards, like compensation, for their prot´ g´ s, so it is an e e important variable to control for when looking at differences in compensation. Only one study (Dreher & Cox, 1996) partially considered this explanation, although they looked at the difference in levels between the prot´ g´ and the e e mentor, rather than the absolute level of the mentor. In conclusion, it does not appear that women report having had fewer mentors than men. In contrast, we cannot definitively answer whether women in mentoring relationships receive different amounts or kinds of mentoring functions than men because of inconsistent research findings. Finally, the answer to whether women gain fewer benefits than men from mentoring is unclear. It does appear that there may be a difference between compensation for prot´ g´ s based upon the gender or e e the race and gender of the mentor, but we do not know why this occurs or whether there are differences for other outcome variables. Mentoring and Race Paralleling the impact of the glass ceiling in organizations for women, few minorities hold top positions in organizations (Federal Glass Ceiling Commission, 1995). This has led researchers to ask questions (similar to those asked for gender and mentoring) about the role of mentoring and career success among minorities in organizations. Critical questions that have been posed include: (1) do minorities report having had fewer mentors than Caucasians?; (2) when in a mentoring relationship do minorities receive different amounts or kinds of mentoring than Caucasians?; and (3) do minorities reap fewer career-related benefits from mentoring than Caucasians? While there has been considerable research dedicated to understanding the relationship between mentoring and

Mentoring Research

59

gender, less research has focused on mentoring and race. We found less than 20 empirical studies looking at race, compared to over 50 looking at gender. One basic question is whether minorities are less likely to have a mentor than Caucasians. Most studies have found that minorities are not less likely than Caucasians to have a mentor (Dreher & Cox, 1996; Greenhaus, Parasuraman & Wormley, 1990; McGuire, 1999; Mobley, Jaret, Marsh & Lim, 1994; Steinberg & Foley, 1999; Thomas, 1990; Witt Smith et al., 2000). The samples from these studies were quite diverse, including MBA alumni (Dreher & Cox, 1996), managers (Greenhaus et al., 1990; Thomas, 1990), managers and non-managers (McGuire, 1999), Army personnel (Steinberg & Foley, 1999), professionals (Mobley et al., 1994) and university faculty (Witt Smith et al., 2000), lending generalizability to the conclusion. However, one study did find that African-American public accounting employees were less likely to have informal mentors than Caucasians (Viator, 2001b). There are two possible explanations for this finding. First, as suggested by the author, this may be a special case due to the extreme lack of African-Americans at senior levels in public accounting firms. Second, given that this study distinguished between formal and informal mentors and found that African-Americans were more likely to have formal mentors than Caucasians but less likely to have informal mentors, this finding may reflect the lack of clear delineation between formal and informal mentors in the majority of the literature. A second important question is whether minorities report receiving different types and levels of mentoring functions. In terms of overall levels of mentoring received, we can not draw any conclusions as there has been little research focused on this question and results to date have been inconsistent. While two studies found that African-Americans and Caucasians received similar amounts of mentoring (Blake-Beard, 1999; Steinberg & Foley, 1999), one study found that minorities received less mentoring (Cox & Nkomo, 1991). In terms of specific mentoring functions received, studies have generally found that minorities do not report less career mentoring than Caucasians. The majority of the studies that have investigated the amount of career functions received by prot´ g´ s have either found no difference between Caucasian and Africane e American prot´ g´ s (Thomas, 1990; Viator, 2001b) or no difference between e e homogeneous dyads and heterogeneous dyads, regardless of the gender-race mix (Witt Smith et al., 2000). In fact, one study found that minorities actually received more career mentoring than Caucasians (Koberg et al., 1994). It is not clear whether there is a difference in terms of psychosocial mentoring received, because research results are varied. Two studies found that minorities reported receiving either less psychosocial support (Koberg et al., 1998) or less social support (Viator, 2001b) than Caucasians, but Thomas (1990) found that prot´ g´ race was not related to receipt of psychosocial mentoring. At the dyad e e

60

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

level, two studies have found that prot´ g´ s in same race dyads reported higher e e levels of psychosocial functions than prot´ g´ s in cross race dyads (Koberg et al., e e 1998; Thomas, 1990). These findings are supported by Viator’s (2001b) results that African-American mentors provided more social support and role modeling than other mentors to African-American prot´ g´ s. However, Witt Smith et al. (2000) e e did not find any difference in the amount of psychosocial support reported in dyads that were homogeneous in terms of gender and race compared to heterogeneous dyads. A final critical question is whether minorities receive fewer objective or subjective work-related outcomes from mentoring than Caucasians. Similar to the finding for gender and mentoring, research is suggestive of the possibility that the combination of mentor race and gender may be important to consider. As discussed above, studies have found that Caucasian male mentors are associated with higher compensation for prot´ g´ s, when compared to prot´ g´ s with other e e e e types of mentors and to individuals without mentors (Dreher & Chargois, 1998; Dreher & Cox, 1996; Weil & Kimball, 1996). A caveat to these findings is that we cannot assume a causal relationship as the studies did not control for important alternative explanatory variables, such as the mentor’s level in the organization. Race differences in other objective and subjective outcomes have been examined in a limited number of studies, yielding inconsistent results. In an analysis of university faculty, mentoring was significantly related to subjective outcomes of intent to turnover and affective commitment for Caucasians (Witt Smith et al., 2000) but not for minorities. However, in a study that looked at networking groups and African-American MBA alumni, Friedman, Kane and Cornfield (1998) found that having a mentor was significantly related to career optimism for these African-Americans. In addition, Blake-Beard (1999) found that for satisfaction with career progress, which was associated with mentoring, there was no significant difference between Caucasian and African-American female prot´ g´ s. e e In conclusion, it does not appear that minorities report having had fewer mentors than Caucasians, although minorities may have had fewer informal mentors. In terms of amounts or kinds of mentoring, it appears from a limited number of studies that there may be no difference between Caucasians and minorities in the level of career-related mentoring reported, although results for psychosocial mentoring are mixed and so no conclusion can be drawn. Finally, the answer to whether minorities gain fewer benefits than Caucasians from mentoring is unclear. It does appear that there may be a difference between compensation for prot´ g´ s based upon the race and gender of the mentor, but we do not know e e why this occurs or whether there are differences for other outcome variables.

Mentoring Research

61

Research Priorities: Mentoring and Diversity Diversity has primarily been examined in terms of gender and race in the mentoring literature. While other types of diversity may be interesting to study, including age and sexual orientation, more mentoring research examining gender and race is still necessary. For mentoring and gender, targeted research is needed that focuses on whether there is a link between gender and outcomes such as compensation, and why such linkages may occur. More research on moderators of gender/mentoring relationships is especially appealing. Are there situations or conditions under which one gender receives less mentoring than another? When conducting this research, potentially confounding factors that correlate with gender, and may be the underlying cause of any differences that might be found, must be carefully considered. For example, as O’Neill (2002) stated in her review of gender and race in mentoring, “what appear(s) to be stereotypical gender differences may be misleading when the mentor’s gender is confounded with other factors, such as rank” (p. 7). Given the paucity of research on mentoring and race, the first and most important research priority is more research. Questions requiring further research include whether minorities receive equivalent mentoring to Caucasians, and whether the outcomes of mentoring are different for minorities and Caucasians. In this context, researchers must attend to three critical issues. First, researchers must differentiate between formal and informal mentoring. It is possible that the proportion of formal mentoring relationships to total mentoring relationships is higher for minorities than for Caucasians and this difference may distort empirical results. Second, it is important that racial groups in addition to African-Americans be examined. Most research to date has studied African-Americans (Blake-Beard, 1999; Bridges & Perotti, 1993; Cox & Nkomo, 1991; Dreher & Chargois, 1998; Friedman et al., 1998; Greenhaus et al., 1990; Kalbfleisch & Davies, 1991; Steinberg & Foley, 1999; Thomas, 1990; Viator, 2001b; Weil & Kimball, 1996). Only one study was found that examined Asians (Goto, 1999), two looked at minority groups separately (Dreher & Cox, 1996; Koberg et al., 1998) and two used a variety of groupings depending on the analysis being conducted (Turban et al., 2002; Witt Smith et al., 2000). Some studies combined different minorities into one group (see Koberg et al., 1994; McGuire, 1999; Mobley et al., 1994; Turban et al., 2002; Witt Smith et al., 2000). Although we recognize the difficulty of finding sufficient samples of minorities to study, it should be recognized that the experiences of minority groups may not be similar. Finally, when conducting this research, potentially confounding factors that correlate with race (e.g. organizational level of prot´ g´ ), and may be the underlying cause of any differences e e that might be found, must be carefully considered to help understand why any differences appear.

62

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Other Correlates of Mentoring Researchers have investigated an informative array of prot´ g´ and mentor chare e acteristics as predictors of mentoring in addition to gender and race. Consistent with other research on mentoring, more attention has been devoted to prot´ g´ e e characteristics than mentor characteristics. Prot´ g´ Characteristics e e Knowledge of how prot´ g´ characteristics influence mentoring informs our undere e standing of the development and process of mentoring relationships. From a practical vantage point, this information may be useful in identifying employees who will flourish as prot´ g´ s or in assisting employees who may find it challenging to ese e tablish and maintain productive mentoring relationships. Questions about prot´ g´ e e characteristics that have been investigated include: (1) what motivates individuals to seek mentors?; (2) what characteristics do mentors seek in prot´ g´ s?; (3) what e e differentiates individuals who have mentors from those who don’t?; and (4) how are prot´ g´ characteristics related to mentoring received? Table 3 provides examples of e e the many prot´ g´ characteristics that have been studied across these four questions e e and introduces the general rationale for considering each variable for inclusion in the nomological network of mentoring. It should be noted that data for some of these variables (i.e. demographics and job/career history) have been included as control variables in several studies, rather than their being the focus of research. Characteristics associated with motivation to seek mentors. Relatively little research has been directed towards the first issue – understanding employees’ motivation to seek out mentors. Initial research indicated having a mentor was a universal task of early adulthood (Levinson et al., 1978), suggesting there were not individual differences in motivation to be a prot´ g´ . Darling (1986) reported being e e surprised when her interviews with successful business people revealed that some individuals were not interested in having a mentor. She determined individuals without mentors often had childhood experiences with adults marked by a lack of trust, respect, and/or relevance. Consistent with these findings, Kram (1988) suggested a number of variables that might affect the kind of developmental relationships individuals sought, including attitudes towards authority, conflict, and intimacy; values about work and learning; and self-competence perceptions. More recently, a few quantitative studies have investigated how personality traits, demographic characteristics, and job/career history variables relate to motivation to find a mentor. Two studies were located that examined how personality traits are related to behaviors directed towards initiating mentoring relationships (Aryee, Lo & Kang,

Mentoring Research

Table 3. Hypotheses on Prot´ g´ Characteristics and Mentoring: A Summary of the Informal Mentoring Literature. e e
Prot´ g´ Characteristic e e Competence Personality traits Negative affectivity Extraversion Self-monitoring Type A personality Locus of control Need for affiliation Need for autonomy Need for achievement Need for power Femininity Masculinity Self-esteem Learning goal orientation Attitudes Demographics Marital status Hypothesis Social exchange theory: Working with a more competent prot´ g´ has more e e short- and long-term benefits. Those high in negative affectivity will attempt to avoid the stress and tension likely to arise during the challenging opportunities provided by mentors. Due to their proactivity and sociability, those high in extraversion will be more involved in mentoring relationships. High self-monitors will be more aware of the value of mentoring for career success. Their drive to excel will lead those with a Type A personality to have mentoring relationships. Those with an internal locus of control are more likely to perceive that they can improve their skills. Mentoring relationships provide the opportunity to affiliate. Prot´ g´ s depend upon their mentors. e e Mentoring relationships are instrumental for achievement. Mentoring relationships are an avenue for attaining power. Data on mentoring and gender roles available from research on gender differences in career progression. Data on mentoring and gender roles available from research on gender differences in career progression. Individuals high in self-esteem will have the confidence to handle the challenging opportunities provided by mentors. Stronger learning goal orientation will be associated with greater appreciation of mentor functions and enhanced motivation to learn from the mentor. Individuals whose work and careers are more important to their identity will be more involved at work and in mentoring. Greater work-life conflict will limit married individuals’ capacity to participate in mentoring relationships. Sample Researchers Olian, Carroll and Giannantonio (1993) Turban and Dougherty (1994) Aryee, Lo and Kang (1999) Turban and Dougherty (1994) Aryee et al. (1999) Colarelli and Bishop (1990) Fagenson (1992) Fagenson (1992) Fagenson (1992) Fagenson (1992) Kirchmeyer (1998, 2002) Kirchmeyer (1998, 2002) Turban and Dougherty (1994) Godshalk and Sosik (in press) Aryee, Wyatt and Stone (1996) Research Volumea Small

Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Minimal Small

Olian et al. (1993)

Small

63

64 CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Table 3. (Continued )
Prot´ g´ Characteristic e e Education Hypothesis Education credentials may be used as a proxy for potential in allocating scarce developmental resources. Individuals with more education will be more involved in their jobs. Younger employees are less experienced and want more mentoring to address feelings of insecurity. Informal norms discourage older employees from being prot´ g´ s. e e Perceived similarity encourages senior managers to work with prot´ g´ s of e e higher SES. Primarily studied as a control variable; by definition those with greater experience are less likely to need mentoring. Negative, if the supply of mentors is smaller at higher levels. Positive, if mentoring is a reward or if the need for mentoring is greater. By definition, less experienced employees should be more likely than those with greater experience to be a prot´ g´ . e e Studied as a control variable. Since it may be a proxy for drive to excel or job involvement, the same arguments apply. Sample Researchers Whitely, Dougherty and Dreher (1992) Koberg, Boss, Chappell and Ringer (1994) Olian, Carroll, Giannantonio and Ferren (1988) Whitely et al. (1992) Whitely et al. (1992) Research Volumea Moderately large

Age

Large

SES Job/career history Tenure Organizational rank Work experience Continuity of work history Average hours worked
a

Small

Fagenson (1989) Whitely et al. (1992), Koberg et al. (1994) Olian et al. (1988) Dreher and Ash (1990) Whitely and Coetsier (1993), Whitely et al. (1992)

Moderate Moderately large Moderate Moderate Small

This column is intended to provide readers with a sense of how much attention has been devoted to studying different characteristics, rather than give an exact count of the number of independent samples. Categorization is based on the number of published studies on informal mentoring relationships that reported relationships between the predictor variable and mentoring criteria: minimal (1–3 studies), small (4–9 studies), moderate (10–19 studies), moderately large (20–29 studies); large (30 or more studies). A minimal number of studies included more than one independent sample; some samples were used in more than one study.

Mentoring Research

65

1999; Turban & Dougherty, 1994). Individuals with less negative affectivity, greater extraversion, higher self-monitoring, a propensity towards Type A personality, and greater self-esteem were more likely to report performing behaviors likely to lead to a mentoring relationship. Across the two studies, the relationship between locus of control and initiating behaviors diverged. Multivariate analyses suggested initiating behaviors mediate the relationship between personality traits and mentoring received (Aryee et al., 1999; Turban & Dougherty, 1994). A few studies have scrutinized how demographics relate to employees’ motivation to be a prot´ g´ . Based on one study, it appears that employees who e e are not married put more effort into initiating mentoring relationships (Turban & Dougherty, 1994). Evidence on whether education is associated with initiating behaviors is mixed, with no relationship reported in one study (Turban & Dougherty, 1994) and better educated individuals reporting more initiating behaviors in another (Aryee et al., 1999). The relationship between age and motivation to seek a mentor is also ambiguous. Age does not appear to be related to fears about initiating a mentoring relationship, perceptions of others’ willingness to provide mentoring, or concerns about how others’ view the relationship (Burke & McKeen, 1997; Ragins & Cotton, 1991). However, despite experimental evidence that younger individuals find having a mentor more appealing (Olian, Carroll, Giannantonio & Feren, 1988), a single study found older employees report more behaviors directed towards initiating mentoring relationships (Aryee et al., 1999). This study also suggested SES is not related to initiating behaviors. Based on this body of research, it is difficult to firmly connect many demographic characteristics of employees with their motivation to have mentors. Researchers also have explored the link between employees’ career or job history variables and their motivation to be prot´ g´ s. Ragins and Cotton (1991) found eme e ployees with more organizational tenure and higher rank perceived fewer barriers to gaining access to mentors. On the other hand, neither tenure nor rank was associated with fears about initiating relationships. Unexpectedly, a couple of studies found work experience was not related to interest in having a mentor (Olian et al., 1988) or attempts to initiate mentoring relationships (Turban & Dougherty, 1994). In conclusion, preliminary evidence seems to suggest there are individual differences in motivation to seek a mentor. Based on only single studies, some employee attributes have not been linked to efforts to find a mentor (e.g. SES, work experience). Other research has found personality traits (i.e. negative affectivity, extraversion, self-monitoring, Type A personality, and self-esteem) and one demographic variable (i.e. marital status) account for variance in behaviors directed towards initiating informal mentoring relationships. Two studies suggest these behaviors mediate the relationship between employee personality traits and mentoring received.

66

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Prot´ g´ characteristics sought by mentors. Researchers have devoted relatively e e little attention to identifying the attributes mentors seek in prot´ g´ s. Based on an e e analysis of interviews with 27 mentors, Allen, Poteet and Borroughs (1997) identified six higher-order factors that attract mentors to prot´ g´ s. Mentors look for e e prot´ g´ s who are competent, motivated, display a strong learning orientation, and e e have certain personality traits (e.g. people-oriented, honest, confident, dependable, patient, and flexible). In addition, employees who arouse mentors’ desire to help and remind mentors of themselves are appealing as prot´ g´ s. In a second e e study, Allen and her colleagues (Allen, Poteet & Russell, 2000) found mentors (particularly female mentors) were more likely to base their selection of prot´ g´ s e e on perceived ability and motivation than on perceived need for help. An experiment supports the importance of employees’ competence in their attractiveness to potential mentors. Bank managers anticipated greater rewards and indicated greater interest in providing mentoring functions when a hypothetical scenario described a high-performing, rather than a moderately-performing, potential prot´ g´ (Olian, Carroll & Giannantonio, 1993). Another finding of this experiment e e was that managers anticipated greater rewards and intended to provide more career mentoring to married men and unmarried women. Although there have been few studies conducted on the topic, ability and motivation stand out as desirable prot´ g´ characteristics. Yet, mentors also seem to have e e interest in working with prot´ g´ s who need help, remind mentors of themselves, e e have a strong learning orientation, and demonstrate key personality traits. Characteristics differentiating employees with and without mentors. In attempting to differentiate employees with and without mentors, researchers have investigated a variety of variables. Factors that have not distinguished those with and without mentors include early promotion rate (Fagenson, 1989), academic performance in post-secondary school (Judge & Bretz, 1994; Laband & Lentz, 1995), LSAT scores (Laband & Lentz, 1995), locus of control (Colarelli & Bishop, 1990), need for affiliation, need for autonomy (Fagenson, 1992), femininity (Kirchmeyer, 1998, 2002), and career continuity (Judge & Bretz, 1994; Kirchmeyer, 1998). In contrast, prot´ g´ s appear to be higher on need for achievement, need for power e e (Fagenson, 1992), and job involvement (Aryee & Chay, 1994). Job tenure may have a weak, negative relationship with access to mentors (Judge & Bretz, 1994). The relationship of masculinity – the constellation of attributes traditionally comprising the male gender role (e.g. assertiveness, individualism, and instrumentality) – to having a mentor is unclear. Concurrent data suggests a positive relationship, but longitudinal data indicates no relationship (Kirchmeyer, 1998, 2002). Evidence also is mixed regarding whether individuals with and without mentors differ in terms of marital status (Colarelli & Bishop, 1990; Judge & Bretz,

Mentoring Research

67

1994; Kirchmeyer, 1998; Laband & Lentz, 1995), education (Colarelli & Bishop, 1990; Judge & Bretz, 1994; Lankau & Scandura, 2002), age (Broadbridge, 1999; Colarelli & Bishop, 1990; Fagenson, 1989; Lankau & Scandura, 2002; Nielson et al., 2001; Ragins & Cotton, 1991; Ragins et al., 2000), SES (Dreher & Cox, 1996; Judge & Bretz, 1994), organizational tenure (Fagenson, 1989; Kirchmeyer, 1998; Lankau & Scandura, 2002; Ragins & Cotton, 1991; Ragins et al., 2000), organizational rank (Broadbridge, 1999; Fagan & Fagan, 1983; Fagan & Walter, 1982; Gaskill & Sibley, 1990; Ragins & Cotton, 1991; Ragins et al., 2000; Steinberg & Foley, 1999; Viator & Scandura, 1991), work experience (Kirchmeyer, 1998; Laband & Lentz, 1995), and hours worked per week (Judge & Bretz, 1994; Nielson et al., 2001). In conclusion, demographic and job/career history variables do not consistently seem to distinguish between employees with and without mentors. On the other hand, a limited number of studies suggest prot´ g´ s may differ from non-prot´ g´ s e e e e on a few personality traits (i.e. need for achievement and need for power), attitudes (i.e. job involvement), and career/job history variables (i.e. job tenure). Prot´ g´ characteristics and mentoring received. A relatively large proportion of e e research on prot´ g´ characteristics has explored how these attributes relate to the e e nature of mentoring received. A majority of these studies have been focused on the level of mentoring functions received, but a few have investigated frequency of mentor-prot´ g´ contact. Prot´ g´ attributes studied included individual differe e e e ence variables, such as abilities, personality, and attitudes, as well as demographic characteristics and job/career history variables. The few studies examining the link between prot´ g´ s’ competence and e e mentoring functions received have obtained mixed results (Godshalk & Sosik, in press; Mullen, 1998; Whitely et al., 1992). The inconsistencies may arise from differences in how competency was conceptualized and measured. In one sample, a moderate, positive correlation suggested prot´ g´ s perceived as more competent e e by their mentors tend to receive more mentoring functions (Mullen, 1998; Mullen & Noe, 1999). In a second study, employees’ job search strategies – argued to be an indicator of their interpersonal skills – were also positively and significantly associated with career mentoring (Whitely et al., 1992), albeit weakly. On the other hand, prot´ g´ s’ scores on cognitive tests were not associated with mentoring e e functions received (Godshalk & Sosik, in press). Three studies examining the relationship between prot´ g´ s’ personality traits e e and mentoring functions received have found some meaningful relationships. Higher prot´ g´ extraversion, greater self-esteem, less negative affectivity, and e e Type A personality have been associated with receiving more of at least some mentoring functions (Aryee et al., 1999; Turban & Dougherty, 1994). The

68

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

evidence on whether or not locus of control and self-monitoring are related to mentoring functions has diverged across studies (Aryee et al., 1999; Turban & Dougherty, 1994). Other traits, such as need for power, achievement, autonomy, and affiliation were unrelated to mentoring functions and the frequency of mentorprot´ g´ interactions (Fagenson, 1992). In related research on mentoring and e e individual differences, prot´ g´ s higher on learning goal orientation received more e e career, psychosocial, and role modeling functions (Godshalk & Sosik, in press). A handful of studies have reported relationships between job attitudes and mentoring received. Based on a single study, job involvement may have a small, positive relationship with psychosocial functions (Koberg et al., 1998). Four studies on the relationship between prot´ g´ attitudes (job involvement, e e work centrality, and career identity salience) and career mentoring have yielded diverging results (Aryee & Chay, 1994; Aryee, Wyatt & Stone, 1996; Cox & Nkomo, 1991; Whitely et al., 1991). More data are available on the relationship between demographic characteristics and mentoring received. Based on five studies, it is unclear whether or not marital status is related to the level of mentoring functions received (Burke & McKeen, 1996; Dreher & Ash, 1990; Seibert, Kraimer & Liden, 2001; Turban & Dougherty, 1994; Whitely et al., 1991). A relatively large number of studies have investigated the relationship between education and mentoring functions. Although there are exceptions (e.g. Godshalk & Sosik, 2000; Sosik & Godshalk, 2000a, b), most have found little relationship between prot´ g´ s’ education level and the mentoring e e functions they receive (e.g. Aryee, Wyatt & Stone, 1996; Lankau & Scandura, 2002; Lyness & Thompson, 2000; Whitely & Coetsier, 1993). This research, however, often has been based on samples with restricted levels of education. One study of more heterogeneous prot´ g´ s reported education level had a small positive e e relationship with psychosocial functions (Koberg et al., 1998). The relationship between prot´ g´ s’ age and mentoring functions has been studied relatively often, e e but this research has yielded contradictory results. Although a number of studies have found no relationship between age and mentoring functions (e.g. BlakeBeard, 1999; Chao, 1997; Gilbert & Ivancevich, 1999; Scandura & Ragins, 1993), at least five have reported older employees receive less of at least some mentoring functions (e.g. Cox & Nkomo, 1991; Fagenson, 1992; Ragins & McFarlin, 1990; Scandura & Williams, 2001; Sosik & Godshalk, 2000a). Several have indicated the opposite (Aryee et al., 1999; Burke & McKeen, 1997). Note that if the relationship between prot´ g´ age and mentoring functions received is, in fact, e e curvilinear, the observed pattern of results might arise if the age ranges of prot´ g´ s e e varied across studies. Research on the relationship between age and the frequency of mentor-prot´ g´ interactions also is divided (Fagenson, 1992; Mullen, 1998). e e A few studies have examined how prot´ g´ s’ SES relates to mentoring received. e e

Mentoring Research

69

Most (Aryee et al., 1999; Dreher & Ash, 1990; Whitely & Coetsier, 1993; Whitely et al., 1992), but not all (Blake-Beard, 1999), of this research suggests individuals with higher SES receive more mentoring. In general, the relationships between prot´ g´ s’ demographic characteristics and mentoring received remain unclear. e e A moderate proportion of the studies examining the question of how prot´ g´ e e characteristics relate to mentoring received have scrutinized variables that are a function of prot´ g´ s’ job or career history. These variables include tenure, e e rank, work experience, continuity of work history and average hours worked per week. Research linking organizational tenure to mentoring functions has not yielded completely consistent results. Although in more than half a dozen studies, prot´ g´ s’ organizational tenure has not been substantially associated with receive e ing mentoring (Aryee, Wyatt & Stone, 1996; Burke & McKeen, 1997; Fagenson, 1989; Koberg et al., 1994, 1998; Lankau & Scandura, 2002; Lyness & Thompson, 2000), both significant positive and significant negative relationships have been observed for at least some functions in other samples (Gilbert & Ivancevich, 1999; Higgins & Thomas, 2001; Wayne et al., 1999). Among expatriates, assignment tenure does not appear to be related to receiving mentoring (Bolino & Feldman, 2000). The relationship between prot´ g´ s’ organizational rank and mentoring e e functions also is unclear. A few studies have found prot´ g´ s with higher rank e e receive more career (Cox & Nkomo, 1991; Koberg et al., 1994; Whitely et al., 1992) and psychosocial (Koberg et al., 1998) mentoring functions; others suggest rank is unrelated to mentoring received (e.g. Bahniuk, Dobos & Hill, 1990; Bolino & Feldman, 2000; Fagenson, 1992; Gaskill & Sibley, 1990; Ragins & McFarlin, 1990; Scandura & Ragins, 1993). One study of accountants found prot´ g´ rank was e e unrelated to career mentoring, positively related to psychosocial mentoring, and had a curvilinear relationship with role modeling (Barker et al., 1999). Research relating work experience to mentoring functions also has been inconclusive. Although the results of a few studies suggest prot´ g´ s with more work experience e e may receive fewer career mentoring functions (Dreher & Ash, 1990; Turban & Dougherty, 1994; Whitely et al., 1992), more than half a dozen studies have found no evidence that work experience is related to mentoring functions (e.g. Aryee, Wyatt & Stone, 1996; Blake-Beard, 1999; Whitely et al., 1991). In two studies, a positive relationship was reported between professional experience and career mentoring (Higgins & Thomas, 2001; Lankau & Scandura, 2002). Although most studies have found that continuity of work history is unrelated to mentoring functions (e.g. Burke & McKeen, 1996; Dreher & Ash, 1990; Seibert, Kraimer & Liden, 2001; Turban & Dougherty, 1994; Whitely & Coetsier, 1993), there are exceptions suggesting those without career interruptions receive slightly more career mentoring (Aryee, Wyatt & Stone, 1996). The results of a handful of studies reporting relationships between the average hours prot´ g´ s work each week and e e

70

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

career mentoring have been mixed (Aryee, Wyatt & Stone, 1996; Nielson et al., 2001; Wayne et al., 1999; Whitely & Coetsier, 1993; Whitely et al., 1991, 1992). In conclusion, a few studies have hinted at intriguing relationships between mentoring functions received and several prot´ g´ characteristics, including e e ability, personality, and learning orientation. Results regarding demographic and job/career history variables and mentoring functions received have been inconsistent. Differences in range restriction across samples may account for some of the variability in these results. For example, studying a cohort of graduates from one or two business schools constrains the possible values of education, and, to a lesser extent related variables, such as age, tenure and work experience, attenuating correlations with mentoring functions. In future research, careful attention to the impact of sampling procedures on range restriction may be helpful in clarifying how demographic and job/career history variables relate to the nature of mentoring received. Mentor Characteristics A comprehensive understanding of the role of mentor characteristics in mentoring relationships would be of tremendous practical value, enabling prot´ g´ s to seek out e e effective mentors, permitting seasoned employees to assess their capacity to serve as mentors, and informing organizations how to select and train mentors. However, only a small proportion of the literature on mentoring has considered mentors’ attributes. Mentor characteristics studied have included competence (i.e. abilities, knowledge, and skills), personality, self-esteem, demographics, and job/career history variables, including past experiences in mentoring relationships. Table 4 provides an overview of hypotheses regarding these attributes. Key questions that have been explored include: (1) what characteristics do prot´ g´ s seek in mentors?; e e (2) what affects experienced employees’ motivation to serve as mentors; and (3) how do experienced employees’ characteristics relate to the mentoring they provide? The following review is organized around these questions. Mentor characteristics sought by prot´ g´ s. The limited information available suge e gests prot´ g´ s consider a wide array of characteristics when seeking a mentor. From e e surveys of retail managers, Gaskill (1991) concluded the characteristics prot´ g´ s e e seek in mentors include abilities (ability to develop subordinates, take risks, and share knowledge), knowledge (related to the organization, individuals in the organization, and to organizational power), personality traits (open-mindedness, upwardly oriented), and job membership characteristics (high position in the organization, rank/status in the organization, and respect from others). This information may be weighed and combined in a complex fashion by employees seeking mentors. In three experiments, Olian et al. (1988) provided undergraduates with a

Mentoring Research

Table 4. Hypotheses on Mentor Characteristics and Mentoring Provided: A Summary of the Informal Mentoring Literature.
Mentor Characteristic Competence Hypothesis Experienced employees with better skills (e.g. interpersonal) will be more appealing as mentors because they will be better able to serve mentoring functions. Included as a control variable. Senior employees who tend to be in a good mood would be seen as more approachable. Mentors high in self-monitoring will be more likely to seek information from their prot´ g´ s. e e Experienced employees with a more internal locus of control would be more likely to perceive they could have a positive impact on others’ professional growth. Due to identification with potential prot´ g´ s’ desire to advance and e e the possibility of improving their own power, those higher in upward striving will be more willing to mentor others. Being generous, kind, and helpful, those higher in altruism would be more motivated to mentor. Individuals high in self-esteem and those who are experiencing fewer career challenges will have more psychological resources to devote to mentoring. Due to their own love of learning and mastery focus, those higher on learning goal orientation will provide more mentoring. Sample Researchers Olian, Carroll, Giannantonio and Ferren (1988) Research Volumea Minimal

Personality traits Negative affectivity Positive affectivity Self-monitoring Locus of control

Aryee, Chay and Chew (1996) Mullen and Noe (1999) Allen, Poteet, Russell and Dobbins (1997) Allen, Poteet, Russell and Dobbins (1997) Aryee, Chay and Chew (1996) Kram (1988), Aryee, Chay and Chew (1996) Godshalk and Sosik (in press)

Minimal Minimal Minimal Minimal

Upward striving

Minimal

Altruism, helpfulness Self-esteem/Career challenges Learning goal orientation

Minimal Small

Minimal

71

72

Table 4. (Continued )
Mentor Characteristic Demographics Education Hypothesis Sample Researchers Research Volumea

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Age

Better educated individuals would be more qualified to serve as mentors and may have more familiarity with mentoring, leading them to feel more comfortable in the role. Younger employees seek mentors; mid-career employees seek prot´ g´ s to gain a sense of fulfillment from teaching others. e e Employees nearing retirement may be less interested in mentoring as they begin to withdraw from work. Employees with more tenure may feel more qualified to serve as mentors. Employees higher in the organizational hierarchy are likely to have more experience and feel more comfortable as mentors. More powerful senior employees would be sought out as mentors because they would be able to help prot´ g´ s’ careers more. e e Included as a control variable or in exploratory study.

Allen, Poteet, Russell and Dobbins (1997) Levinson, Darrow, Klein, Levinson and McKee (1978)

Small

Moderate

Job/career history Tenure Organizational rank Power Span of control

Ragins and Cotton (1993) Ragins and Cotton (1993) Olian et al. (1988) Tepper, Brown and Hunt (1993), Burke, McKeen and McKenna (1993) Ragins and Cotton (1993)

Small Moderate Minimal Minimal

Experience in mentoring relationships

Those who have experienced the benefits of being a prot´ g´ and/or e e mentor will see providing mentoring as more valuable.

Moderate

a This column is intended to provide readers with a sense of how much attention has been devoted to studying different characteristics, rather than give

an exact count of the number of independent samples. Categorization is based on the number of published studies on informal mentoring relationships that reported relationships between the predictor variable and mentoring criteria: minimal (1–3 studies), small (4–9 studies), moderate (10–19 studies), moderately large (20–29 studies); large (30 or more studies). A minimal number of studies included more than one independent sample; some samples were used in more than one study.

Mentoring Research

73

transcript of an interaction between a manager and a subordinate and asked them to rate how attracted they were to the manager as a mentor. Managers with stronger interpersonal skills were seen as more appealing. Managers’ power, depicted in terms of their involvement in the decision-making structure, only came into play when manager’s interpersonal competence was moderate. When the manager had strong interpersonal skills, involvement in the decision-making network did not affect his or her attractiveness as a mentor. Contrary to the researchers’ prediction, age did not affect the attractiveness of potential mentors. These preliminary findings suggest further research on the characteristics of potential mentors – particularly their skills and job membership characteristics – may be useful in developing a better understanding of the development of informal mentoring relationships. Characteristics associated with motivation to mentor. In studying motivation to mentor, researchers have examined personality, demographics, and job/career history variables. Initial data suggest motivation to mentor is positively related to some personality traits, including positive affectivity, (internal) locus of control, upward striving, and altruism (Allen, Poteet, Russell & Dobbins, 1997; Aryee, Chay & Chew, 1996). On the other hand, one study found organizational self-esteem was not associated with the perceived appeal of providing mentoring functions (Aryee, Chay & Chew, 1996). With regard to demographic variables, supervisors with more education appear to perceive fewer barriers to mentoring and have greater intentions to serve as a mentor than those with less education (Allen et al., 2000; Allen, Poteet, Russell & Dobbins, 1997). Although age is not related to perceived barriers to mentoring, older employees may be slightly less likely to intend to serve as mentors (Allen et al., 2000; Allen, Poteet, Russell & Dobbins, 1997; Ragins & Cotton, 1993). A similar pattern of relationships has been observed for organizational tenure (Allen, Poteet, Russell & Dobbins, 1997; Ragins & Cotton, 1993). In contrast, Ragins and Cotton (1993) found a very weak, but significant, trend for employees at higher levels in the organizational hierarchy to have greater intentions to mentor others. Higher ranking employees also reported slightly fewer drawbacks in being a mentor than those at lower levels. The size of this effect was small. Taken together, these results suggest motivation to mentor is affected by a variety of individual difference variables. One career history variable that has been prominent in the handful of studies that have attempted to predict experienced managers’ motivation to mentor is prior experience in mentoring relationships. Although there are exceptions (Olian et al., 1993), most studies suggest experience in mentoring relationships fosters more favorable views of being a mentor. Experience as a prot´ g´ , as a mentor, or e e both has been associated with lower perceived costs of being a mentor and greater

74

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

intentions to mentor (Allen, Poteet, Russell & Dobbins, 1997; Olian et al., 1993; Ragins & Cotton, 1993; Ragins & Scandura, 1999). The relationship between the perceived costs of mentoring and intentions to mentor also appear to be moderated by experience in mentoring relationships. Among those without experience in mentoring relationships, intentions to mentor may not be related to perceived costs. In contrast, among those with experience in mentoring relationships, executives who perceived mentoring as having high costs had much lower intentions to mentor than those who perceived mentoring as having low costs (Ragins & Scandura, 1999). In addition to underscoring the importance of prior experience in mentoring relationships, this research highlights the complexity of the construct of motivation to mentor. Implicit in researchers’ use of willingness or intentions to mentor as dependent variables is the idea that these variables are important determinants of mentoring. However, the one study that has tested this assumption found no relationship between superiors’ willingness to mentor and subordinates’ reports of mentoring functions received (Tepper, Brown & Hunt, 1993). Additional research should be conducted to evaluate the generalizability of this finding and assess whether motivation to mentor, in fact, mediates the relationship between the individual differences of experienced employees and providing mentoring. Mentor characteristics and mentoring provided. A handful of studies have investigated relationships between the characteristics of experienced employees and the mentoring they provide. In these investigations, the criteria used have included serving as a mentor (vs. not), mentoring functions, and the frequency of mentorprot´ g´ communication. The predictors studied parallel those examined in research e e on motivation to mentor. They are competence, personality, demographics, and job/career history variables, including experience in mentoring relationships. Despite descriptive research suggesting open-mindedness, patience, and honesty are important traits for mentors to possess (Allen & Poteet, 1999; Allen, Poteet & Burroughs, 1997; Gaskill, 1991), few studies have examined the relationships among mentors’ stable individual differences and the mentoring they provide. In the single study considering mentors’ cognitive ability, meaningful relationships with mentoring functions were not observed (Godshalk & Sosik, in press). With regard to personality, results have varied depending upon the traits assessed. For example, Rotundo and Perrew´ (2000) examined the relationship e between employees’ negative affectivity and their self-reported attempts to mentor younger employees. Employees higher on negative affectivity were less likely to provide mentoring, but these results were significant only for employees who felt they were experiencing a career plateau (i.e. were unlikely to be promoted). Mullen and Noe (1999) found neither mentors’ nor prot´ g´ s’ ratings of mentors’ e e

Mentoring Research

75

self-monitoring were related to career or psychosocial functions. In contrast, Fagenson (1992) observed that mentors rated by their prot´ g´ s as being more helpe e ful communicated more frequently with their prot´ g´ s and were seen as providing e e more career, psychosocial, and role modeling functions (Fagenson, 1992). In related research, a handful of studies have explored the premise that individuals faced with problems in their own careers are unlikely to have the psychological resources to serve as mentors. The results of this research have been divided, varying according to how “psychological resources” has been conceptualized and measured. Two studies suggest employees with higher work-related self-esteem are more likely to have prot´ g´ s and serve both career and psychosocial mentoring e e functions (Horgan & Simeon, 1990a; Mullen, 1998). On the other hand, studies investigating career experiences suggest career satisfaction and perceptions of experiencing a career plateau are unrelated to acting as a mentor. Career success is related to serving as a role model, but not to career and psychosocial functions (Burke, 1984; Campion & Goldfinch, 1983; Rotundo & Perrew´ , 2000). Additional e research in this area would be helpful in clarifying how career challenges affect mentoring. From a practical viewpoint, it would be worthwhile to know whether managers must have positive career experiences in order to be effective mentors. Preliminary research suggests mentors’ learning goal orientation may play a role in the mentoring process. In a recent study, mentors’ learning goal orientation was positively associated with psychosocial, career, and role-modeling functions (Godshalk & Sosik, in press). Further research on this individual difference construct should prove interesting. Mentors’ demographic characteristics – specifically education and age – have been more extensively studied than other mentor attributes. Four studies of the relationship between mentor’s education level and mentoring provided were located. The two studies examining the relationship of mentors’ education with the frequency of mentor-prot´ g´ communication yielded diverging results (Fagensone e Eland, Marks & Amendola, 1997; Mullen, 1998). There is little evidence that education is related to mentoring functions served (Burke, McKeen & McKenna, 1993; Fagenson-Eland et al., 1997; Mullen, 1998; Sosik & Godshalk, 2000b). Mentor’s age has been scrutinized in a handful of studies, yielding a pattern of results that is difficult to interpret. A study of water-management plant employees indicated that older employees are more likely to report that they have tried to serve as a mentor (Rotundo & Perrew´ , 2000), but two studies have observed no e significant differences between the ages of those who act as mentors and those who do not (Broadbridge, 1999; Campion & Goldfinch, 1983). Age appears to be unrelated to how much mentors interact with their prot´ g´ s (Fagenson-Eland et al., e e 1997; Mullen, 1998). Some, but not all, studies have found older mentors provide fewer psychosocial functions (Burke, 1984; Burke et al., 1993; Fagenson-Eland

76

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

et al., 1997; Godshalk & Sosik, 2000; Sosik & Godshalk, 2000b, in press). Most studies have found no relationship between mentors’ age and other mentoring functions (Burke, 1984; Burke et al., 1993; Fagenson-Eland et al., 1997; Godshalk & Sosik, 2000; Sosik & Godshalk, in press), although there are exceptions to this trend (Sosik & Godshalk, 2000b). One study suggests mentors’ age is not related to prot´ g´ s’ perceptions of the relationship’s effectiveness (Godshalk & e e Sosik, 2000). Restriction of range should be evaluated as a contributor to the lack of observed relationships between mentors’ demographic characteristics and mentoring provided. A m´ lange of variables that are a function of senior manager’s career and job e history have been related to the amount of mentoring they provide. The small number of studies investigating mentors’ tenure suggest job tenure is not associated with efforts to provide mentoring (Rotundo & Perrew´ , 2000), career tenure e is unrelated to serving as a mentor (Campion & Goldfinch, 1983), organizational tenure is not associated with mentor-prot´ g´ contact (Fagenson-Eland et al., e e 1997), and job and organizational tenure are not substantially related to mentoring functions (Burke et al., 1993; Fagenson-Eland et al., 1997). In contrast, one study found organizational tenure is positively related to employees’ efforts to provide mentoring (Rotundo & Perrew´ , 2000). From the handful of studies investigating e mentors’ organizational rank, it is unclear whether individuals at higher organizational levels are more likely to serve as mentors (Broadbridge, 1999; Steinberg & Foley, 1999). Results generally suggest rank is unrelated to mentoring functions (Burke et al., 1993; Fagenson-Eland et al., 1997; Sosik & Godshalk, 2000b), although there are exceptions to this trend (Scandura & Viator, 1994; Struthers, 1995). One study was located on a related variable: mentors’ power (Fagenson, 1992). This study suggests prot´ g´ s’ perceptions of their mentors’ influence are e e negatively related to psychosocial functions, but unrelated to career functions, role-modeling and communication frequency. Two studies considering mentors’ span of control yielded diverging results on how number of subordinates relates to mentoring functions (Burke et al., 1993; Tepper et al., 1993). Only a few studies have examined the relationship between experiences in mentoring relationships and providing mentoring. Compared to employees without mentors, those who have been prot´ g´ s appear to be more likely to mentor e e others (Campion & Goldfinch, 1983; Fagan & Fagan, 1983; Fagan & Walter, 1982). The results of one study suggest that this effect is ongoing. Managers who had received mentoring were more likely than those who did not to mentor more than one person (Broadbridge, 1999). There is also some evidence that individuals who have participated in more mentoring relationships provide more career functions to their current prot´ g´ s. However, in the same study, experience e e in mentoring relationships was not significantly associated with psychosocial or

Mentoring Research

77

role modeling functions (Fagenson-Eland et al., 1997). Although tentative, these results suggest being a prot´ g´ may help prepare individuals to serve as mentors. e e In conclusion, initial data suggest mentoring provided is positively associated with mentors’ helpfulness, learning goal orientation, and past experience in mentoring relationships. In contrast, research to date suggests general cognitive ability, self-monitoring, education, and tenure may not account for meaningful variance in mentoring functions. Research Priorities: Other Correlates of Mentoring The influence of individual characteristics other than race and gender in mentoring relationships has received a modest amount of attention within the literature on mentoring. More research has been dedicated to understanding the influence of prot´ g´ characteristics than mentor characteristics. As a caveat, this research e e has relied on cross-sectional, correlational methods. To the extent that prot´ g´ e e and mentor characteristics are malleable, observed relationships may indicate mentoring is a cause, rather than a consequence of these variables. Longitudinal research in this area would be highly useful. A few prot´ g´ characteristics stand out as meriting additional study. First, as e e mentioned earlier, relatively little research has been devoted to understanding how employees’ abilities and skills affect their experiences as prot´ g´ s, particularly e e the nature of mentoring functions they receive. Second, research on prot´ g´ s’ e e personality traits has uncovered some thought-provoking relationships. As will be discussed in more detail later, research examining whether proactive individuals receive more mentoring will be informative. In addition, it is striking research has generally not taken advantage of current models of personality being utilized in research in work settings – specifically, the five factor model of personality and related frameworks (Hough & Ones, 2001; Hough & Schneider, 1996; Schneider & Hough, 1995). Studies examining individual traits from the five factor model have yielded promising results. There are also gaps in the mentor characteristics that have been examined. First, empirical research on the cognitive attributes of mentors is lacking. Only one study was located that attempted to assess mentors’ abilities (Godshalk & Sosik, in press). Although substantial relationships between mentors’ self-reported scores on a general test of cognitive ability and prot´ g´ reports of mentoring functions were not e e observed, it seems likely that more specific mentor knowledge, skills, and abilities affect their attractiveness and effectiveness as mentors. Second, it may be valuable to investigate more systematically the effect of less stable mentor characteristics on mentoring. Implicit in research on mentor’s self-esteem, career situation, job stress, and number of subordinates is the idea that individuals whose resources, whether psychological or temporal, are being stretched will be less willing and less

78

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

able to mentor others. To date, research on this topic has yielded mixed results. In future research, it would be informative to clarify what hinders mentoring. Third, research on some mentor demographics (e.g. marital status) and some predictorcriterion combinations (e.g. the relationship between age and serving as a mentor) is sparse. In the future, it would be informative to address these gaps, where there is a theoretical reason to expect meaningful relationships. Finally, quantitative research has not been directed towards understanding how interests and values affect the decision to be a mentor and the mentoring functions served. It has been argued that from the mentor’s perspective, the benefits of being a mentor include helping others succeed and leaving a legacy. It seems likely that individuals whose values or interests are not aligned with the benefits will not devote time and energy to mentoring. While most research has examined prot´ g´ or mentor characteristics as e e predictors of the mentoring received in the relationship, an increased amount of research with data collected from matched mentor/prot´ g´ pairs would also be e e helpful. With the exception of research on diversity, most research has collected data from only one member of the mentor/prot´ g´ dyad (for examples of other e e exceptions, see Ensher & Murphy, 1997; Godshalk & Sosik, in press; Scandura & Schriesheim, 1994; Wayne et al., 1999). There is a wealth of information that could be gained from studying both members of the mentoring relationship, and we urge research in this area to collect data from mentor/prot´ g´ pairs more e e frequently. Analysis of dyads would allow us to delve into a deeper understanding of mentoring relationships (Kashy & Levesque, 2000) and would help to decrease mono-method bias concerns. Based on prior theory and research, four categories of dyad characteristics appear to be particularly promising foci for future research. First, the concept of similarity underpins many of the arguments advanced for why mentoring relationships composed of mentors and prot´ g´ s of different genders or races e e will be more challenging and less productive than those involving mentors and prot´ g´ s of the same race or gender (e.g. Dreher & Cox, 1996; Noe, 1988b). e e Differences are thought to hamper mentoring relationships; similarity to facilitate them. Prot´ g´ perceptions of similarity are positively associated with mentoring e e functions (Burke et al., 1993; Ensher & Murphy, 1997; Nielson et al., 2001). One of these studies also reported the correlation between mentor perceptions of similarity to the prot´ g´ and prot´ g´ perceptions of similarity to the mentor was e e e e only 0.22 (Ensher & Murphy, 1997). Clarification of what dimensions or attributes affect prot´ g´ s’ and mentors’ perceptions of similarity would be helpful. e e A second important dyad characteristic to investigate is the pattern of prot´ g´ e e and mentor personality traits. Research on other kinds of relationships suggests that mentoring relationships may be affected by the combination of traits

Mentoring Research

79

demonstrated by mentors and prot´ g´ s. For example, in a study of roommates, e e mean levels of certain traits (extraversion and conscientiousness) were related to conflicts about the relationship. Task-related conflicts, as well as the frequency of conflicts, were associated with differences in extraversion (Bono, Boles, Judge & Lauver, 2002). It seems likely that the interaction of prot´ g´ and mentor e e characteristics will affect an array of relationships characteristics, such as those specified in our conceptual model. A third critical dyad characteristic may be the congruence between mentors’ and prot´ g´ s’ approach to handling potentially sensitive issues. In a study of e e cross-race work relationships, Thomas (1993) identified two strategies that relationship participants used to handle the issue of race: denial/suppression and direct engagement. When both parties in the relationship used the same strategies, both career and psychosocial functions were present. In contrast, only career functions were provided when the preferred style of the parties diverged. Additional research on how styles or approaches to handling conflict affect relationship characteristics and mentoring received would be valuable. A final dyad characteristic that may be worthwhile to study further is whether or not both members of the relationship work within the same organization. Mentors who are external to the prot´ g´ s’ organization may not be able to provide the same e e kinds of support as those internal to the organization (Ostroff & Kozlowski, 1993; Ragins, 1997, 1999). Based on this reasoning, prot´ g´ s with “external” mentors e e have occasionally been excluded from research studies (e.g. Ostroff & Kozlowski, 1993). Preliminary data supports this conclusion. For example, one longitudinal study found new insurance agents with “internal” mentors had a higher first year “survival” rate than those with “external” mentors. Of the survivors, those with “internal” mentors also sold more policies (Silverhart, 1994). There is also some evidence that internal and external mentors may differ in terms of the mentoring functions they serve (Godshalk & Sosik, in press). More research is needed to clarify how the employment of mentors and prot´ g´ s with the same vs. different e e organizations shapes relationship characteristics, mentoring functions, and – ultimately – outcomes experienced.

Dynamics of Mentoring Relationships A small body of research has been directed towards understanding the dynamics of mentoring relationships. Two broad questions have been investigated. Researchers taking a more “macro” view have asked: How do mentoring relationships evolve over time? Those taking a more “micro” perspective have inquired: How do mentors and prot´ g´ s interact with and influence each other? e e

80

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Phases of Mentoring Relationships Researchers addressing the first question have attempted to identify the phases through which mentoring relationships progress. Based on interviews with mentors and/or prot´ g´ s (Kram, 1985b; Missirian, 1982; Phillips, 1978, cited in Carden, e e 1990; Pollock, 1995), several models describing the course of informal mentoring relationships have been developed (Pollock, 1995). Although these models diverge in some ways, all recognize a preliminary stage in which the mentor and prot´ g´ e e meet and begin evaluating the possibility of having a relationship, an “active” stage in which the mentor assists the prot´ g´ , and a termination phase in which the e e mentor and prot´ g´ either transform the nature of their relationship or end it (Kram, e e 1985b; Missirian, 1982; Phillips, 1977, cited in Carden, 1990; Pollock, 1995). Two studies have directly tested models of the phases of mentoring relationships (Chao, 1997; Pollock, 1995). Both examined how career and psychosocial functions varied across the phases of mentoring relationships. Each provided some support for the idea that mentoring relationships have a preliminary initiation phase before entering a more active phase. Career and psychosocial functions increased as mentoring relationships progressed from the early to middle phase. Contrary to the models of mentoring phases, neither study found evidence that career and psychosocial functions decrease as mentoring relationships move from the middle to late stages. However, both studies had methodological limitations that made it difficult to observe changes in the later stages of mentoring relationships. Chao (1997) pointed out the past-tense wording of the items in her study may have led prot´ g´ s in the later phases of mentoring relationships to respond e e based on the mentoring functions they had experienced during the “height” of their mentoring relationships. Pollock’s (1995) use of cut-scores to differentiate individuals with and without mentors may have resulted in prot´ g´ s in the late e e stages of mentoring relationships being categorized as not having mentors. Overall, there is relatively little empirical evidence with which to evaluate models describing the phases of mentoring relationships. Mentor-Prot´ g´ Interactions e e Despite the fact that definitions emphasize mentoring as involving intense, interpersonal relationships, research on how mentors and prot´ g´ s interact is e e limited. A handful of studies have illustrated the complexity of mentor-prot´ g´ e e interactions, highlighting a number of ways in which prot´ g´ s and mentors e e influence each others’ perceptions, affect, and behaviors. The work of several researchers indicates that the amount and kind of ingratiation and influence tactics used by employees affect the mentoring they receive. Judge and Bretz (1994) reported a small positive correlation between supervisor-focused influence tactics and having access to a mentor. Supervisorfocused influence tactics, such as praising one’s supervisor and agreeing with his

Mentoring Research

81

or her opinions, are directed towards improving the supervisor’s affect toward the subordinate. Although significant, the relationship between job-focused influence tactics and access to a mentor was weak. Job-focused influence tactics involve self-promotional efforts aimed at making one appear more competent, such as making others aware of job-related accomplishments. Consistent with these results, employees in Hong Kong who used more ingratiation tactics reported receiving more career mentoring (Aryee, Wyatt & Stone, 1996). Ingratiation included four kinds of assertive behaviors used to gain the approval of those who control significant rewards, including doing favors, conforming to the target’s opinions, making positive evaluative comments about the target, and using self-presentation to convince the target of the speaker’s competence. Note that these behaviors reflect elements of both the supervisor-focused and job-focused tactics studied by Judge and Bretz (1994). One study suggests that gender may moderate the effectiveness of prot´ g´ s’ influence tactics. According to the gender e e congruency theory of power, women are perceived favorably when – congruent with the gender stereotype of women – they use power indirectly (Tepper et al., 1993). Consistent with this theory, Tepper and her colleagues observed women who used weaker upward influence tactics received slightly more psychosocial mentoring from their supervisors, while men who used stronger influence tactics received more career mentoring. Taken together, these three studies indicate that the receipt of mentoring is actively shaped by prot´ g´ s’ use of influence e e and ingratiation tactics. One innovative line of inquiry has explored how prot´ g´ s and mentors influence e e one specific kind of behavior: information seeking. Mullen (1994) proposed that mentors and prot´ g´ s mutually seek a variety of information from each other, e e including technical, referent, and normative information, along with performance and social feedback. Mentoring functions, relationship closeness, method of relationship initiation (mentor-, prot´ g´ -, or organization-initiated), contextual e e variables, mentor characteristics, and prot´ g´ characteristics influence the extent e e of information seeking. In a study of 161 mentors and 140 prot´ g´ s from 17 e e organizations in the U.S., Mullen and Noe (1999) tested hypotheses derived from this framework. Partial support was obtained. Based on information provided by mentors, the results of a series of hierarchical regression analyses suggested mentors’ information seeking was related to the extent to which the mentor was influenced by his or her prot´ g´ , mentors’ perceptions of whether the prot´ g´ e e e e thought information seeking was appropriate, perceived prot´ g´ competence, and e e by career mentoring. On the other hand, based on similar analyses using data from prot´ g´ s, only perceptions of the appropriateness of information seeking were sige e nificantly associated with information seeking. This thought-provoking research tentatively suggests how mentors interact with their prot´ g´ s may be shaped by e e a variety of factors.

82

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Other researchers have investigated the relationship between mentoring functions and relationship characteristics, such as trust, reciprocity, and mentorprot´ g´ interdependence. In a study of 108 faculty mentors and 215 prot´ g´ s who e e e e were currently doctoral students or assistant professors in management, Young and Perrew´ (2000) found met expectations partially mediated the relationship e between mentoring functions and perceptions of relationship effectiveness and trust. An interesting finding was that the mentoring function associated with met expectations differed for mentors and prot´ g´ s. Psychosocial support was e e associated with prot´ g´ s’ met expectations, while mentors’ met expectations e e appeared to be driven by perceptions of the prot´ g´ s responding appropriately e e to career functions. In related research, positive associations between mentoring functions and satisfaction with mentoring relationships have been reported in other studies (Ensher & Murphy, 1997; Ragins & Cotton, 1999). Other researchers have observed mentoring functions have moderate, positive relationships with prot´ g´ s’ perceptions of mentor-prot´ g´ reciprocity (Ensher, Thomas & Murphy, e e e e 2001) and with mentors’ reports of the extent to which their prot´ g´ s influence e e them – a measure of relationship closeness (Mullen, 1998; Mullen & Noe, 1999). Allen, Day and Lentz (2002) found prot´ g´ s’ interpersonal comfort with their e e mentors was strongly and positively related to the mentoring functions they reported receiving and their perceptions of the quality of their relationships. On the other hand, Burke and McKeen (1997) reported emotional distance had a small association with psychosocial functions and, at most, a weak association with four career mentoring functions. Further research is needed to clarify how mentoring functions and other relationship characteristics influence one another. Research Priorities: Dynamics of Mentoring Relationships Only a small amount of research has been directed towards understanding the dynamics of mentoring relationships. Both “macro” and “micro” approaches have been used, providing a preliminary understanding of how mentoring relationships evolve and how mentors and prot´ g´ relate to one another. The “macro” research e e suggests mentoring functions may not be constant over time, increasing from early to middle phases of the relationship. As noted in our discussion of the role of gender in mentoring relationships, how mentoring functions change over the duration of the relationship may depend upon the characteristics of mentors and prot´ g´ s e e (Turban et al., 2002). Additional research on factors influencing the evolution of mentoring relationships would be helpful. A particularly interesting question is whether formal and informal mentoring relationships have similar phases. The small number of “micro” studies have linked mentoring functions to relationship characteristics and prot´ g´ influence tactics. An important theme of e e this research is that the prot´ g´ is an active participant in mentoring relationships, e e

Mentoring Research

83

rather than a passive beneficiary of the mentor’s benevolent attention. Preliminary evidence suggests mentors’ behaviors and views of their mentoring relationships may be shaped by a variety of factors, including mentoring functions, prot´ g´ e e behaviors, and prot´ g´ characteristics. Social psychological literature on relatione e ships can be used to contribute guiding frameworks and hypotheses to this line of research. We provide examples of the application of the literature on relationships to mentoring in our conceptual model introduced later in this review. In studying mentoring dynamics from either a “macro” or “micro” perspective, research with multiple time waves is critical.

Mentoring in the Context of Formal Mentoring Programs Due to differences between informal and formal mentoring, and because formal mentoring has gained the attention of organizations as a potential management developmental tool, we discuss research on formal mentoring separately in this section. We begin with a general overview, responding to two questions: (1) how are formal and informal mentoring different?; and (2) how are organizations using formal mentoring programs? We then provide a summary of available research on formal mentoring programs. Overview According to Ragins and Cotton (1999), formal and informal mentoring relationships differ according to three key dimensions: the initiation of the relationship, the structure surrounding the relationship, and other aspects of the relationship. While informal mentoring relationships develop naturally over time, formal mentoring relationships are typically initiated through some type of organizational matching process. While there are no structural rules governing informal relationships, formal mentoring programs may include several elements of structure, such as suggested guidelines about how often dyads should meet, suggestions about possible topics to discuss, a goal setting process for the prot´ g´ , training sessions to prepare e e both mentors and prot´ g´ s for the experience, and a specified duration for the e e relationship. Although some formal relationships may last longer, the minimum duration as specified by the program is typically between 6 and 12 months. Other aspects of the relationship may differ as well (Ragins & Cotton, 1999). For example, because the relationship did not develop naturally, some formal mentors may be less motivated to be in the relationship than informal mentors. Matches of formal mentors with prot´ g´ s may result in dyads from different functional e e units, possibly impeding the ability of the mentor to fully provide assistance to the prot´ g´ . Also, it is possible that some of the mentors identified do not have e e

84

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

appropriate coaching or communication skills. While a prot´ g´ would likely not e e be naturally attracted to a mentor without good coaching or communication skills, it is plausible that a mentor without these skills could be assigned in a formal mentoring situation. Finally, because formal programs increase the visibility of a mentoring relationship, mentors in a formal program may be less willing than informal mentors to engage in mentoring that might be perceived by others as favoritism. While not all organizational mentoring programs have specified goals, a common goal of formal mentoring programs is to promote the careers, development, and performance of prot´ g´ s at a managerial level. While organizations turn to e e an array of techniques to develop their leaders, including 360-degree feedback, assessment centers, executive coaching, action learning, classroom training, and e-learning (see, for example, McCauley & Hezlett, 2001), mentoring uniquely involves the sharing of experience and information between current leaders and future leaders. An accompanying goal for some of the formal programs focused on careers, development, and/or performance is a desire to increase diversity at higher levels of the organization by placing a program focus on women and minorities (Douglas & McCauley, 1999). Organizations also frequently use formal mentoring programs with new hires (usually managers or college graduates) to give individuals a head start in acquiring an understanding of the organization and how to best be effective (Douglas & McCauley, 1999; Ragins et al., 2000). According to Chao, O’LearyKelly, Wold, Klein and Gardner (1994), organizational newcomers encounter a barrage of issues to learn about and do, including the need to: (1) learn job tasks and responsibilities and become proficient on the job; (2) learn about organizational politics, organizational goals and values, and the organization’s history; (3) become familiar with the organization’s technical language, jargon, traditions, customs, and stories; and (4) establish relationships with others within the organization. While other means of socialization are available (e.g. formal orientation training programs), mentoring is viewed by some organizations as an effective means of assisting and developing new employees. Formal mentoring programs vary in their quality, or the extent to which they are planful efforts versus haphazard pretenses of a formal mentoring program (Single & Muller, 2001). At the haphazard extreme, an organization might assign mentors to prot´ g´ s, have a social for the pairs to meet one another, and then passively e e hope that the mentors and prot´ g´ s will meet again to discuss something of an e e unspecified nature. A more planful approach involves making careful matches of mentors and prot´ g´ s, making suggestions about how often mentors and prot´ g´ s e e e e should meet, conducting training to offer suggestions about what mentors and prot´ g´ s might discuss and how mentoring relationships work, conducting e e

Mentoring Research

85

follow-up checks to see how the relationship is working, and having mentors and prot´ g´ s develop objectives for the relationship (Gaskill, 1993; Newby & Heide, e e 1992; Tyler, 1998; Zey, 1985). Kram and Bragar (1992) suggest six components that are essential to a high quality formal mentoring program: (1) specific objectives and an identified target population; (2) a process to select and match prot´ g´ s with mentors; (3) an orientation that involves suggestions on maintaining e e the relationship as well as expectation setting; (4) communication with involved parties about the intent of the program; (5) a monitoring and evaluation process; and (6) a coordinator to provide support to participants. The authors further note that formal programs will be most effective when they are clearly linked to business goals, when they are consistent with other human resources practices and policies, and when they are implemented with “effectiveness rather than expediency in mind” (p. 228). Despite salient differences between informal and formal mentoring (Ragins & Cotton, 1999) and the widespread use of formal mentoring on the part of organizations (Barbian, 2002), two issues were apparent in our review of the literature. First, a large proportion of researchers have not asked the prot´ g´ s in their studies whether their mentors were acquired naturally or instead e e through a formal mentoring program. As mentioned earlier in this review, this failure to assess the origin of mentoring relationships acts as a significant barrier to the clear understanding of the effectiveness of formal and informal relationships. Second, there is a striking dearth of research on formal mentoring. Our review identified less than twenty-five empirical studies focused on the outcomes associated with formal mentoring programs in organizations. Of these, almost half were focused on peer mentoring or alternatively did not provide sufficient methodological rigor, sample size, and/or measurement information to summarize. The remaining thirteen studies provided information regarding three important questions about formal mentoring programs in organizations. First, how do outcomes for formal program participants compare to outcomes for individuals without mentors? Second, how do outcomes for formal program participants compare to outcomes for individuals with informal mentors? Finally, what does empirical research suggest about the association between formal mentoring program characteristics and program outcomes? Formal Mentoring Versus No Mentoring Four studies have compared the outcomes reported by individuals with formal mentors to those of individuals reporting no mentor. The goal of these studies has been to evaluate what formal mentoring programs can achieve in comparison to a person not having had a mentor. For example, Chao, Walz and Gardner (1992) compared individuals with formal mentors to those without mentors on the outcomes of organizational socialization, job satisfaction, and salary. Ragins and

86

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Cotton (1999) compared individuals with formal mentors to those without mentors on levels of compensation and promotion rates. In the Chao et al. (1992) study, formal prot´ g´ s reported more favorable e e outcomes than employees without mentors on three dimensions of socialization (understanding the goals/values of the organization, politics, and people). Formal prot´ g´ s did not differ from those without mentors on three other dimensions of e e socialization (language, history, and performance proficiency) or on levels of job satisfaction or salary. Ragins and Cotton (1999) found individuals with formal mentors did not have higher levels of compensation or promotion rates than individuals without mentors. It is critical, when evaluating and interpreting these first two comparisons between formal mentoring and no mentoring, to recognize that both studies included individuals across a variety of organizations and that there was no control for the quality or purpose of the formal mentoring programs that were evaluated. Some participants were undoubtedly part of very high quality mentoring programs, while others were likely involved in programs that did little more that match individuals together and send them on their way with little or no guidance. If there were more participants from haphazardly developed formal mentoring programs than from carefully implemented programs, the results of these studies may be biased toward portraying formal mentoring as less effective. It is important, then, to interpret the Chao et al. (1992) and Ragins and Cotton (1999) studies as comparisons of individuals in formal mentoring programs in general (regardless of the quality or purpose of the program) to individuals without mentors. We cannot generalize the results of these studies to a carefully planned formal mentoring program without additional data. Ragins et al. (2000) recognized the importance of taking into account the quality and purpose of formal mentoring programs. Using the same sample as Ragins and Cotton (1999), these authors found that individuals with satisfying formal mentoring relationships (as opposed to unsatisfying or marginal) had higher levels of career commitment, organizational commitment, and organizational self-esteem than individuals without mentors. In the only other study to date that has compared outcomes of formally mentored individuals to those without mentors, Seibert (1999) noted that certain types of people may participate (or be asked to participate) in formal mentoring programs, making it difficult to compare participants to non-participants with a non-experimental and/or cross-sectional examination. Using a quasi-experimental design, Seibert (1999) examined the effectiveness of a formal mentoring program for new employees at a Fortune 100 company by comparing 43 participants with a control group of 18 individuals that had reported never having had a mentor. Participants and control group members were assigned to their conditions by graduation

Mentoring Research

87

major and date; analyses showed no differences between the two groups at Time 1 of the study. One year later, the employees participating in the mentoring program reported higher levels of job satisfaction than the non-participants in the control group. There were no differences between the participants and non-participants on organizational commitment, work-role stress, or self-esteem at work. In conclusion, although few in number, studies comparing formal mentoring to no mentoring have generally portrayed positive outcomes associated with formal mentoring, including higher levels of socialization, career commitment, organizational commitment, and organizational self-esteem. While one study (Ragins & Cotton, 1999) did not portray positive outcomes, the outcomes examined were limited and the quality of the participant’s programs was not controlled. More controlled research involving random assignment of participants to formal mentoring versus no mentoring would be useful in further comparing the benefits associated with formal mentoring. Formal Mentoring Versus Informal Mentoring Three of the studies noted in the previous section (Chao et al., 1992; Ragins & Cotton, 1999; Ragins et al., 2000) along with three additional studies (Allen, Day & Lentz, 2002; Fagenson-Eland et al., 1997; Viator, 2001a) have compared the outcomes received by prot´ g´ s with informal mentors and those received by e e prot´ g´ s with formal mentors. The goal of these comparisons has been to evaluate e e to what extent formal mentoring relationships can mimic the outcomes of informal mentoring relationships. Allen, Day and Lentz’s (2002) study included 53 formal prot´ g´ s and 65 infore e mal prot´ g´ s from two organizations. Although results showed that individuals e e in formal and informal relationships had similar levels of interpersonal comfort with their mentors, individuals in informal relationships reported higher levels of career mentoring and higher quality mentoring relationships than individuals in formal relationships. Controlling for length of mentorship, Chao et al. (1992) also found that individuals in formal mentoring relationships reported receiving lower levels of career-related mentoring functions than individuals in informal mentoring relationships. While results from Allen, Day and Lentz (2002) and Chao et al. (1992) suggest individuals in formal relationships receive lower levels of career functions from their mentors, Fagenson-Eland et al. (1997) found whether the relationship was informal or formal was not associated with the level of career functions reported by prot´ g´ s. Instead, formal prot´ g´ s reported lower levels of psychosocial funce e e e tions than informal prot´ g´ s in this study. Viator’s (2001a) analysis of multiple e e subsamples showed that, in comparison to informal mentoring, formal mentoring was less frequently associated with lower levels of role ambiguity, role conflict,

88

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

perceived environmental uncertainty, and turnover intentions. His conclusions, however, should be accepted with caution. Specifically, the number of individuals with formal mentors in some of his groups was extremely small (e.g. 14 senior managers reported having formal mentors, and Viator analyzes female and male senior managers separately, further reducing the sample size). Furthermore, his results are difficult to interpret as several participants had both formal and informal mentors. Ragins and Cotton (1999) is the only study to date that has compared informal and formal prot´ g´ s on mentoring received at the specific function level, rather e e than aggregating function subscales into the total of career-related functions or the total of psychosocial-related functions. This is an important step toward understanding the nature of the differences in mentoring between informal and formal mentoring relationships. Formal prot´ g´ s in this study reported e e lower levels of mentoring in comparison to informal prot´ g´ s on almost every e e mentoring function (e.g. on the dimensions of sponsoring, coaching, protection, challenging assignments, exposure, friendship, social support, role modeling, and acceptance). No differences were found between formal and informal prot´ g´ s on e e the dimensions of parenting and counseling. Formal prot´ g´ s also reported lower e e compensation than individuals with informal mentors. Again, as was the case with formal/no mentoring comparisons, these studies have not controlled for the characteristics of the formal mentoring programs or have not specified the characteristics of the formal mentoring programs that were used in the comparisons. Although Ragins et al. (2000) did not compare the functions received by individuals with informal mentors to those with formal mentors, this study found that individuals with high levels of satisfaction with their formal mentors did not differ from individuals with high satisfaction with their informal mentors on all six outcomes in their study, including career commitment, job satisfaction, organizational commitment, organizational self-esteem, promotion satisfaction, intentions to quit, and procedural justice. These authors concluded “. . . the view that informal mentoring relationships will automatically be more beneficial than formal mentoring relationships is apparently too simplistic; the level of satisfaction in a relationship appears to be the key variable” (p. 1187). In conclusion, studies comparing formal mentoring to informal mentoring generally portray informal mentoring as more effective, although recent research by Ragins et al. (2000) suggests that individuals with high levels of satisfaction with their formal mentors did not differ in outcomes observed from individuals with high satisfaction with their informal mentors. This result suggests that formal mentoring relationships have the potential to reap the same benefits as informal mentoring relationships.

Mentoring Research

89

Program Characteristics Three studies to date have directly addressed the relationship between formal mentoring program characteristics and program outcomes. Ragins et al. (2000) examined the hypothesis that formal programs that are designed to be most similar to informal relationships would be the ones perceived as most effective. Specifically, their hypothesis was that programs would be viewed as most effective and would result in more positive career and job attitudes when: (1) participation was voluntary; (2) mentors and prot´ g´ s were allowed to participate in the e e matching process; (3) the program was aimed at career development rather than general job orientation; (4) guidelines for frequency of meetings were offered; and (5) recognition for mentors was offered. Results showed that while programs were perceived as marginally more effective if the mentor entered the program voluntarily, prot´ g´ choice was not significantly related to perceived effectiveness. e e The method of matching mentors and prot´ g´ s was not predictive of perceived e e effectiveness. Programs aimed at career development were associated with higher levels of prot´ g´ satisfaction with opportunities for promotion, but the objectives e e of the program had no relationship to the overall perceived effectiveness of the program. Formal programs that offered guidelines for meeting frequency were perceived as more effective than those without guidelines. Mentor recognition was not related to perceived effectiveness or career outcomes for the prot´ g´ . e e Viator (1999) examined the role of matching process, goal setting, and meeting frequency guidelines in formal mentoring programs by surveying certified public accountants working in large public accounting firms. Most prot´ g´ s indicated e e that their formal mentoring program had meeting frequency guidelines (69.2%) and required the prot´ g´ s to set goals and objectives (62.8%). Viator reported e e that meeting more frequently and setting goals was related to higher prot´ g´ e e satisfaction with the mentorship. His results showed variability in how prot´ g´ s e e were matched with their mentors, with 197 participants indicating they had input into the selection of their mentor and 113 participants indicating no input. For prot´ g´ s indicating they had input, the type of input varied. For example, some e e firms assigned the prot´ g´ s a mentor from a “top choices” list; others surveyed e e the prot´ g´ regarding needs and objectives. Prot´ g´ s that were not allowed input e e e e into the matching process were less satisfied with their mentors and were more likely to report their needs were not met in comparison with prot´ g´ s given input. e e Results from Klauss (1981), although based on a very small sample size (n = 18), similarly suggested participant input into the matching process was associated with higher perceived relationship effectiveness. Four additional studies, although not specifically focused on understanding the role of program characteristics, highlight the importance of opportunity to interact and frequency of meeting within the formal mentoring environment. In

90

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Noe’s (1988a) study of 39 prot´ g´ s and 43 mentors from a formal mentoring e e program for educators, mentors spent an average of four hours with their one to four prot´ g´ s over a period of six months. More frequent meeting was related e e to perceived quality of interaction with the mentor and higher psychosocial benefits. Three primary barriers to interaction were identified: lack of time, conflicting work schedules, and geographic separation. Studies by Heimann and Pittenger (1996) and Orpen (1997), although based on small samples, also suggest the importance of opportunity to interact and dyad relationship closeness in a formal mentoring program. Finally, an investigation of 77 prot´ g´ s starting e e new businesses found that prot´ g´ perceptions of business success were predicted e e by the frequency with which they had contact with their mentors and the level of advice on legal, technical, financial, and marketing issues provided by their mentors (Waters, McCabe, Kiellerup & Kiellerup, 2002). In conclusion, research has suggested that formal program characteristics such as frequency of meeting guidelines, specified program objectives, or participant input into the matching process may be related to the outcomes of a formal mentoring program and should be examined in future studies of formal mentoring. Specifically, it is important both to understand the role of program characteristics in outcomes achieved by formal programs as well as to control for the quality of programs in research comparing formal mentoring to informal mentoring or no mentoring. However, results on program characteristics are so preliminary, with few studies having addressed this issue, that our understanding of the role of program characteristics and other antecedents of effective mentoring outcomes is still very limited. Research Priorities: Formal Mentoring Continued research examining the benefits of formal mentoring are needed. A primary need, however, is for researchers to make attempts to differentiate between “quality” and more poorly planned mentoring programs when doing so. Otherwise, the conclusion may be that formal mentoring programs do not work, when actually it is simply that poorly planned mentoring programs do not work but more organized, planful programs do work. To conduct these examinations, we need a clearer idea of what constitutes a quality mentoring program. Furthermore, a conceptual framework to guide examinations of formal mentoring is sorely needed. In the next section, we attempt to provide such a framework to guide future research on formal mentoring.

FORMAL MENTORING: A CONCEPTUAL FRAMEWORK
Research in the area of formal mentoring programs is at an exciting stage. A small number of studies on formal mentoring have now been conducted and have

Mentoring Research

91

helped to develop a preliminary understanding of formal mentoring programs and their outcomes. To assist further understanding of and research on formal mentoring programs, we develop a conceptual model that allows us to suggest propositions relevant to four salient questions. First, what outcomes might formal mentoring programs achieve? Second, what is the nature of the mentoring that occurs in a formal mentoring relationship? Third, what are the antecedents of positive outcomes in a formal mentoring context? Finally, what is the process through which formal mentoring leads to positive outcomes? Our conceptual model was developed with reference to a diverse body of literature, including that on interpersonal relationships, career success, training and development, and informal mentoring. The result of our efforts, shown in Fig. 1, is applicable in informal, as well as formal contexts; a primary exception is the box labeled “program antecedents,” surrounded by a double-dashed box. While we hope that the model also extends research and thinking in the informal domain, due to the critical dearth of discussion and research on formal mentoring programs in stark contrast to the widespread organizational use of these programs (Barbian, 2002), we focus the following discussion of this conceptual model within the formal mentoring context.

Mentoring Outcomes We differentiate between two broad classes of formal mentoring outcomes. The first are several distal outcomes that may be attained by the prot´ g´ , mentor, and e e organization. The second are more proximal outcomes including prot´ g´ change e e and satisfaction with the assigned mentor and the mentoring program. Distal Outcomes Findings from Ragins et al. (2000) suggest that, assuming individuals are in a satisfactory relationship, formal mentoring has the potential to reap the same distal outcomes as informal mentoring. Following the example of Hunt and Michael (1983), we conceptualize these outcomes at three levels of analysis: prot´ g´ oute e comes, mentor outcomes, and organizational outcomes. At the prot´ g´ level, we conceptualize the formal mentoring process as a e e potential means for advancing an individual’s career success. The career success literature differentiates between extrinsic career success (represented by more observable career achievement indices, such as promotions or ascendancy and compensation level) and intrinsic success (represented by more subjective reports of job satisfaction, career satisfaction and commitment, and life satisfaction) (see, for example, Boudreau, Boswell & Judge, 2001; Judge, Cable, Boudreau & Bretz, 1995; Judge, Higgins, Thoresen & Barrick, 1999; Seibert, Kraimer &

92

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Fig. 1. A Conceptual Process Model of Formal Mentoring.

Mentoring Research

93

Crant, 2001). The distinction clarifies that more promotions and higher compensation are not the only route to success, and recognizes that some individuals are more concerned with “psychological success,” or whether they have lived a satisfying life (Hall & Mirvis, 1996). Our model shown in Fig. 1 also portrays outcomes at the mentor and organizational level. The processes and specific means by which mentors benefit from the formal mentoring process are not elaborated in our model; instead our focus will surround the target of the formal mentoring program, the prot´ g´ , and to a lesser e e extent the resulting outcomes for the organization. We note, however, that mentor benefits such as personal satisfaction that have been reported by mentors in the informal mentoring context are also expected to appear in the formal mentoring context. Organizational outcomes are expected to be generated from the proximal and distal outcomes of the prot´ g´ and mentor, as well as from the people relevant e e to the organization that the prot´ g´ and mentor impact and interact with (e.g. e e customers and co-workers). These organizational outcomes may be as diverse as improved organizational commitment, retention, organizational communication, managerial succession, institutional memory, productivity or job performance, and perceived justice. As a caveat, organizational level outcomes are limited not only by the effectiveness of the formal mentoring program and the relationships that are formed, but also by the number of participants in the program (Kram & Bragar, 1992). If the organization launches only ten partnerships, organizational-level outcomes will be commensurate with the small number of pairs. Furthermore, the net organizational outcome could be negative if program costs are high and active participation in the program low, if matched pairs develop only superficial relationships, or if there is perceived injustice on the part of individuals who wanted to be in the mentoring program (Kram, 1985a). This suggests assessments of the organizational benefits of mentoring programs need to be conducted at the aggregate level, taking into account all stakeholders. Proximal Prot´ g´ Outcomes e e Research to date has not carefully delineated the proximal areas of impact that mentoring – formal or informal – may have upon the prot´ g´ , although exciting e e new advances into this area have recently appeared (see, for example, Lankau & Scandura, 2002). Our model first suggests four potential areas of prot´ g´ change e e (cognitive learning, skill-based learning, affective-related learning, and social networks) that may, at least in part, drive the achievement of more distal extrinsic and intrinsic career success outcomes in a formal mentoring context. Three of the areas of prot´ g´ change (cognitive, skill-based, and affective e e learning) were derived from Kraiger, Ford and Salas’s (1993) classification of

94

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

learning outcomes that stem from training interventions. Following Kraiger et al.’s conceptualization, we use cognitive learning to represent enhancements in declarative knowledge, procedural knowledge, strategic or tacit knowledge, knowledge organization, or cognitive strategies that may occur as a result of the mentoring relationship. For example, the mentoring process may result in the prot´ g´ gaining new declarative knowledge (e.g. information about the company’s e e history and politics), developing job-related mental models (e.g. about how to be successful in his or her organization), or learning new cognitive strategies (e.g. planning, problem solving, decision making, creative thinking, self-regulation). Skill-based learning would involve the improvement of technical (e.g. report writing, running a meeting) or motor skills (e.g. jewelry repair, stitching up a head wound). Affective-based learning could be attitudinal (e.g. changes in self-awareness or values, improved tolerance for diversity, or reconciliation of work/life balance issues) or motivational (e.g. representing changes in the prot´ g´ ’s motivational disposition, self-efficacy, or goal setting). e e The last area of prot´ g´ change (social networks), rather than being an e e individual learning outcome, involves changes in relationships, specifically the prot´ g´ ’s social integration and development of social capital. This increased e e social integration and development of social capital will be expected to occur if the formal mentor helps the prot´ g´ develop his or her network, for example e e by taking the prot´ g´ along to business functions, or introducing the prot´ g´ to e e e e colleagues. These introductions may result in an expanded social network that provides the prot´ g´ with other sources of contacts, advice, social support, or e e strategic information (Podolny & Baron, 1997). The specification of these change outcomes helps to unfold the mentoring process by better describing the mechanisms through which mentoring may lead to increased extrinsic and intrinsic career success (i.e. mentoring leads to increased career success, at least in part, through prot´ g´ change). For example, e e a prot´ g´ , as a result of discussions with his or her mentor, may develop a e e strong mental model for what it takes to succeed in the organization. It is this new understanding that may be largely responsible for the prot´ g´ ’s later e e success in getting a promotion. Similarly, mentor-prot´ g´ discussions on topics e e such as improved role clarity, how best to pursue one’s career objectives, and strategies for maintaining work-life balance have the potential to increase the prot´ g´ ’s job, career, and life satisfaction. The number and placement of one’s e e social resources have also been empirically linked to career success measures including current salary, number of promotions over one’s career, and career satisfaction (see, for example, Seibert, Kraimer & Liden, 2001). Consequently, in regard to the relationship between prot´ g´ change and career success we e e suggest:

Mentoring Research

95

Proposition 1. Higher levels of prot´ g´ cognitive, skill-based, and affective e e learning and the expansion of social networks will be associated with increased probability of promotion and higher compensation as well as higher job satisfaction, career satisfaction, career commitment, and life satisfaction. Importantly, not all learning outcome areas need to be affected for distal outcomes to be influenced. For example, a prot´ g´ may be strongly impacted by the e e mentoring process only in regard to his or her self-efficacy and self-presentation skills. Although the mentoring may not affect the individual in other areas, the changes in the prot´ g´ on self-efficacy and self-presentation are sufficient to e e improve his or her job performance and chances for promotion. Proposition 2. Significant prot´ g´ change in one area, rather than multiple e e areas, may positively impact extrinsic and/or intrinsic distal outcomes. Our model specifies two additional proximal outcomes, satisfaction with mentor and the mentoring program. Although listed together, these outcomes are distinct; one can be high and the other low or both could be high or low. Ragins et al. (2000) found prot´ g´ satisfaction with mentor was related to an array of pere e ceptual outcomes, including higher prot´ g´ career commitment, job satisfaction, e e organizational commitment, and perceived opportunities for promotion. Perceived program effectiveness was related to all of these outcomes with the exception of perceived opportunities for promotion. We suggest that mentor and program satisfaction are likely to impact, in a positive direction, the career success outcomes for the prot´ g´ that are more affective in nature (i.e. job satisfaction). The more objece e tive career success outcomes are proposed to be affected through the prot´ g´ change e e outcomes or directly through mentor actions (to be described later). We propose: Proposition 3. Prot´ g´ satisfaction with one’s mentor and the mentoring proe e gram will have a positive impact on prot´ g´ job satisfaction, career satisfaction, e e career commitment, and life satisfaction. Mentoring Received Central to the conceptual model is the mentoring received in the relationship. Drawing upon the interpersonal relationship literature and specific conceptualizations of relationship closeness (Berscheid, Snyder & Omoto, 1989), we suggest formal mentoring can be described in terms of its frequency, scope, and strength of influence. Frequency, Scope, and Strength of Influence. Frequency refers to the prevalence of meetings between the mentor and prot´ g´ , influencing the amount of time mentors e e and prot´ g´ s spend together. Berscheid et al. (1989) reasoned that “the more time e e

96

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

people spend together, the more opportunity they have to influence each other’s thoughts and behaviors” (p. 794). We similarly suggest that simply spending time together is an important starting point for mentors and prot´ g´ s in a formal mene e toring program. Scope indicates the breadth of mentoring functions received by the prot´ g´ e e (Kram, 1985b) in tandem with the breadth of subjects addressed or discussed during the duration of the mentoring relationship. At the highest level of scope of mentoring received would be a relationship that involved a realm of career and psychosocial-related mentoring functions and covered a variety of subjects. At a low level of scope of mentoring received, discussions may be focused on one topic of interest to the prot´ g´ (e.g. increasing assertiveness), and the functions provided e e by the mentor might fall primarily into the coaching and counseling domains. Recognition of the subject aspect of the scope of mentoring permits a fuller understanding of the range of issues being discussed in the mentoring relationship and helps to ensure the content validity of assessments of mentoring functions that involve subject content. For example, the Mentor Role Instrument examines coaching through a three item scale. The items include: My mentor . . . “Helps me learn about other parts of the organization”; “Gives me advice on how to attain recognition in the organization”; and “Suggests specific strategies for achieving career aspirations” (Ragins & Cotton, 1999, p. 550). Other coaching topics that may be salient in formal relationships (e.g. resolving a current work challenge, work life balance) are not assessed through these items. When used, this may make it appear as if a relationship did not involve coaching if the coaching focused on areas that are not represented in the scale’s items. Strength of influence refers to the degree to which individuals are influenced by the mentoring received. Specifically, the strength of the influence of the mentoring received can vary. Some formal mentors may meet frequently with their prot´ g´ s, e e sharing ideas and counsel, but their suggestions and ideas are superficial, they are not discussed in sufficient depth, or they do not meet prot´ g´ needs and are e e thus not influential. Some (e.g. “mentor assigned responsibilities to you that have increased your contact with people in the district who may judge your potential for future advancement”), but not all (e.g. “mentor has shared history of his/her career with you”) current mentoring items capture the concept of strength of influence (examples taken from Noe, 1988a, p. 469). The concepts of frequency, scope, and strength of influence are important to understanding the nature of the mentoring received by an individual. We highlight the importance of these concepts with the following proposition. Proposition 4. Higher frequency, scope, and influence of mentoring received within a formal mentoring program will be associated with more extensive prot´ g´ change and higher satisfaction with the mentor and program. e e

Mentoring Research

97

We suggest that the relationship between mentoring received and more distal career success outcomes for the prot´ g´ is partially, but not fully, mediated by e e prot´ g´ change and mentor and program satisfaction. First, it is our expectation e e that formal mentoring exerts at least part of its impact on the more distal career outcomes through its effect on prot´ g´ change (cognitive learning, skill-based e e learning, affective learning, and improving social networks) and mentor and program satisfaction. However, it is also possible for formal mentoring to impact the distal career outcomes directly. For example, if the mentor were to exert any protection or sponsorship functions without the knowledge of the prot´ g´ , these e e functions could directly influence the prot´ g´ ’s career success without impacting e e the prot´ g´ ’s learning or mentor or program satisfaction. We propose: e e Proposition 5. The relationship between mentoring received and more distal career success outcomes for the prot´ g´ is partially, but not fully, mediated by e e prot´ g´ change and satisfaction. e e Sponsorship, protection, and exposure. Here we elaborate on our expectations regarding the receipt of three specific mentoring functions within a formal mentoring relationship. While acknowledging that all traditionally accepted mentoring functions (e.g. Kram, 1985b) can occur in a formal mentoring relationship, we suggest that the level of sponsorship, protection, and exposure received by prot´ g´ s in formal relationships will be quite low, especially in comparison to their e e receipt of other mentoring functions (e.g. coaching, challenging, role modeling, acceptance and confirmation, counseling, and friendship). Sponsorship, protection, and exposure all require the mentor to use his or her influence on behalf of the prot´ g´ , something less likely to take place within a limited duration (e.g. one e e year) formal mentoring program that may also involve prot´ g´ /mentor dyads that e e are separated geographically. Furthermore, while sponsorship, protection, and exposure functions are not prohibited, these functions are not explicitly encouraged or communicated as being part of relationships in formal mentoring programs. The emphasis of formal mentoring programs is typically on the prot´ g´ s achieving e e self-reliance, rather than teaching them to depend on the sponsorship or protection of a senior person (Clutterbuck, 2002). Furthermore, an organization’s suggestion that formal mentors engage in functions such as sponsorship, protection, or exposure could lead to organizational justice problems among non-participants or participants not receiving such functions (Kram, 1988). Indeed, many organizations use the term “mentee” rather than “prot´ g´ ” to refer to the more junior e e participants in a formal mentoring program. The formal definition of prot´ g´ , “one e e who is protected, trained, or guided by an influential person” (Merriam Webster Dictionary, 1995, p. 418) is distasteful to organizations that do not endorse “protection” as a formalized mentoring function. Specifically, to delineate our

98

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

expectation regarding the scope of mentoring received in a formal relationship, we propose: Proposition 6. Sponsorship, protection, and exposure will occur only rarely in a formal mentoring environment. Given Proposition 6, studies comparing formal and informal mentoring programs in terms of their effectiveness should examine mentoring functions at the specific function level (e.g. sponsorship) rather than combining the functions to form broader career-related and psychosocial-related function scales. Sponsorship, protection, and exposure are typically regarded as career-related mentoring functions. If career-related mentoring functions are computed as a sum of sponsorship, protection, and exposure along with coaching and challenging work, formal programs evaluated with current mentoring function measures will likely appear as less effective than informal relationships even when formal programs do not desire their dyads to be entering into the sponsorship, protection, and exposure domains.

Participant/Relationship Antecedents Our conceptual model portrays prot´ g´ , mentor, and dyad characteristics, along e e with characteristics of the prot´ g´ /mentor relationship, as important antecedents e e to the mentoring received in a formal mentoring program. We describe the role of these antecedents, beginning with the characteristics of the prot´ g´ /mentor relae e tionship. Relationship Characteristics Drawing upon the relationship literature (see, for example, Hinde, 1995, 1997), we identified four primary relationship characteristics (intimacy, interpersonal perception, conflict, and complementary nature of interactions) that are expected to be critical antecedents to the mentoring that is received by prot´ g´ s in formal e e mentoring relationships. Intimacy refers to the closeness that the formal mentor and the prot´ g´ are e e able to achieve in their assigned relationship. This closeness is indexed by the extent to which the two individuals reveal themselves to each other cognitively or emotionally, and is related to the level of trust and comfort in the relationship (Hinde, 1995). Ragins and Cotton (1999) suggest that establishing intimacy or closeness within a formal relationship is a unique challenge. They note, for example, that the mentor and prot´ g´ “recognize that the relationship is short-term e e and that the mentor may be assigned to another prot´ g´ after the relationship is e e over” (p. 531). To the extent that intimacy in the formal relationship is achieved,

Mentoring Research

99

we expect that a prot´ g´ will be more likely to meet with the mentor and to disclose e e concerns about work-related issues, even when the issues are not comfortable ones to discuss. Likewise, within relationships with higher levels of intimacy it is expected that the mentor will be more likely to confess previous work-related mistakes or personal “tricks of the trade” (Kalbfleisch & Davies, 1993, p. 403). Hinde (1995) conceptualizes interpersonal perception as the extent to which two individuals in a relationship feel understood. Translating to the formal mentoring context, his ideas suggests that mentors and prot´ g´ s will meet more frequently e e and engage in more meaningful discussion if they understand one another and feel understood by one another. Hinde’s work suggests several levels of understanding and perceived understanding in the relationship are important. For example, it is ideal if the mentor understands how the prot´ g´ sees himself (e.g. the mentor e e knows that the prot´ g´ believes his presentations are effective), yet sees the e e prot´ g´ as he/she really is (e.g. the mentor recognizes that the prot´ g´ needs to e e e e improve his presentation style). It is furthermore important that the prot´ g´ feels e e understood. For example, a prot´ g´ discussing work/life balance issues might feel e e misunderstood if he or she feels that the mentor interprets the discussion as a lack of work commitment. Finally, the relationship has higher levels of interpersonal perception to the extent the mentor also feels understood by the prot´ g´ , and the e e extent to which the mentor and prot´ g´ see their relationship in similar terms. e e We define conflict as “a process in which one party perceives that its interest are being opposed or negatively affected by another party” (Wall & Callister, 1995, p. 517). Although Kalbfleisch (1997) focused on conflicts within student/mentor relationships, we speculate that the types of conflict that were identified in these relationships could also translate to the formal context, including conflicts stemming from disagreement, embarrassment, negativity, or requests that are made on the part of the mentor that are perceived to be inappropriate or illegitimate to the prot´ g´ . We subsume under the umbrella of conflict any “lack of agreement or ace e ceptance of where power lies” that may occur in the mentoring relationship (Hinde, 1995, p. 7) as well as dysfunctional behaviors leading to conflict in the relationship (Scandura, 1998). Conflict may generate anger, stress, or distrust of the other party, and may reduce both the quality and amount of communication between the two individuals (Wall & Callister, 1995). Despite its negative connotation, conflict can be constructive if the issues are trivial, important to discuss, and/or if the two members are committed to maintaining the relationship (Hinde, 1995). Complementary nature of interactions reflects the extent to which mentor and prot´ g´ exchanges involve a rich and satisfactory interplay between each e e members’ needs and offerings (Hinde, 1995). To the extent that a prot´ g´ has e e certain development goals that the mentor has the experience and ability to assist him or her with, a relationship has the potential to include interactions that

100

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

are complementary in nature. Specifically, one partner has a need (e.g. how to handle a particularly demanding client) that the other partner can address. The interactions are even more complementary if they fulfill needs for both partners, for example, if the mentor feels fulfilled by sharing his or her knowledge. Consistent with our review of the handful of studies exploring the dynamics of informal and formal mentoring relationships (e.g. Ensher, Thomas & Murphy, 2001; Heimann & Pittenger, 1996; Mullen, 1998; Mullen & Noe, 1999; Orpen, 1997), these relationship characteristics are expected to have a reciprocal relationship with mentoring received by the prot´ g´ . First, relationships characterized e e at any given time by more intimacy, greater feelings of being understood, the absence of destructive conflict, and higher complementarity are likely to lead to more mentoring. Reciprocally, the more frequently individuals meet, and the more mentoring that occurs in the relationship, the more likely intimacy and interpersonal perception will increase. However, more mentoring received may not automatically lead to reduced conflict or increased complementarity. We propose: Proposition 7. Higher intimacy, higher interpersonal perception, lower levels of conflict, and higher complementarity within the relationship are expected to facilitate a higher frequency, broader scope, and stronger influence of mentoring. Proposition 8. Higher levels of frequency, scope, and influence of mentoring are expected, in turn, to increase the intimacy and interpersonal perception within the relationship. Mentor, Prot´ g´ , and Dyad Characteristics e e Each member of the formal mentoring relationship brings to the relationship his or her unique demographic background, ability levels, personality, attitudes, and job/career history. The relationship literature (Berscheid, Lopes, Ammazzalorso & Langenfeld, 2001; Hinde, 1997) suggests that while the unique characteristics of each individual are important, the interaction of the characteristics of two individuals is particularly critical in determining the characteristics of the relationship. For example, Kelley et al. (1983) gave the example that the unique interaction between the characteristics of the two individuals is predictive of whether two individuals strike “it off well” or have “a special chemistry for each other” (p. 55). There has been relatively little attention given to the interaction among prot´ g´ e e and mentor characteristics beyond interactions among the race and gender of the prot´ g´ and mentor (Ragins, 1999). An exception to this is recent work by e e Godshalk and Sosik (in press). In a study of mentors and prot´ g´ s who may have e e been engaged in either formal or informal relationships, Godshalk and Sosik found that when both the mentor and prot´ g´ had high levels of learning goal e e orientation, a high level of mentoring was reported. In contrast, when members

Mentoring Research

101

of the dyad had different or low levels of learning goal orientation, lower levels of mentoring were reported. Our model would suggest that this unique interaction of mentor and prot´ g´ characteristics (including demographic background and other e e individual differences) influences the mentoring received through its influence on the characteristics of the relationship. Specifically, a nice mesh of the multitude of characteristics of the mentor and prot´ g´ influences the development of a e e relationship that is more intimate, where individuals understand one another, that is lower in conflict, and higher in interactions that are complementary. Consider a mix of race, ethnicity, gender, age, disability, or sexual orientation between the mentor and prot´ g´ (Ragins, 2002). While dissimilarities on some of these chare e acteristics might make it difficult for pairs to achieve intimacy and interpersonal perception, this dissimilarity may be overcome if both members of the dyad have high levels of openness to experience and agreeableness (McCrae & Costa, 1996) or use similar strategies to address potentially divisive issues (Thomas, 1993). In summary, we propose that characteristics of the dyad will play an important role in influencing the characteristics of the relationship that is formed between the mentor and the prot´ g´ . e e Proposition 9. The characteristics of the dyad, represented by the rich interaction of the myriad prot´ g´ and mentor characteristics, will influence the level e e of intimacy, interpersonal perception, conflict, and complementarity in a formal mentoring relationship. We do not wish to discount the role that that the unique characteristics of the mentor and prot´ g´ can play in the mentoring process. Our model portrays mentor e e and prot´ g´ characteristics as impacting the mentoring received not only through e e the dyad and relationship characteristics, but also directly. We additionally portray prot´ g´ characteristics as having a direct effect on both the proximal and distal e e outcome variables. These direct paths between the prot´ g´ ’s characteristics and e e the developmental outcomes are supported by recent meta-analytic work in the training literature (Colquitt, LePine & Noe, 2000). Specifically, these authors found that individual characteristics had both direct and indirect relationships with training motivation, learning levels, transfer of training, and job performance, rather than being fully mediated through more proximal variables. We expect the same prot´ g´ and mentor characteristics that have been found in e e the informal mentoring literature to be associated with higher levels of mentoring received, and refer the reader back to our section on predictors within the informal mentoring context. We wish to highlight, however, our expectations around one personality characteristic, proactivity, that has not been widely discussed within the informal mentoring literature. Our review of the formal mentoring process along with the career success literature highlighted proactivity as an important

102

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

prot´ g´ characteristic that would seem to drive the frequency of meetings between e e the mentor and the prot´ g´ , as well as the scope of topics discussed. e e Proactivity is an individual’s tendency to engage in activities meant to influence one’s environment (Bateman & Crant, 1993). Proactivity is significantly related to two of the five dimensions of the Big Five model of personality, extraversion (0.25) and conscientiousness (0.43) and is also related to need for achievement, need for dominance, and participation in charitable activities and professional associations (Bateman & Crant, 1993). Because individuals with proactive personalities tend to identify opportunities and follow through with them, seek solutions to barriers in their way, and show initiative in other important ways (Seibert, Crant & Kraimer, 1999), we expect that proactivity is likely to impact the frequency of meetings that the prot´ g´ requests with the mentor. Many formal e e mentoring programs suggest that the prot´ g´ take the initiative in the mentoring e e relationship; that the prot´ g´ should not rely on the mentor to schedule meetings e e and goals for the relationship (Coley, 1996). This may be difficult for a prot´ g´ e e who feels uncomfortable “bothering” the mentor, or who tends to shy away from seeking out advice or counsel. We expect that prot´ g´ s with a proactive disposition e e will be more likely to schedule regular meetings with the mentor, rather than waiting to hear from the mentor. We further expect that proactive individuals will be more likely to help guide the mentoring process, bringing questions and setting goals for the relationship. Finally, because proactivity has been shown to be associated with both intrinsic and extrinsic career success outcomes (Seibert et al., 1999), we expect that this characteristic will also have a direct impact on both the proximal and distal mentoring outcomes. We propose: Proposition 10. Higher levels of prot´ g´ proactivity will be associated with a e e higher frequency, scope, and influence of mentoring, increased prot´ g´ change, e e and more positive intrinsic and extrinsic career success outcomes.

Program Characteristics Based on our literature review, we identified six primary program characteristics proposed to have direct implications for the mentoring that will occur in a formal relationship. In addition, we propose that certain program components may be in the position to influence relationship characteristics. First, criteria for the selection of participants seems to vary across formal mentoring programs. While the issue of who should participate in the program (e.g. new hires, high potentials, or all managers) is specific to each program’s objectives, an issue that is relevant to all programs is whether participation

Mentoring Research

103

is voluntary or required. Practitioners seem to agree that results of formal mentoring programs will be best if participation is voluntary on the part of the mentor (Gaskill, 1993; Gray & Gray, 1990; Newby & Heide, 1992). Mentors participating voluntarily are expected to be more motivated to be involved in the relationship, and will make themselves more accessible to their prot´ g´ s than e e those “arm-twisted” by their organizations to be in the program. Although the effect size was small, Ragins et al. (2000) found that prot´ g´ s perceived programs e e as more effective if their mentor had entered the program voluntarily. In contrast, there is less support for the idea that it is critical for prot´ g´ s e e to participate in formal mentoring programs voluntarily. Prot´ g´ voluntary e e participation was not significantly related to perceived program effectiveness in the Ragins et al. (2000) study, and it is not uncommon for organizations to include all individuals in a given group (e.g. new hires) in a formal program. Most individuals included in a formal program are likely pleased that they have been included in the program, given the program’s intent is commonly to benefit their careers. Nevertheless, research has developed solid support for the importance of allowing individuals to participate in decisions regarding their training and development (Quinones, 1997). We surmise that voluntary participation may impact frequency of meeting for both parties, with fewer meetings occurring on the average among mentors and prot´ g´ s required to be in the program. e e Proposition 11. Voluntary participation on the part of mentors and prot´ g´ s e e will be related to higher frequency of meeting. Another program characteristic is the method of matching prot´ g´ s and mentors. e e Two issues are of interest: (1) how much participant input to use in the matching process; and (2) what makes a good match. First, organizations vary in the extent that their participants contribute to the matching process (Viator, 1999). On the higher extreme of participant input, organizations allow mentors or prot´ g´ s to e e choose one another or allow participants to specify their three top partner choices (Coley, 1996). In the middle of the spectrum, an organization may survey and then match the prot´ g´ and mentor on business-related criteria (e.g. area prot´ g´ e e e e wants to gain experience in), human interest factors (e.g. hobbies, interests), learning styles, and/or geographical location (Forret, 1996; Gaskill, 1993; Hale, 2000; Tyler, 1998). At the low participation extreme (no input), mentors and prot´ g´ s are matched by the program planner without being solicited for any type e e of information. Expert opinion seems to favor allowing the prot´ g´ to have some level of ine e put into the matching process. For example, Chao et al. (1992) endorsed participant input by comparing “random assignment of prot´ g´ s to mentors” to “blind e e dates” (p. 634). Research by Viator (1999) suggests that allowing the prot´ g´ to e e

104

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

have input into the matching process, whether that input is more extensive (where prot´ g´ s are allowed to choose among mentors) or less extensive (where prot´ g´ s e e e e are instead surveyed for information that is used in the matching process) is associated with higher prot´ g´ satisfaction with the mentor. While Ragins et al. e e (2000) found no relationship between method of matching (prot´ g´ choice versus e e no choice) and prot´ g´ satisfaction with the mentor, their measurement combined e e mentors and prot´ g´ s that were assigned to each other without any input at all with e e those that were assigned following a survey of interests into a “no choice” category. It is likely that prot´ g´ input leads to increased prot´ g´ psychological ownership e e e e of and commitment to the mentoring process, and that it facilitates the relationship by matching individuals having more in common. Furthermore, prot´ g´ s matched e e with mentors based on their business goals and interests may be more likely to be in a relationship that meets their developmental needs. In following, we suggest that prot´ g´ input in the matching process is likely to be associated with increased mene e tor/prot´ g´ intimacy, interpersonal perception, complementarity of interactions, e e meeting frequency, and strength of influence. Specifically, we propose: Proposition 12. Formal mentoring programs that match individuals through a method that solicits participant input (i.e. either through allowing choice of partner or by eliciting interests and goals from the participants) will have pairs with higher levels of intimacy, interpersonal perception, complementarity, meeting frequency, and strength of influence. Assuming the organization is going to make matches based on prot´ g´ input, e e the literature gives little guidance about what concrete factors should be used to develop an effective match. The mentoring literature is supportive of the mentor and prot´ g´ having some similarities (e.g. to facilitate comfort in the relationship). e e For example, the results of three empirical studies found that perceptions of similarity are positively associated with mentoring functions (Burke et al., 1993; Ensher & Murphy, 1997; Nielson et al., 2001). The dimensions on which similarity was rated were fairly broad in scope, including intelligence, personality, background, procedures, and activities outside of work (Burke et al., 1993), values and attitudes concerning work and family balance (Nielson et al., 2001), and “outlook, perspective, and values” (Ensher & Murphy, 1997, p. 469). Despite this support for similarity between the mentor and prot´ g´ , it is also e e believed that individuals will develop most if exposed to views and experiences unlike their own (see, for example, Hale, 2000). Although our recommendation is generic in nature (e.g. we are unable to propose specific scales or items to make this recommendation more concrete), we propose that the most effective matches will be those where the mentor and prot´ g´ are matched on at least one dimension e e so they have something in common as a basis for the relationship (whether it be

Mentoring Research

105

a common interest in baseball or both having children of the same age) and on at least one dimension that is suggestive of the mentor having a skill or background that complements the prot´ g´ ’s needs. We propose that this approach will enhance e e the level of intimacy, interpersonal perception, and complementarity of interactions in a relationship. Specifically, by matching individuals so that they have at least something in common, their level of intimacy and interpersonal perception may be enhanced. By matching prot´ g´ s to mentors that may be able to meet their e e needs, the complementarity of the interactions will be enhanced. We propose: Proposition 13. Organizations that are more effective at matching prot´ g´ s e e and mentors with a basis of similarity (e.g. an interest in common) but some complement of each others’ needs will have mentor/prot´ g´ pairs with higher e e levels of intimacy, interpersonal perception, and complementary interactions. We put forth an additional proposition about geographical proximity. Many formal mentoring programs now match prot´ g´ s with mentors in other locations, e e sometimes overseas, and contact has to be primarily over the phone and through e-mail. Indeed, there are an increasing number of e-mentoring programs, programs through which mentors and prot´ g´ s are connected through e-mail, web sites, and e e electronic discussion lists (Single & Muller, 2001). We acknowledge the necessity of assigning prot´ g´ s to mentors in other locations; this is sometimes unavoidable e e (Jossi, 1997). However, evidence is suggestive that formal mentoring programs that involve long-distance mentoring will not be as effective as mentoring programs whose participants can meet face-to-face. For example, Burke et al. (1993) found that office proximity was related to provision of more mentoring, and opportunities for interaction on the job have been identified as an important predictor of the level of mentoring provided in a number of studies (Aryee, Chay & Chew, 1996; Noe, 1988a). Based on this research, we suggest that pairs in more proximal locations will report higher meeting frequency, mentoring scope, and strength of influence. Proposition 14. Prot´ g´ s more proximally located to their mentors will report e e higher levels of meeting frequency, mentoring scope, and strength of influence. Berscheid et al. (1989) suggest that even though individuals can be close when separated by distance, they will likely not be as close as individuals who spend time together in person. Although pairs can be encouraged to communicate frequently despite geographical separation, we expect that pairs in more proximal locations will report higher levels of intimacy and interpersonal perception than individuals separated by greater distance. Proposition 15. Prot´ g´ s more proximally located to their mentors will report e e higher levels of relationship intimacy and interpersonal perception.

106

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Authors outlining “best practice” guidelines for formal mentoring programs suggest that an orientation or training session be provided for mentors and prot´ g´ s e e (e.g. Phillips-Jones, 1983). Orientations for prot´ g´ s may include information e e about the purpose of the program; discussions about mentoring functions, benefits, and limitations; and exercises or handouts on handling conflicts that may occur during the relationship (Forret, 1996). Forret suggests that it is important that the program facilitator clarifies expectations, for example by discussing whether the formal mentoring program is meant to help the prot´ g´ s’ general introduction e e to the organization or their career development. Orientations for mentors might similarly include discussions about mentoring responsibilities, functions, benefits, and limitations, along with workshops on mentorship skills such as listening and problem solving methods (Gaskill, 1993). We first propose three means by which program orientations might have an impact on relationship characteristics. To the extent that the orientation provides quality training on avoiding destructive conflict in the mentoring relationship (e.g. by providing a discussion of appropriate roles and potentials for abuse of power in the relationship) along with conflict resolution information, we propose that program orientations can reduce the occurrence, level, and seriousness of destructive conflict in the mentor/prot´ g´ relationships. To the extent that the e e orientation provides quality information about listening skills and skills that can enhance interpersonal perception, we propose that orientations can improve the level of interpersonal perception in mentor/prot´ g´ relationships. Last, we e e propose that programs that suggest activities that mentors may engage in with prot´ g´ s (see for example, Coley, 1996) and means by which the prot´ g´ and e e e e mentor can begin to develop their relationship can positively enhance the level of intimacy experienced by individuals in the relationship. Proposition 16. Effective orientations for the mentor and prot´ g´ will reduce e e destructive conflict and increase the intimacy and interpersonal perception in the prot´ g´ /mentor relationship. e e Some organizations discuss guidelines for frequency of meeting at orientation (e.g. suggesting meetings of once or twice a month over the duration of the program). Findings by both Viator (1999) and Ragins et al. (2000), along with studies suggesting more frequent meetings are associated with more mentoring (Burke et al., 1993; Mullen, 1998; Waters et al., 2002), are supportive of setting frequency of meeting guidelines. The mentor is typically a person of more seniority and influence than the prot´ g´ . Although some prot´ g´ s may not be bothered by this seniority, we proe e e e pose that frequency of meeting guidelines give prot´ g´ s both the permission and e e expectation that they are to contact their mentors, reducing issues related to intimidation or reluctance to bother the mentor. Frequency of meeting guidelines are also

Mentoring Research

107

proposed to guard against the prot´ g´ and mentor getting so busy with their work e e that they feel they cannot take the time out to meet (Noe, 1988a); the guidelines may act as a work assignment or meeting that has been previously agreed upon. In the absence of meeting guidelines it is easy for time to pass quickly and for prot´ g´ s e e and mentors to more easily discount the need to meet. We propose the following: Proposition 17. Prot´ g´ s in programs with frequency of meeting guidelines e e will meet with their mentors more often than prot´ g´ s in programs without e e frequency of meeting guidelines. Orientation sessions for prot´ g´ s in mentoring programs focused on development e e may request that participants conduct a skills assessment. A goal setting process can then be used to encourage prot´ g´ s to specify competencies or issues they e e wish to work on in the mentoring relationship (Coley, 1996). Newby and Heide (1992) note that prot´ g´ goals for the relationship “help to delineate differences e e between present and desired levels of performance, identify the needed resources, and serve as reference points for comparison and adjustment as changes are attempted” (p. 10). A wealth of research in work, educational, and training contexts supports the notion that goals are motivational in nature, showing individuals who develop challenging and concrete goals exert more effort (Kanfer, 1990; Locke & Latham, 2002) and that more motivated individuals learn more and acquire more knowledge and skills (Colquitt et al., 2000). The nature of prot´ g´ goals are also e e expected to influence the mentoring received. Kram and Bragar (1992) note that some prot´ g´ s are more interested in receiving career-related functions such as e e coaching, while others are more interested in receiving psychosocial functions, such as support and counseling. We propose that individuals who set goals will meet more frequently with their mentors, due to having a clear picture of what they wish to achieve. We also propose that the nature of the goal will influence the nature of the mentoring received. Proposition 18. Prot´ g´ s that set specific and challenging goals to work on e e through their mentoring relationship will meet more frequently with their mentor. Proposition 19. The nature of prot´ g´ goals will influence the scope of the e e mentoring received. Last, we propose that program objectives (the purpose of the mentoring program) will influence the nature of the mentoring that occurs and the outcomes achieved. Consistent with Kram and Bragar (1992), we feel it is important to recognize that the focus of mentoring within programs that are for newcomers (e.g. having a socialization objective) will be different from the focus of mentoring within

108

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

other programs, such as those developed for high potential managers (e.g. having a management development or succession planning objective). As in any other training or development evaluation effort, the outcomes examined should parallel the program objectives (Gray & Gray, 1990). Indeed, Ragins et al. (2000) found that programs aimed at career development were associated with higher levels of prot´ g´ satisfaction with opportunities for promotion than programs aimed at e e socialization. We propose the following: Proposition 20. The nature of the program objectives will influence the scope of the mentoring received and thus ultimately the outcomes achieved.

Organizational Context Importantly, we have enclosed our model within an outer box representing the “organizational context” in order to portray formal mentoring relationships as embedded within a larger organizational context. Salient to the organizational context is the organization’s culture, or its members’ characteristic values and attitudes (Ashkanasy & Jackson, 2001). For example, relevant to mentoring programs focused on employee development are the organization’s deeply ingrained values and beliefs about the importance of continuous learning and development (London & Smither, 1999). Relevant to mentoring programs focused on socialization would be the messages the organization sends about the importance of newcomers gaining a facilitated introduction to the organization. To illustrate the importance of culture, consider a prot´ g´ whose job involves e e demanding, heavily scheduled days and many evenings of working late. This prot´ g´ may place priority on his or her pressing job demands, postponing meete e ings with his or her mentor until soon the year devoted to the mentoring program has passed. Prot´ g´ s surrounded by a climate emphasizing continuous learning e e may be more likely to make the time for the meetings with the mentor, having been persuaded that development time is critical. Similarly, mentors may have more motivation to spend time with prot´ g´ s when they perceive that mentoring e e is a valued organizational activity (Aryee, Chay & Chew, 1996). We propose: Proposition 21. Organizations with cultures that communicate values and attitudes supportive of the mentoring program’s objectives will have mentor/prot´ g´ e e dyads that will meet more frequently. At a more micro level, it is also important that mentors and prot´ g´ s perceive e e that their supervisors are supportive of their involvement in the mentoring program. Research, for example, suggests that perceived supervisory support

Mentoring Research

109

for development activity is associated with higher levels of voluntary learning and development activity on the part of employees (see, for example Maurer & Tarulli, 1994; Noe & Wilk, 1993). Proposition 22. To the extent that the mentor’s and prot´ g´ ’s superiors are e e supportive of their involvement in the mentoring program, mentoring program outcomes are expected to be enhanced. Also relevant are other training and developmental opportunities that are available to the prot´ g´ (Tharenou, 1997). For example, a mentor’s counseling may help e e the prot´ g´ realize that he or she needs some intensive presentation skills training. e e If presentation skills training or a referral to such training is available within the organization, the prot´ g´ may be more likely to act upon that developmental need. e e In other words, we propose that prot´ g´ change can be enhanced by the existence e e of employee development workshops or related development opportunities. To take this one step further, it is important to consider the existence of other opportunities when predicting career success outcomes to avoid attributing career success uniquely to mentoring when mentoring was supplemented by other development opportunities. Proposition 23. Prot´ g´ change and more distal career success outcomes will e e be enhanced by the existence of employee development workshops or related development opportunities. Finally, several other contextual factors are relevant to the mentoring process and outcomes within the organizational context. An organization’s size, structure, and compensation and promotional opportunities may limit the observation of increased promotions and compensation as a result of the mentoring program. For example, mentoring may have more potential to influence promotions within a large organization with long promotion ladders and high vacancy rates in higher positions (Tharenou, 1997). We suggest: Proposition 24. Other contextual factors are relevant to the outcomes observed as a consequence of a formal mentoring program, including an organization’s size, structure, and compensation and promotional systems.

Dynamic Aspects of Conceptual Model Our model is a dynamic process model. First, it is conceivable that the importance of some predictors in our model varies over the duration of the relationship (Zaheer, Albert & Zaheer, 1999). The predictive role of dyad characteristics may

110

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

be especially dynamic in nature, changing over the duration of the relationship. For example, demographic dissimilarity may result in less mentoring early in the relationship, but may actually be beneficial later in the relationship (see, for example, Turban et al., 2002), although this may depend on the complex interplay of other characteristics mentioned earlier such as the openness to experience and agreeableness of both members. Importantly, levels of any of the variables in the model may change substantially during the relationship. For example, the scope, frequency, and strength of influence of prot´ g´ /mentor interactions may change over time. This change in e e mentoring received could occur due to a wide variety of reasons. As the prot´ g´ e e changes both as a result of the formal mentoring program as well as from becoming more experienced in the organization, his or her needs and goals may change. Mentors or prot´ g´ s may experience an increase in workload, making it difficult e e to meet. If conflict occurs (Scandura, 1998) or if a relationship is not meeting the prot´ g´ ’s or mentor’s expectations (Young & Perrew´ , 2000), frequency of e e e meeting may decline or the relationship may be terminated. One of the two individuals in the partnership may alternatively be relocated, leave the company, or be laid off, leading to termination of the relationship. Finally, most formal programs come to a conclusion after one year’s duration. It is unknown how many of these dyads continue into an informal relationship, or to what extent learning based on the mentor’s advice and teachings continue even after the relationship is ended. A final important issue is that prot´ g´ perceptions of the mentoring experience e e may be dynamic and may not initially reflect its actual benefits. For example, due to low perception accuracy (Atwater & Yammarino, 1997), the prot´ g´ e e may change as a result of the mentoring program but not realize it. The social influence literature, for example, argues that the norms, values, and beliefs of an influence agent may have an effect on an individual without that individual’s realization (Forgas & Williams, 2001). Dijksterhuis (2001) also presents evidence that individuals unknowingly change their behavior to match the behavior of individuals they are in close contact with, mimicking both basic behaviors such as gestures or posture to more complex behaviors such as helpfulness. Alternatively, the prot´ g´ may realize they have changed, but may not attribute the change to the e e mentoring program. This failure to attribute the change to the mentoring program might be either unintentional (the individual might believe they would have changed without the mentoring program) or intentional (individuals motivated to succeed in an organization may be motivated, if asked, to attribute their change to their own efforts, rather than acknowledging the helpfulness of the mentoring program). The prot´ g´ ’s perceptions may, however, change over time (Zaheer e e et al., 1999). A few years after the mentoring experience, the prot´ g´ may be e e better able to appreciate the relationship and the benefits of that relationship. This

Mentoring Research

111

speaks to caution when relying on self-report measures, although assessment of prot´ g´ change through other means is also very challenging. e e

Summary Our conceptual model represents a preliminary attempt to portray the antecedents and outcomes of mentoring received within a formal mentoring context. One contribution of the model is the “unfolding” of the mentoring criterion space, represented in our model through proximal and distal outcomes. Also important is the delineation of variables expected to be predictive of successful formal mentoring experiences and the process through which these antecedents lead to successful outcomes. Finally, another contribution of the model is its integration of useful theory and research from related literatures, such as the career success literature and the social psychology literature on relationships. For example, the social psychology literature on relationships helped to guide our delineation of “relationship characteristics” that emerge as a direct result of the mentor/prot´ g´ e e dyad combination and our conceptualization of mentoring received as involving the tripartite scope, frequency, and strength of influence. Given the limited research on formal mentoring, we hope our model will stimulate more empirical research in this area. Most studies of formal mentoring programs in organizations have compared outcomes for individuals in formal programs to outcomes for individuals with informal mentors or without mentors, rather than looking at factors that may differentiate between formal programs that are successful versus not successful. Although our model was discussed within the context of formal mentoring, researchers will benefit from using the model to direct their research efforts on informal mentoring. Indeed, the model’s components, excluding the program characteristics, are applicable in the informal context. We chose specifically to present the model within the formal mentoring context to stimulate future theory building and research in the formal arena. It is likely our model will also be useful for practitioners managing formal mentoring program efforts. Day and Allen (2002) recently suggested that “if program administrators knew the precise factors that contribute to beneficial outcomes they could put these factors to use in mentor training” (p. 25). While our model is far from precise, it is suggestive and informative of the many factors that may play a role in mentoring program success. On the outcome side, we also hope to convey that “mentoring program success” is highly dependent on the goals of both the organization and the prot´ g´ . e e As a last note, we suggest several caveats. First, we acknowledge that the model may be difficult to test in its entirety due to the large number of variables

112

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

within its realm. Second, refinement of the model will be required to more explicitly incorporate the role of additional concepts and variables relevant to the mentoring experience. Simply due to the vast number of potential variables and paths, it was difficult to include all variables that have been mentioned in the mentoring literature and all conceivable paths in the model. Finally, we note that examination of the model may require development of new measurement scales. As an example, some items in current mentoring function scales incorporate what we conceptualize as relationship characteristics (e.g. trust or intimacy) into the assessment of mentoring functions. For example, the Mentor Role Instrument reprinted in Ragins and Cotton (1999) includes “is someone I can trust,” to assess the friendship dimension (p. 550). Future mentoring function scale revisions may attempt to capture mentoring functions separately from relationship characteristics.

CONCLUSIONS
In 1988(b), Noe suggested that “research regarding the benefits of mentoring relationships is in its infancy” (p. 66). Since then, research has advanced our understanding of the benefits as well as drawbacks of mentoring relationships, and has delved further into related and important branches of research including the construct space of mentoring, mentoring within the context of diversity, other correlates of mentoring received, relationship dynamics, and formal mentoring programs. While a great deal of research has been conducted on mentoring in the last decade, our assessment is that the mentoring literature is still very young; perhaps analogous to a primary school level. As with a child in primary school, continued stimulation and insight is highly critical for the mentoring literature. We hope this review and the conceptual model provided within are useful in directing future research. In addition to relying on conceptual frameworks such as the one we provided, attention to several methodological issues in future mentoring research will help move the mentoring literature forward. Specifically, researchers need to provide clear and consistent definitions of mentoring to study participants (important so a participant would not be considered a prot´ g´ in one study and without a e e mentor in the next), differentiate between formal and informal mentoring, ensure measurement instruments are content valid and psychometrically sound, rely less on prot´ g´ self-reports, incorporate more dyad analysis into research, increase e e the use of longitudinal research, and include appropriate control variables. Researchers’ attention to these methodological issues when evaluating and refining models of mentoring will foster the maturation of the field.

Mentoring Research

113

NOTE
1. This meta-analysis includes studies that examined formal and informal relationships, although the majority were focused on informal mentoring relationships. At the time of our review, the meta-analysis was under review. Given that revisions of the meta-analysis may include computational changes or additional studies, we provide results at a general rather than specific level. The authors reported average correlations weighted by sample size as indicators of effect size (without corrections for other artifacts due to the small number of studies available). We indicate whether the authors found small (i.e. 0.30 or less), medium (i.e. 0.30–0.49) or large (i.e. 0.50 or higher) effect sizes (Cohen, 1988) in the relationships they analyzed rather than providing the exact effect size.

ACKNOWLEDGMENTS
We thank Tammy Allen, Zhaoli Song, and Andy Van de Ven for their comments and discussion relevant to parts of this review. In addition, we thank Naomi Hagen for her early involvement in this project.

REFERENCES
Allen, T. D., Day, R., & Lentz, E. (2002, April). Interpersonal comfort in mentorship: A mediating mechanism. Paper presented at the 17th Annual SIOP Conference, Toronto, ON. Allen, T. D., Eby, L. T., Poteet, M. L., Lentz, E., & Lima, L. (2002). Career benefits associated with mentoring for prot´ g´ s: A meta-analysis. Manuscript submitted for publication. e e Allen, T. D., & Poteet, M. L. (1999). Developing effective mentoring relationships: Strategies from the mentor’s viewpoint. Career Development Quarterly, 48, 59–73. Allen, T. D., Poteet, M. L., & Burroughs, S. M. (1997). The mentor’s perspective: A qualitative inquiry and future research agenda. Journal of Vocational Behavior, 51, 70–89. Allen, T. D., Poteet, M. L., & Russell, J. E. A. (2000). Prot´ g´ selection by mentors: What makes the e e difference? Journal of Organizational Behavior, 21, 271–282. Allen, T. D., Poteet, M. L., Russell, J. E. A., & Dobbins, G. H. (1997). A field study of factors related to supervisors’ willingness to mentor others. Journal of Vocational Behavior, 50, 1–22. Aryee, S., & Chay, Y. W. (1994). An examination of the impact of career-oriented mentoring on work commitment attitudes and career satisfaction among professional and managerial employees. British Journal of Management, 5, 241–249. Aryee, S., Chay, Y. W., & Chew, J. (1996). The motivation to mentor among managerial employees: An interactionist approach. Group & Organization Management, 21, 261–277. Aryee, S., Lo, S., & Kang, I.-L. (1999). Antecedents of early career stage mentoring among Chinese employees. Journal of Organizational Behavior, 20, 563–576. Aryee, S., Wyatt, T., & Stone, R. (1996). Early career outcomes of graduate employees: The effect of mentoring and ingratiation. Journal of Management Studies, 33, 95–118.

114

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Ashkanasy, N. M., & Jackson, C. R. A. (2001). Organizational culture and climate. In: N. Anderson, D. S. Ones, H. K. Sinangil & C. Viswesvaran (Eds), Handbook of Industrial, Work, & Organizational Psychology (Vol. 2, pp. 398–415). London: Sage Publications. Atwater, L. E., & Yammarino, F. J. (1997). Self-other rating agreement: A review and model. In: G. R. Ferris (Ed.), Research in Personnel and Human Resources Management (Vol. 15, pp. 121–174). Greenwich, CT: JAI Press. Bahniuk, M. H., Dobos, J., & Hill, S. E. K. (1990). The impact of mentoring, collegial support, and information adequacy on career success: A replication. Journal of Social Behavior and Personality, 5, 431–451. Bahniuk, M. H., Hill, S. E. K., & Darus, H. J. (1996). The relationship of power-gaining communication strategies to career success. Western Journal of Communication, 60, 358–378. Barbian, J. (2002, March). A little help from your friends. Training, 39, 38–60. Barker, P., Monks, K., & Buckley, F. (1999). The role of mentoring in the career progression of chartered accountants. British Accounting Review, 31, 297–312. Bateman, T. S., & Crant, J. M. (1993). The proactive component of organizational behavior: A measure and correlates. Journal of Organizational Behavior, 14, 103–118. Baugh, S. G., & Scandura, T. A. (1999). The effects of multiple mentors on prot´ g´ attitudes toward e e the work setting. Journal of Social Behavior and Personality, 14, 503–521. Berscheid, E., Lopes, J., Ammazzalorso, H., & Langenfeld, N. (2001). Causal attributions of relationship quality. In: V. Manusov & J. H. Harvey (Eds), Attribution, Communication Behavior, and Close Relationships. Advances in Personal Relations (pp. 115–133). Cambridge, England: Cambridge University Press. Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The relationship closeness inventory: Assessing the closeness of interpersonal relationships. Journal of Personality and Social Psychology, 57, 792–807. Blake-Beard, S. D. (1999). The cost of living as an outsider within: An analysis of the mentoring relationships and career success of Black and White women in the corporate sector. Journal of Career Development, 26, 21–36. Bolino, M. C., & Feldman, D. C. (2000). The antecedents and consequences of underemployment among expatriates. Journal of Organizational Behavior, 21, 889–911. Bono, J. E., Boles, T. L., Judge, T. A., & Lauver, K. J. (2002). The role of personality in task and relationship conflict. Journal of Personality, 70, 311–344. Boudreau, J. W., Boswell, W. R., & Judge, T. A. (2001). Effects of personality on executive career success in the United States and Europe. Journal of Vocational Behavior, 58, 53–81. Bridges, C. R., & Perotti, V. S. (1993). Characteristics of career achievement: Perceptions of AfricanAmerican corporate executives. Mid American Journal of Business, 8, 61–64. Broadbridge, A. (1999). Mentoring in retailing: A tool for success? Personnel Review, 28, 336–355. Burke, R. J. (1984). Mentors in organizations. Group & Organization Studies, 9, 353–372. Burke, R. J., & McKeen, C. A. (1996). Gender effects in mentoring relationships. Journal of Social Behavior & Personality, 11(5), 91–104. Burke, R. J., & McKeen, C. A. (1997). Benefits of mentoring relationships among managerial and professional women: A cautionary tale. Journal of Vocational Behavior, 51, 43–57. Burke, R. J., McKeen, C. A., & McKenna, C. S. (1990). Sex differences and cross-sex effects on mentoring: Some preliminary data. Psychological Reports, 67, 1011–1023. Burke, R. J., McKeen, C. A., & McKenna, C. (1993). Correlates of mentoring in organizations: The mentor’s perspective. Psychological Reports, 72, 883–896.

Mentoring Research

115

Burke, R. J., McKenna, C. S., & McKeen, C. A. (1991). How do mentorships differ from typical supervisory relationships? Psychological Reports, 68, 459–466. Busch, J. W. (1985). Mentoring in graduate schools of education: Mentors’ perceptions. American Educational Research Journal, 22, 257–265. Campion, M. A., & Goldfinch, J. R. (1983). Mentoring among hospital administrators. Hospital and Health Services Administration, 28(4), 77–93. Carden, A. D. (1990). Mentoring and adult career development: The evolution of a theory. The Counseling Psychologist, 18, 275–299. Chao, G. T. (1997). Mentoring phases and outcomes. Journal of Vocational Behavior, 51, 15–28. Chao, G. T. (1998). Invited reaction: Challenging research in mentoring. Human Resource Development Quarterly, 9, 333–338. Chao, G. T., O’Leary-Kelly, A. M., Wolf, S., Klein, H. J., & Gardner, P. D. (1994). Organizational socialization: Its content and consequences. Journal of Applied Psychology, 79, 730–743. Chao, G. T., Walz, P. M., & Gardner, P. D. (1992). Formal and informal mentorships: A comparison on mentoring functions and contrast with nonmentored counterparts. Personnel Psychology, 45, 619–636. Clutterbuck, D. (2002). Establishing and sustaining a formal mentoring program for working with diversified groups. In: D. Clutterbuck & B. R. Ragins (Eds), Mentoring and Diversity: An International Perspective (pp. 54–86). Woburn, MA: Butterworth Heinemann. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Colarelli, S. M., & Bishop, R. C. (1990). Career commitment: Functions, correlates, and management. Group & Organization Studies, 15, 158–176. Coley, D. B. (1996, July). Mentoring two-by-two. Training & Development, 50, 46–48. Collins, P. M. (1994). Does mentorship among social workers make a difference? An empirical investigation of career outcomes. Social Work, 39, 413–419. Colquitt, J. A., LePine, J. A., & Noe, R. A. (2000). Toward an integrative theory of training motivation: A meta-analytic path analysis of 20 years of research. Journal of Applied Psychology, 85, 678–707. Cox, T. H., & Nkomo, S. M. (1991). A race and gender-group analysis of the early career experience of MBAs. Work & Occupations, 18, 431–446. Darling, L. A. W. (1986). The mentoring mosaic: A new theory of mentoring. In: W. A. Gray & M. M. Gray (Eds), Mentoring: Aid to Excellence in Career Development, Business and the Professions, Proceedings of the First International Conference on Mentoring (Vol. 2, pp. 1–7). Vancouver, BC: International Association for Mentoring. Day, R., & Allen, T. (2002, April). Why are prot´ g´ s more successful: Explaining the mentoring-career e e success relationship. Paper presented at the 17th Annual SIOP Conference, Toronto, ON. Dijksterhuis, A. (2001). Automatic social influence: The perception-behavior links as an explanatory mechanism for behavior matching. In: J. P. Forgas & K. D. Williams (Eds), Social Influence: Direct and Indirect Processes (pp. 95–108). Philadelphia, PA: Psychology Press. Donaldson, S. I., Ensher, E. A., & Grant-Vallone, E. J. (2000). Longitudinal examination of mentoring relationships on organizational commitment and citizenship behavior. Journal of Career Development, 26, 233–249. Douglas, C. A., & McCauley, C. D. (1999). Formal developmental relationships: A survey of organizational practices. Human Resource Development Quarterly, 10, 203–224.

116

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Dreher, G. F., & Ash, R. A. (1990). A comparative study of mentoring among men and women in managerial, professional, and technical positions. Journal of Applied Psychology, 75, 539–546. Dreher, G. F., & Chargois, J. A. (1998). Gender, mentoring experiences, and salary attainment among graduates of an historically Black university. Journal of Vocational Behavior, 53, 401–416. Dreher, G. F., & Cox, T. H., Jr. (1996). Race, gender, and opportunity: A study of compensation attainment and the establishment of mentoring relationships. Journal of Applied Psychology, 81, 297–308. Eby, L. T. (1997). Alternative forms of mentoring in changing organizational environments: A conceptual extension of the mentoring literature. Journal of Vocational Behavior, 51, 125–144. Eby, L. T., McManus, S. E., Simon, S. A., & Russell, J. E. A. (2000). The prot´ g´ ’s perspective regarde e ing negative mentoring experiences: The development of a taxonomy. Journal of Vocational Behavior, 57, 1–21. Ensher, E. A., & Murphy, S. E. (1997). Effects of race, gender, perceived similarity, and contact on mentor relationships. Journal of Vocational Behavior, 50, 460–481. Ensher, E. A., Thomas, C., & Murphy, S. E. (2001). Comparison of traditional, step-ahead, and peer mentoring on prot´ g´ s’ support, satisfaction, and perceptions of career success: A social e e exchange perspective. Journal of Business and Psychology, 15, 419–438. Fagan, M. M., & Fagan, P. D. (1983). Mentoring among nurses. Nursing and Health Care, 4, 80–82. Fagan, M. M., & Walter, B. (1982). Mentoring among teachers. Journal of Educational Research, 76, 113–118. Fagenson, E. A. (1988). The power of a mentor: Prot´ g´ s and non-prot´ g´ s perceptions of their own e e e e power in organizations. Group & Organization Studies, 13, 182–194. Fagenson, E. A. (1989). The mentor advantage: Perceived career/job experiences of prot´ g´ s versus e e non-prot´ g´ s. Journal of Organizational Behavior, 10, 309–320. e e Fagenson, E. A. (1992). Mentoring: Who needs it? A comparison of prot´ g´ s’ and non-prot´ g´ s’ needs e e e e for power, achievement, affiliation, and autonomy. Journal of Vocational Behavior, 41, 48–60. Fagenson, E. A. (1994). Perceptions of prot´ g´ s’ vs. non-prot´ g´ s’ relationships with their peers, e e e e superiors, and departments. Journal of Vocational Behavior, 45, 55–78. Fagenson-Eland, E. A., Marks, M. A., & Amendola, K. L. (1997). Perceptions of mentoring relationships. Journal of Vocational Behavior, 51, 29–42. Federal Glass Ceiling Commission (1995, March). Good for business: Making full use of the nation’s human capital (executive summary). Retrieved from http://www.dol.gov/asp/ programs/history/reich/reports/ceiling1.pdf Forgas, J. P., & Williams, K. D. (2001). Social influence: Introduction and overview. In: J. P. Forgas & K. D. Williams (Eds), Social Influence: Direct and Indirect Processes (pp. 3–24). Philadelphia, PA: Psychology Press. Forret, M. L. (1996). Issues facing organizations when implementing formal mentoring programmes. Leadership & Organization Development Journal, 17(3), 27–30. Friedman, R., Kane, M., & Cornfield, D. B. (1998). Social support and career optimism: Examining the effectiveness of network groups among Black managers. Human Relations, 51, 1155–1177. Gaskill, L. R. (1991). Same-sex and cross-sex mentoring of female prot´ g´ s: A comparative analysis. e e Career Development Quarterly, 40, 48–63. Gaskill, L. R. (1993). A conceptual framework for the development, implementation, and evaluation of formal mentoring programs. Journal of Career Development, 20, 147–160. Gaskill, L. R., & Sibley, L. R. (1990). Mentoring relationships for women in retailing: Prevalence, perceived importance, and characteristics. Clothing and Textiles Research Journal, 9(1), 1–10.

Mentoring Research

117

Gilbert, J. A., & Ivancevich, J. M. (1999). A re-examination of organizational commitment. Journal of Social Behavior and Personality, 14, 385–396. Godshalk, V. M., & Sosik, J. J. (2000). Does mentor-prot´ g´ agreement on mentor leadership behavior e e influence the quality of a mentoring relationship? Group & Organization Management, 25, 291–317. Godshalk, V. M., & Sosik, J. J. (in press). Aiming for career success: The role of learning goal orientation in mentoring relationships. Journal of Vocational Behavior. Goto, S. (1999). Asian Americans and developmental relationships. In: A. J. Murrell, F. J. Crosby & R. J. Ely (Eds), Mentoring Dilemmas: Developmental Relationships within Multicultural Organizations. Applied Social Research (pp. 47–62). Mahwah, NJ: Lawrence Erlbaum Associates. Gray, M. M., & Gray, W. A. (1990). Planned mentoring: Aiding key transitions in career development. Mentoring International, 4(3), 27–32. Green, S. G., & Bauer, T. N. (1995). Supervisory mentoring by advisers: Relationships with doctoral student potential, productivity, and commitment. Personnel Psychology, 48, 537–561. Greenhaus, J. H., Parasuraman, S., & Wormley, W. M. (1990). Effects of race on organizational experiences, job performance evaluations and career outcomes. Academy of Management Journal, 33, 64–86. Hale, R. (2000). To match or mis-match? The dynamics of mentoring as a route to personal and organizational learning. Career Development International, 5, 223–234. Hall, D. T., & Mirvis, P. H. (1996). The new protean career: Psychological success and the path with a heart. In: D. T. Hall (Ed.), The Career is Dead Long Live the Career: A Relational Approach to Careers (pp. 15–45). San Francisco: Jossey-Bass. Heimann, B., & Pittenger, K. K. S. (1996). The impact of formal mentorship on socialization and commitment of newcomers. Journal of Managerial Issues, 8, 108–117. Higgins, M. C. (2000). The more the merrier? Multiple developmental relationships and work satisfaction. Journal of Management Development, 19, 277–296. Higgins, M. C., & Kram, K. E. (2001). Reconceptualizing mentoring at work: A developmental network perspective. Academy of Management Review, 26, 264–288. Higgins, M. C., & Thomas, D. A. (2001). Constellations and careers: Toward understanding the effects of multiple developmental relationships. Journal of Organizational Behavior, 22, 223–247. Hinde, R. A. (1995). A suggested structure for a science of relationships. Personal Relationships, 2, 1–15. Hinde, R. A. (1997). Relationships: A dialectical perspective. East Sussex, England: Psychology Press. Horgan, D. D., & Simeon, R. J. (1990a). Gender, mentoring, and tacit knowledge. Journal of Social Behavior and Personality, 5, 453–471. Horgan, D. D., & Simeon, R. J. (1990b). Mentoring and participation: An application of the VroomYetton model. Journal of Business and Psychology, 5, 63–84. Hough, L. M., & Ones, D. S. (2001). The structure, measurement, validity and use of personality variables in industrial, work, and organizational psychology. In: N. Anderson, D. S. Ones, H. K. Sinangil & C. Viswesvaran (Eds), Handbook of Industrial, Work & Organizational Psychology (Vol. 1, pp. 233–277). London: Sage Publications. Hough, L. M., & Schneider, R. J. (1996). Personality traits, taxonomies, and applications in organizations. In: K. R. Murphy (Ed.), Individual Differences and Behavior in Organizations (pp. 31–88). San Francisco: Jossey-Bass. Howard, G. S. (1994). Why do people say nasty things about self-reports? Journal of Organizational Behavior, 15, 399–404.

118

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Hubbard, S. S., & Robinson, J. P. (1998). Mentoring: A catalyst for advancement in administration. Journal of Career Development, 24, 289–299. Hunt, D. M., & Michael, C. (1983). Mentorship: A career training and development tool. Academy of Management Review, 8, 475–485. Johnson, N. B., & Scandura, T. A. (1994). The effect of mentorship and sex-role style on male-female earnings. Industrial Relations, 33, 263–274. Johnson, K. K. P., Yust, B. L., & Fritchie, L. L. (2001). Views on mentoring by clothing and textiles faculty. Clothing and Textiles Research Journal, 19(1), 31–40. Jossi, F. (1997, August). Mentoring in changing times. Training, 34, 50–54. Judge, T. A., & Bretz, R. D., Jr. (1994). Political influence behavior and career success. Journal of Management, 20, 43–65. Judge, T. A., Cable, D. M., Boudreau, J. W., & Bretz, R. D., Jr. (1995). An empirical investigation of the predictors of executive career success. Personnel Psychology, 48, 485–519. Judge, T. A., Higgins, C. A., Thoreson, C. J., & Barrick, M. R. (1999). The Big Five Personality traits, general mental ability, and career success across the life span. Personnel Psychology, 52, 621–652. Kalbfleisch, P. J. (1997). Appeasing the mentor. Aggressive Behavior, 23, 389–403. Kalbfleisch, P. J., & Davies, A. B. (1991). Minorities and mentoring: Managing the multicultural institution. Communication Education, 40, 266–271. Kalbfleisch, P. J., & Davies, A. B. (1993). An interpersonal model for participation in mentoring relationships. Western Journal of Communication, 57, 399–415. Kanfer, R. (1990). Motivational theory and industrial and organizational psychology. In: M. D. Dunnette & L. M. Hough (Eds), Handbook of Industrial and Organizational Psychology (2nd ed., Vol. 1, pp. 75–170). Palo Alto, CA: Consulting Psychologists Press. Kashy, D. A., & Levesque, M. J. (2000). Quantitative methods in close relationships research. In: C. Hendrick & S. Hendrick (Eds), Close Relationships: A Sourcebook (pp. 3–17). Thousand Oaks, CA: Sage Publications. Kelley, H. H., Berscheid, E., Christensen, A., Harvey, J. H., Huston, T. L., Levinger, G. et al. (1983). Analyzing close relationships. In: H. H. Kelley, E. Berscheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E. McClintock, L. A. Peplau & D. R. Peterson (Eds), Close Relationships (pp. 20–67). New York: W. H. Freeman and Company. Kirchmeyer, C. (1998). Determinants of managerial career success: Evidence and explanation of male/female differences. Journal of Management, 24, 673–692. Kirchmeyer, C. (2002). Change and stability in managers’ gender roles. Journal of Applied Psychology, 87, 929–939. Klauss, R. (1981). Formalized mentor relationships for management and executive development programs in the federal government. Public Administration Review, 41, 489–496. Koberg, C. S., Boss, R. W., Chappell, D., & Ringer, R. C. (1994). Correlates and consequences of prot´ g´ mentoring in a large hospital. Group & Organization Management, 19, e e 219–239. Koberg, C. S., Boss, R. W., & Goodman, E. (1998). Factors and outcomes associated with mentoring among health-care professionals. Journal of Vocational Behavior, 53, 58–72. Kraiger, K., Ford, J. K., & Salas, E. (1993). Application of cognitive, skill-based, and affective theories of learning outcomes to new methods of training evaluation. Journal of Applied Psychology, 78, 311–328. Kram, K. E. (1985a). Improving the mentoring process. Training and Development Journal, 39(4), 40–43.

Mentoring Research

119

Kram, K. E. (1985b). Mentoring at work: Developmental relationships in organizational life. Glenview, IL: Scott, Foresman & Company. Kram, K. E. (1988). Mentoring in the workplace. In: D. T. Hall (Ed.), Career Development in Organizations (pp. 160–201). San Francisco: Jossey-Bass. Kram, K. E., & Bragar, M. C. (1992). Development through mentoring: A strategic approach. In: D. H. Montross & C. J. Shinkman (Eds), Career Development: Theory and Practice (pp. 221–254). Springfield, IL: Charles C. Thomas. Laband, D. N., & Lentz, B. F. (1995). Workplace mentoring in the legal profession. Southern Economic Journal, 61, 783–802. Lankau, M. J., & Scandura, T. A. (2002). An investigation of personal learning in mentoring relationships: Content, antecedents, and consequences. Academy of Management Journal, 45, 779–790. Levinson, D. J., Darrow, C. N., Klein, E. B., Levinson, M. H., & McKee, B. (1978). Seasons of a man’s life. New York: Knopf. Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and task motivation. American Psychologist, 57, 705–717. London, M., & Smither, J. (1999). Career-related continuous learning: Defining the construct and mapping the process. In: G. R. Ferris (Ed.), Research in Personnel and Human Resources Management (Vol. 17, pp. 81–121). Stamford, CT: JAI Press. Lyness, K. S., & Thompson, D. E. (2000). Climbing the corporate ladder: Do female and male executives follow the same route? Journal of Applied Psychology, 85, 86–101. Maurer, T. J., & Tarulli, B. A. (1994). Investigation of perceived environment, perceived outcome, and person variables in relationship to voluntary development activity by employees. Journal of Applied Psychology, 79, 3–14. McCauley, C. D., & Hezlett, S. A. (2001). Individual development in the workplace. In: N. Anderson, D. S. Ones, H. K. Sinangil & C. Viswesvaran (Eds), Handbook of Industrial, Work, & Organizational Psychology (Vol. 2, pp. 313–335). London: Sage Publications. McCrae, R. R., & Costa, P. T. (1996). Toward a new generation of personality theories: Theoretical contexts for the five-factor model. In: J. S. Wiggins (Ed.), The Five Factor Model of Personality (pp. 51–87). New York: Guildford Press. McGuire, G. M. (1999). Do race and sex affect employees’ access to and help from mentors? Insights from the study of a large corporation. In: A. J. Murrell, F. J. Crosby & R. J. Ely (Eds), Mentoring Dilemmas: Developmental Relationships within Multicultural Organizations. Applied Social Research (pp. 105–120). Mahwah, NJ: Lawrence Erlbaum Associates. McManus, S. E., & Russell, J. E. A. (1997). New directions for mentoring research: An examination of related constructs. Journal of Vocational Behavior, 51, 145–161. Merriam Webster Dictionary (1995). Springfield, MA: Merriam Webster. Missirian, A. K. (1982). The corporate connection. Why executive women need mentors to reach the top. Englewood Cliffs, NJ: Prentice-Hall. Mobley, G. M., Jaret, C., Marsh, K., & Lim, Y. Y. (1994). Mentoring, job satisfaction, gender, and the legal profession. Sex Roles, 31, 79–98. Mullen, E. J. (1994). Framing the mentoring relationship as an information exchange. Human Resource Management Review, 4, 257–281. Mullen, E. J. (1998). Vocational and psychosocial mentoring functions: Identifying mentors who serve both. Human Resource Development Quarterly, 9, 319–331. Mullen, E. J., & Noe, R. A. (1999). The mentoring information exchange: When do mentors seek information from their prot´ g´ s? Journal of Organizational Behavior, 20, 233–242. e e

120

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Newby, T. J., & Heide, A. (1992). The value of mentoring. Performance Improvement Quarterly, 5(4), 2–15. Nielson, T. R., Carlson, D. S., & Lankau, M. J. (2001). The supportive mentor as a means of reducing work-family conflict. Journal of Vocational Behavior, 59, 364–381. Noe, R. A. (1988a). An investigation of the determinants of successful assigned mentoring relationships. Personnel Psychology, 41, 457–479. Noe, R. A. (1988b). Women and mentoring: A review and research agenda. Academy of Management Review, 13, 65–78. Noe, R. A., Greenberger, D. B., & Wang, S. (2002). Mentoring: What we know and where we might go. In: G. R. Ferris & J. J. Martocchio (Eds), Research in Personnel and Human Resources Management (Vol. 21, pp. 129–174). Oxford, England: Elsevier Science Ltd. Noe, R. A., & Wilk, S. L. (1993). Investigation of the factors that influence employees’ participation in developmental activities. Journal of Applied Psychology, 78, 291–302. Ochberg, R. L., Tischler, G. L., & Schulberg, H. C. (1986). Mentoring relationships in the careers of mental health administrators. Hospital and Community Psychiatry, 37, 939–941. Olian, J. D., Carroll, S. J., & Giannantonio, C. M. (1993). Mentor reactions to prot´ g´ s: An experiment e e with managers. Journal of Vocational Behavior, 43, 266–278. Olian, J. D., Carroll, S. J., Giannantonio, C. M., & Feren, D. B. (1988). What do prot´ g´ s look e e for in a mentor? Results of three experimental studies. Journal of Vocational Behavior, 33, 15–37. O’Neill, R. M. (2002). Gender and race in mentoring relationships: A review of the literature. In: D. Clutterbuck & B. R. Ragins (Eds), Mentoring and Diversity: An International Perspective (pp. 1–22). Woburn, MA: Butterworth Heineman. Orpen, C. (1995). The effects of mentoring on employees’ career success. Journal of Social Psychology, 135, 667–668. Orpen, C. (1997). The effects of formal mentoring on employee work motivation, organizational commitment and job performance. The Learning Organization, 4(2), 53–60. Ostroff, C., & Kozlowski, S. W. (1993). The role of mentoring in the information gathering processes of newcomers during early organizational socialization. Journal of Vocational Behavior, 42(2), 170–183. Peluchette, J. V. E., & Jeanquart, S. (2000). Professionals’ use of different mentor sources at various career stages: Implications for career success. Journal of Social Psychology, 140, 549–564. Phillips, L. (1978). Mentors and prot´ g´ s: A study of the career development of women managers e e and executives in business and industry (Doctoral dissertation, University of California at Los Angeles, 1977). Dissertation Abstracts International, 38(11), 6414A. Phillips-Jones, L. (1983, February). Establishing a formalized mentoring program. Training and Development Journal, 37, 38–42. Podolny, J. M., & Baron, J. N. (1997). Resources and relationships: Social networks and mobility in the workplace. American Sociological Review, 62, 673–693. Pollock, R. (1995). A test of conceptual models depicting the developmental course of informal mentorprot´ g´ relationships in the work place. Journal of Vocational Behavior, 46, 144–162. e e Prevosto, P. (2001). The effect of “mentored” relationships on satisfaction and intent to stay of companygrade U.S. Army reserve nurses. Military Medicine, 166, 21–26. Quinones, M. A. (1997). Contextual influences on training effectiveness. In: M. A. Quinones & A. Ehrenstein (Eds), Training for a Rapidly Changing Workplace (pp. 177–200). Washington, DC: American Psychological Association.

Mentoring Research

121

Ragins, B. R. (1989). Barriers to mentoring: The female manager’s dilemma. Human Relations, 42, 1–22. Ragins, B. R. (1997). Diversified mentoring relationships in organizations: A power perspective. Academy of Management Review, 22, 482–521. Ragins, B. R. (1999). Gender and mentoring relationships: A review and research agenda for the next decade. In: G. N. Powell (Ed.), Handbook of Gender and Work (pp. 347–370). Thousand Oaks, CA: Sage Publications. Ragins, B. R. (2002). Understanding diversified mentoring relationships: Definitions, challenges, and strategies. In: D. Clutterbuck & B. R. Ragins (Eds), Mentoring and Diversity: An International Perspective (pp. 23–53). Woburn, MA: Butterworth Heinemann. Ragins, B. R., & Cotton, J. L. (1991). Easier said than done: Gender differences in perceived barriers to gaining a mentor. Academy of Management Journal, 34, 939–951. Ragins, B. R., & Cotton, J. L. (1993). Gender and willingness to mentor in organizations. Journal of Management, 19, 97–111. Ragins, B. R., & Cotton, J. L. (1999). Mentor functions and outcomes: A comparison of men and women in formal and informal mentoring relationships. Journal of Applied Psychology, 84, 529–550. Ragins, B. R., Cotton, J. L., & Miller, J. S. (2000). Marginal mentoring: The effects of type of mentor, quality of relationship, and program design on work and career attitudes. Academy of Management Journal, 43, 1177–1194. Ragins, B. R., & McFarlin, D. B. (1990). Perceptions of mentor roles in cross-gender mentoring relationships. Journal of Vocational Behavior, 37, 321–339. Ragins, B. R., & Scandura, T. A. (1997). The way we were: Gender and the termination of mentoring relationships. Journal of Applied Psychology, 82, 945–953. Ragins, B. R., & Scandura, T. A. (1999). Burden or blessing? Expected costs and benefits of being a mentor. Journal of Organizational Behavior, 20, 493–509. Riley, S., & Wrench, D. (1985). Mentoring among women lawyers. Journal of Applied Social Psychology, 15, 374–386. Rotundo, D. M., & Perrew´ , P. L. (2000). Coping with a career plateau: An empirical examie nation of what works and what doesn’t. Journal of Applied Social Psychology, 30, 2622– 2646. Sackett, P. R., Gruys, M. L., & Ellingson, J. E. (1998). Ability-personality interactions when predicting job performance. Journal of Applied Psychology, 83, 545–556. Saks, A. M., & Ashforth, B. E. (1997). Organizational socialization: Making sense of the past and present as a prologue for the future. Journal of Vocational Behavior, 51, 234–279. Scandura, T. A. (1992). Mentorship and career mobility: An empirical investigation. Journal of Organizational Behavior, 13, 169–174. Scandura, T. A. (1997). Mentoring and organizational justice: An empirical investigation. Journal of Vocational Behavior, 51, 58–69. Scandura, T. A. (1998). Dysfunctional mentoring relationships and outcomes. Journal of Management, 24, 449–467. Scandura, T. A., & Ragins, B. R. (1993). The effects of sex and gender role orientation on mentorship in male-dominated occupations. Journal of Vocational Behavior, 43, 251–265. Scandura, T. A., & Schriesheim, C. A. (1994). Leader-member exchange and supervisor career mentoring as complementary constructs in leadership research. Academy of Management Journal, 37, 1588–1602.

122

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Scandura, T. A., & Viator, R. E. (1994). Mentoring in public accounting firms: An analysis of mentorprot´ g´ relationships, mentorship functions, and prot´ g´ turnover intentions. Accounting, e e e e Organizations, and Society, 19, 717–734. Scandura, T. A., & Williams, E. A. (2001). An investigation of the moderating effects of gender on the relationships between mentorship initiation and prot´ g´ perceptions of mentoring functions. e e Journal of Vocational Behavior, 59, 342–363. Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124, 262–274. Schneider, R. J., & Hough, L. M. (1995). Personality and industrial/organizational psychology. International Review of Industrial and Organizational Psychology, 10, 75–129. Seibert, S. (1999). The effectiveness of facilitated mentoring: A longitudinal quasi-experiment. Journal of Vocational Behavior, 54, 483–502. Seibert, S. E., Crant, J. M., & Kraimer, M. L. (1999). Proactive personality and career success. Journal of Applied Psychology, 84, 416–427. Seibert, S. E., Kraimer, M. L., & Crant, J. M. (2001). What do proactive people do? A longitudinal model linking proactive personality and career success. Personnel Psychology, 54, 845–874. Seibert, S. E., Kraimer, M. L., & Liden, R. C. (2001). A social capital theory of career success. Academy of Management Journal, 44, 219–237. Shapiro, E. C., Haseltine, F. P., & Rowe, M. P. (1978). Moving up: Role models, mentors, and the “Patron System”. Sloan Management Review, 19(3), 51–58. Silverhart, T. A. (1994, October). It works: Mentoring drives productivity higher. Managers Magazine, 69, 14–15. Single, P. B., & Muller, C. B. (2001). When email and mentoring unite: The implementation of a nationwide electronic mentoring program. In: J. J. Phillips & L. K. Stromei (Eds), Creating Mentoring and Coaching Programs (pp. 107–122). Alexandria, VA: ASTD. Sosik, J. J., & Godshalk, V. M. (2000a). Leadership styles, mentoring functions received, and jobrelated stress: A conceptual model and preliminary study. Journal of Organizational Behavior, 21, 365–390. Sosik, J. J., & Godshalk, V. M. (2000b). The role of gender in mentoring: Implications for diversified and homogenous mentoring relationships. Journal of Vocational Behavior, 57, 102–122. Sosik, J. J., & Godshalk, V. M. (in press). Self-other rating agreement in mentoring: Meeting prot´ g´ e e expectations for development and career advancement. Group & Organization Management. Steinberg, A. G., & Foley, D. M. (1999). Mentoring in the army: From buzzword to practice. Military Psychology, 11, 365–379. Struthers, N. J. (1995). Differences in mentoring: A function of gender or organizational rank? Journal of Social Behavior & Personality, 10, 265–272. Tepper, B. J. (1995). Upward maintenance tactics in supervisory mentoring and nonmentoring relationships. Academy of Management Journal, 38, 1191–1205. Tepper, B. J., Brown, S. J., & Hunt, M. D. (1993). Strength of subordinates’ upward influence tactics and gender congruency effects. Journal of Applied Social Psychology, 23, 1903–1919. Tepper, K., Shaffer, B. C., & Tepper, B. J. (1996). Latent structure of mentoring function scales. Educational and Psychological Measurement, 56, 848–857. Tharenou, P. (1997). Managerial career advancement. In: C. L. Cooper & I. T. Robertson (Eds), International Review of Industrial and Organizational Psychology (Vol. 12, pp. 39–93). Chichester, England: John Wiley & Sons.

Mentoring Research

123

Thibodeaux, H. F., III, & Lowe, R. H. (1996). Convergence of leader-member exchange and mentoring: An investigation of social influence patterns. Journal of Social Behavior & Personality, 11, 97–114. Thomas, D. A. (1990). The impact of race on managers’ experiences of developmental relationships (mentoring and sponsorship): An intra-organizational study. Journal of Organizational Behavior, 11, 479–492. Thomas, D. A. (1993). Racial dynamics in cross-race developmental relationships. Administrative Science Quarterly, 38, 169–194. Turban, D. B., & Dougherty, T. W. (1994). Role of prot´ g´ personality in receipt of mentoring and e e career success. Academy of Management Journal, 37, 688–702. Turban, D. B., Dougherty, T. W., & Lee, F. K. (2002). Gender, race and perceived similarity effects in developmental relationships: The moderating role of relationship duration. Journal of Vocational Behavior, 61, 240–262. Tyler, K. (1998, April). Mentoring programs link employees and experienced execs. HR Magazine, 43, 98–103. Viator, R. E. (1999). An analysis of formal mentoring programs and perceived barriers to obtaining a mentor at large public accounting firms. Accounting Horizons, 13(1), 37–53. Viator, R. E. (2001a). The association of formal and informal public accounting mentoring with role stress and related job outcomes. Accounting, Organizations and Society, 26, 73–93. Viator, R. E. (2001b). An examination of African Americans’ access to public accounting mentors: Perceived barriers and intentions to leave. Accounting, Organizations and Society, 26, 541–561. Viator, R. E., & Scandura, T. A. (1991). A study of mentor-prot´ g´ relationships in large public e e accounting firms. Accounting Horizons, 5(3), 20–30. Wall, J. A., & Callister, R. R. (1995). Conflict and its management. Journal of Management, 21, 515–558. Wallace, J. E. (2001). The benefits of mentoring for female lawyers. Journal of Vocational Behavior, 58, 366–391. Waters, L., McCabe, M., Kiellerup, D., & Kiellerup, S. (2002). The role of formal mentoring on business success and self-esteem in participants of a new business start-up program. Journal of Business and Psychology, 17, 107–121. Watson, E., & Clark, L. A. (1984). Negative affectivity: The disposition to experience aversive emotional states. Psychological Bulletin, 96, 465–490. Wayne, S. J., Liden, R. C., Kraimer, M. L., & Graf, I. K. (1999). The role of human capital, motivation and supervisor sponsorship in predicting career success. Journal of Organizational Behavior, 20, 577–595. Weil, P. A., & Kimball, P. A. (1996). Determinants of salaries earned: Comparisons of Black and White healthcare managers. Journal of Organizational Behavior, 17, 657–678. Whitely, W. T., & Coetsier, P. (1993). The relationship of career mentoring to early career outcomes. Organization Studies, 14, 419–441. Whitely, W., Dougherty, T. W., & Dreher, G. F. (1991). Relationship of career mentoring and socioeconomic origin to managers’ and professionals’ early career progress. Academy of Management Journal, 34, 331–351. Whitely, W., Dougherty, T. W., & Dreher, G. F. (1992). Correlates of career-oriented mentoring for early career managers and professionals. Journal of Organizational Behavior, 13, 141–154. Witt Smith, J., Smith, W. J., & Markham, S. E. (2000). Diversity issues in mentoring academic faculty. Journal of Career Development, 26, 251–262.

124

CONNIE R. WANBERG, ELIZABETH T. WELSH AND SARAH A. HEZLETT

Young, A. M., & Perrew´ , P. L. (2000). What did you expect? An examination of career-related support e and social support among mentors and proteges. Journal of Management, 26, 611–632. Zaheer, S., Albert, S., & Zaheer, A. (1999). Time scales and organizational theory. Academy of Management Review, 24, 725–741. Zey, M. G. (1984). The mentor connection. Homewood, IL: Dow Jones-Irwin. Zey, M. G. (1985). Mentor programs: Making the right moves. Personnel Journal, 64(2), 53–57.

THE IMPACT OF TELECOMMUTING DESIGN ON SOCIAL SYSTEMS, SELF-REGULATION, AND ROLE BOUNDARIES
David G. Allen, Robert W. Renn and Rodger W. Griffeth
ABSTRACT
As more companies and employees become involved in telecommuting, researchers and managers will need to understand the effects of this relatively new working arrangement on the work perceptions and behaviors of the individual telecommuter. The extant empirical literature provides mixed results and is limited by a lack of theory; consequently, neither researchers nor managers can rely on this literature for clear direction on how telecommuting will likely affect individual telecommuters. There is a critical need for theoretical frameworks to guide research on how telecommuting may affect the telecommuter’s job perceptions, working relations, and work outcomes. We present a multi-dimensional framework of telecommuting design, and focus on how telecommuting design may affect the telecommuter’s work environment and outcomes through its effects on the social system of the telecommuter, autonomy and self-management opportunities and requirements, and role boundaries, particularly in terms of the work and non-work interface. Our goal is to provide a framework to assist managers and researchers
Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 125–163 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22003-X

125

126

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

in systematically addressing questions of how to design telecommuting arrangements to maximize their potential benefits while minimizing their potential drawbacks.

INTRODUCTION
Enabled by the increasing power, availability, and affordability of telecommunications technology, many American companies (51%) have adopted telecommuting programs (McCune, 1998). Telecommuting represents an alternative work arrangement facilitated by such technology that allows the employee to work physically away from the employer’s premises (Feldman & Gainey, 1997; O’Mahony & Barley, 1999). Most estimates of the number of individuals who telecommute indicate that there are at least seven to 19 million telecommuters (Armstrong, 1993; Dunkin, 1995; Raghuram, Wiesenfeld & Garud, 2000). The International Teleworkers Association recently estimated that 16.5 million Americans telecommute at least one day per month (ITA, 2000). A recent survey by the Society for Human Resource Management found that 37% of 754 human resource professionals reported offering telecommuting in their organization, up from 20% in 1997 (SHRM, 2001). Further, telecommuters represent the fastest growing segment of home-based workers (Dunkin, 1995), and it is expected that more companies will embrace telecommuting in the future (Kirkman, Rosen, Gibson, Tesluk & McPherson, 2002; Maynard, 1994). As more companies and employees become involved in telecommuting, researchers and managers will need to understand the effects of this relatively new working arrangement on the work perceptions and behaviors of the individual telecommuter (Duxbury & Neufeld, 1999; O’Mahony & Barley, 1999). Bartol and Liu (2002) suggest that telecommuting presents a number of management challenges surrounding relationships with employees and employee adjustment. Unfortunately, theory and evidence on this important issue is weak. Scott and Timmerman (1999), for example, found only 32 empirical studies of telecommuting, and characterized them as lacking theory and as focusing on subjective estimates of outcomes (e.g. Teo, Lim & Wai, 1998). Bailey and Kurland (2002) recently found 80 published empirical studies. Only a subset of these examined telecommuter outcomes, whereas most focused on issues surrounding the decision whether to adopt telecommuting as a work arrangement. Their review also found that most studies of performance relied on telecommuter self-reports of performance levels. Lamond, Standen and Daniels (2000) found that many telecommuting studies were over a decade old, and excluded new technologies such as e-mail and groupware. Igbaria and Guimaraes (1999) argued

The Impact of Telecommuting Design

127

that there is no consensus regarding the impact of telecommuting on outcomes such as satisfaction and commitment, and characterized the literature as primarily anecdotal and speculative. Hill, Miller, Weiner and Colihan (1998) found that the limited empirical evidence on telecommuting revealed mixed and contradictory results regarding productivity, satisfaction, morale, and work-family balance. For example, it has been suggested that telecommuting increases work flexibility, provides uninterrupted time for cognitive tasks, and reduces distractions and commuting time (Goodrich, 1990; Ladda, 1995; Perlow, 1997). However, others report difficulty adjusting to the lack of structure, isolation, family interruptions, and stress that can accompany telecommuting (Chrobot, Janat & Gerstner, 1997; Raghuram et al., 2000). Furthermore, although telecommuting purportedly improves employee job satisfaction, organizational commitment, and job performance, and also purportedly lowers employee stress, work-family conflict, and turnover (AT&T, 1998), other evidence suggests that telecommuting may lower job and life satisfaction, and organizational commitment and lead to fewer interactions with others (Olson, 1989; Olszewski & Mokhtarian, 1994). The extant empirical literature, then, provides mixed results and is limited by a lack of theory (Clancy, 1994; Elder & Smith, 1999; Tomaskovic-Devey & Risman, 1993). Consequently, neither researchers nor managers can rely on this literature for clear direction on how telecommuting will likely affect individual telecommuters. There is a critical need for theoretical frameworks to guide research on how telecommuting may affect the individual telecommuter’s job perceptions, working relations, and work outcomes (Lamond et al., 2000; Scott & Timmerman, 1999; Sparrow & Daniels, 1999), and Roberts and Grabowski (1996) argue that existing technology frameworks are inadequate for studying new information technologies. Researchers have offered frameworks that offer insights into the impact of electronic communication technology on organizational form and structure, on the use of communication technologies, to explain the types of workers who will pursue telecommuting opportunities, or to link telecommuting directly to work-related outcomes, such as attachment, job satisfaction, withdrawal, and performance (e.g. Feldman & Gainey, 1997; Fulk & DeSanctis, 1995; Higo, Sheng, Shin & Figueredo, 2000; Wiesenfeld, Raghuram & Garud, 1998). They have not, however, explicated the underlying mechanisms by which telecommuting affects attitudinal and behavioral outcomes (Bailey & Kurland, 2002). Raghuram, Garud, Wiesenfeld and Gupta (2001) recently suggested that adjustment to virtual work is a function of both structural (e.g. work and evaluation characteristics) and relational (e.g. trust and connectedness) mechanisms, an important contribution. The present analysis goes further and focuses on how telecommuting may affect the telecommuter’s work environment and outcomes through its effects on the social system of the telecommuter, autonomy and self-management opportunities

128 DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Fig. 1. The Impact of Telecommunicating Design on Social Systems, Self-Regulation, and Role Boundaries.

The Impact of Telecommuting Design

129

and requirements, and role boundaries, particularly in terms of the work and non-work interface. Modeling the effects of telecommuting design on outcomes as mediated by telecommuters’ appraisal of the working arrangement overcomes criticisms that predictions of the effects of telecommunications technology neglect intervening variables (O’Mahony & Barley, 1999). Toward this end, we propose the framework depicted in Fig. 1. Note that there are four major groups of variables in the Fig. 1 model. The first component explicates the multidimensional nature of telecommuting design. Merely classifying an individual as a telecommuter or non-telecommuter does not provide enough information to allow understanding of the likely effects of such an arrangement. The second component consists of variables we have labeled mediating mechanisms, and the third consists of variables we have labeled outcomes. Although telecommuting has been proposed to affect both attitudinal and behavioral outcomes, this model tries to more completely explain how and why telecommuting is likely to influence outcomes such as satisfaction and performance. Our focus is not on the relationships among the mediating mechanisms and outcomes, since those relationships have received substantial research attention and are not endemic to the telecommuting context. Instead, we focus on relationships among telecommuting design and the mediators, since we believe telecommuting will have more direct effects here than on more distal outcomes such as performance and satisfaction. At this point, we are not arguing that these relationships are necessarily fully mediated ones; however, we believe this framework provides greater explanatory potential for understanding the processes by which telecommuting may affect important outcomes. Finally, the fourth component consists of variables proposed to moderate relationships in the model. Again, we focus on variables that may influence relationships between telecommuting design and the mediating mechanisms because of our interest in the effects of telecommuting.

BOUNDARY CONDITIONS
Before addressing specifics of the model, we clarify that the model is explicitly intended to address telecommuting effects on individual telecommuters. It is not intended to address reasons why organizations and individuals decide to adopt telecommuting or make decisions concerning the availability and structure of telecommuting arrangements. Similarly, the model does not address issues of physically distributed teamwork, work group or organization level performance, or cost effectiveness. Finally, the model does not attempt to explain telecommuting effects on managers or co-workers, such as the effects on co-workers who are

130

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

unable or not allowed to telecommute, or the effects on supervisors of managing someone who is not physically present. These issues are important, but have begun to be addressed elsewhere (see Bailey & Kurland, 2002, for a review), and are beyond our scope of interest in this case. Although narrowing the scope of the model necessarily limits the range of phenomena explained, what remains is critically important. For researchers and organizations to understand the effects of allowing and even encouraging and requiring some employees to work outside of the traditional office environment, a systematic framework is needed for explicitly explaining the mechanisms by which telecommuting can influence individual work-related attitudes and behaviors.

CONCEPTUALIZING TELECOMMUTING
It is important to differentiate telecommuting from other work arrangements in which individuals work away from a traditional office setting. Workers who run businesses from their home or who happen to live at their work site (e.g. farmers) are not considered telecommuters because there is no commute trip to eliminate, reduce, or shift. Also, although the terms telecommuting and telework are often used interchangeably, telework is defined as using telecommunications technology to conduct business at a distance; thus, it includes using faxes, e-mail, cellular phones, even ordinary phone calls from any location, such as from the car on the way to work. Telecommuting is a subset of telework that involves using telecommunications technology to work outside the office in lieu of traveling to the traditional office and to allow greater flexibility to individual employees. Current confusion about telecommuting’s individual-level effects may partially result from failing to distinguish telecommuters from other types of teleworkers and home-based workers. Confusion may also stem from unidimensional rather than multidimensional conceptualizations of this construct (e.g. Elder & Smith, 1999). Early conceptualizations viewed telecommuters as workers who worked almost entirely with computers, who telecommuted full-time, and who worked from home (Handy & Mokhtarian, 1995). However, this picture fails to capture the full complexity of telecommuting. An in-depth review of the available literature and analysis of the work practices of several large companies with extensive telecommuting experience reveals that telecommuting designs in practice differ along a number of important dimensions. For example, only a small subset of telecommuters do so full-time; most do so part-time. And, many telecommuters work part of the day in the office and the remainder out of the office. Such telecommuters do not eliminate commute trips; they shift them to off-peak hours. In some cases,

The Impact of Telecommuting Design

131

telecommuters’ work does not require use of a computer. Further, telecommuting need not occur only at home. Telecenters, where equipment and supplies are located in small satellite offices, are a growing telecommuting option. In this case, commutes are reduced, though not eliminated. It would be misleading to assume that a data entry employee required to work full-time at a satellite center under close electronic monitoring in order to reduce real estate costs for the organization is the same as a manager allowed to telecommute from any location at their discretion in order to provide increased flexibility to manage work and non-work demands.

Telecommuting Design Dimensions Unlike unidimensional views, Feldman and Gainey (1997) propose a multidimensional conceptualization of telecommuting arrangements. They suggest that telecommuting arrangements consist of four dimensions: (1) whether the amount of time spent telecommuting is full- or part-time; (2) whether the telecommuting schedule is fixed or flexible; (3) whether telecommuting occurs from home or from a satellite work site; and (4) whether telecommuting was initiated at the request of the organization or the individual. These dimensions are a valuable first step, but expressing them as dichotomous propositions limits their utility. For example, there might be important differences between individuals who telecommute very little and those who do so most but not all of their work time. Similarly, even among telecommuters classified as flexible, there may be important differences between those allowed to choose among certain flexible hours to telecommute and those able to telecommute whenever or wherever they like. We believe it is critical to consider variance in telecommuting arrangements, because most of the empirical research on telecommuting merely compares telecommuters against non-telecommuters without making finer distinctions (Bailey & Kurland, 2002). Because of these potential limitations, we draw from the wider telecommuting literature as well as analysis of the telecommuting arrangements of several large organizations and refine and expand our conceptualization of telecommuting along six design facets: (1) telecommuting frequency; (2) telecommuting location; (3) flexibility of scheduling; (4) formalization of policies; (5) extent and nature of performance monitoring; and (6) initiation of the telecommuting arrangement. We believe these aspects of telecommuting design describe many of the important decisions organizations make when utilizing telecommuting arrangements, and also enable a more fine-grained analysis of the outcomes associated with telecommuting.

132

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Telecommuting Frequency Few individuals are full-time telecommuters. Instead, the amount of time spent telecommuting may range from as little as a few hours per week to full-time telecommuters who rarely, if ever, visit the office. Individuals who work a great deal out of the office are likely to have different experiences than those who do so only a few hours. Feldman and Gainey (1997), for example, suggest full-time telecommuters would have more autonomy and less interdependence and interaction. There is some preliminary evidence to suggest that greater time at the office increases communication satisfaction for telecommuters (Sturgill, 1998). Researchers, however, have often ignored frequency, commonly comparing a telecommuting sample against a non-telecommuting sample (Bailey & Kurland, 2002). Measuring the amount of time an individual telecommutes relative to their total amount of work time would enable more detailed analysis of the effects of working away from the office. Telecommuting Location Telecommuters may work part of the time in the office, part at home, part from alternate work sites such as telecenters, or any combination of locations. Kurland and Bailey (1999) offer four types of telecommuting: home-based; satellite offices, where an employer offers work space closer to locations where employees live; neighborhood work centers, where multiple employers share work space for telecommuters; and mobile teleworkers, who work from a variety of locations, perhaps while traveling. Although empirical evidence is lacking, it seems likely that telecommuting from home would differ from doing so from a telecenter or satellite office. Depending on its location, telecommuting from a satellite office should still shift, reduce, or eliminate stressful commutes, and may also avoid some types of conflicts and distractions possible when working at home, such as dependent children, household chores, and lack of separation between work and home. For such reasons, Shamir and Salomon (1985) suggest that “halfway” arrangements like satellite centers may be ideal options. On the other hand, telecommuting from a satellite center may not provide as much flexibility or increased ability to balance work and family as telecommuting from home. If experiences differ across work sites, whether home, telecenters, or other options, then the relative amount of time spent at specific locations should be important in understanding the effects of telecommuting. Flexibility of Scheduling Telecommuters may differ in terms of the flexibility of their schedules, or the control they have over when and from where they will work. Some telecommuters are required to work certain hours at the office and certain hours from another

The Impact of Telecommuting Design

133

location, such as home. Others may have core hours during which they must be in the office or must be at home. However, during other times, the employee decides from where to work. Still others have complete flexibility in scheduling the timing and location of work. Telecommuters with greater scheduling flexibility seem likely to experience more control and autonomy, and may also perceive that they are better able to balance work and family demands (Feldman & Gainey, 1997). More flexible telecommuters may also perceive differences in social cohesiveness and interaction. Formalization of Policies There are a host of potential policy issues associated with a telecommuting arrangement, such as communication guidelines, equipment ownership, frequency, and location. Yet, there is evidence that organizations vary in the extent to which they provide formal policies regulating the arrangement. One survey indicates that 80% of respondent companies lacked formal telecommuting policies, while others provided extensive documentation of specific policies (Farrah & Dagen, 1993). A more recent survey reported that 24% of respondent companies that offer flexible work arrangement such as telecommuting provide guidelines, down from 26% in 1998 (Wells, 2001). Thus, the existence, detail, and rigidity of such policies are likely to vary, and there is a need for research on the effects of telecommuting policy formalization (Glass & Finley, 2002). Bartol and Liu (2002) suggest that policies will impact relationships with employees and thus outcomes such as attitudes and performance. More formal policies may mitigate ambiguity or uncertainty about role expectations. On the other hand, they may also reduce perceived flexibility or autonomy. Performance Monitoring Management of telecommuters also varies in the extent of performance monitoring that occurs. Concerns about managing individuals who are not physically present and the possibility of shirking are often cited as barriers to telecommuting arrangements. Thus, the evaluation of telecommuter performance is an important issue. Computer-based or electronic performance monitoring allows managers to monitor aspects of telecommuter behavior such as key strokes, telephone calls, log-in times, etc. For telecommuters in professional jobs, the popular press suggests shifting to managing based on objectives and results; however, for clerical, data entry, and other non-professional telecommuting jobs, an alternative is to closely monitor the amount of time spent at the computer or logged into a system, even to the extent of monitoring keystrokes. It seems likely that different amounts of electronic monitoring will be associated with different experiences in terms of flexibility and control. There is some evidence that constant monitoring can lead

134

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

to stress, apprehension, low morale, and less frequent and less rich communication (Fairweather, 1999). Telecommuting Initiation The assumption is often made that telecommuting is a boon to the telecommuter. However, the literature shows that there are clear differences in the reasons why organizations allow or require telecommuting, and these reasons may affect the ways in which telecommuting arrangements are implemented. Telecommuters who make more positive attributions about these motivations are likely to have different experiences than those who make less positive ones. Telecommuting arrangements may be initiated to save overhead costs, for legal compliance, to address environmental concerns, to increase productivity, to allow greater staffing flexibility, to allow employees greater flexibility, to help employees meet non-work responsibilities, and so forth. Feldman and Gainey (1997) suggest that individuals who initiate telecommuting would experience more positive outcomes than those who do so at the organization’s initiative. However, it is not clear that individuals offered the opportunity to telecommute by organizations should necessarily react negatively. Perhaps a more accurate representation of the underlying construct in this case is attributions about the motivation behind adopting telecommuting. These underlying adoption motivations may vary in the extent to which they are intended to benefit the telecommuter versus the organization, and attributions about these differences may result in different experiences and outcomes. Although telecommuting may be intended to benefit both employee and company, the experiences of a lower-level data entry worker required to begin telecommuting to save office space are likely to be quite different from those of a manager who is allowed to telecommute to help accommodate family responsibilities. Underlying adoption motivation more explicitly intended to benefit the telecommuter should result in more positive experiences and outcomes.

THE IMPACT OF TELECOMMUTING DESIGN
It may be tempting to view telecommuting as a simple change in the location from which an individual works. However, this analysis is based on the premise that telecommuting fundamentally alters the ways people work, relationships with peers, supervisors, and organizations, and the boundaries between work and non-work life. Broadly speaking, we propose that telecommuting design facets influence individual-level work-related outcomes through three categories of mechanisms: (1) the social system of the telecommuter, particularly in terms

The Impact of Telecommuting Design

135

of relationships with others and the telecommuter’s own social identity; (2) self-regulatory processes, particularly in terms of autonomy, personal control, self-management, and the feedback environment; and (3) role boundaries, particularly in terms of the work and non-work interface. We focus on these mediating mechanisms for several reasons. First, telecommuting is a technological change whose effects cannot be understood without considering the social system within which it is embedded (Passmore & Sherwood, 1978). The effects of changes in technical systems should not be considered in isolation from the effects of these changes on social systems (e.g. Emery & Trist, 1960; Trist & Bamforth, 1954). This reasoning applies to telecommuting in particular because to varying degrees it physically removes the individual from the social context, thus potentially drastically affecting relationships with supervisors and co-workers, as well as with family and friends, and also the social identity of the telecommuter. Second, telecommuting is by definition both a technological change and an alternate way of designing work and jobs. Although the telecommuter’s general responsibilities and tasks may remain fundamentally similar, the work environment and the ways in which they accomplish work may change, perhaps drastically, as a function of telecommuting. We argue that one of the most important and direct ways that telecommuting affects work design is by providing greater opportunities for autonomy and self-management while simultaneously requiring greater personal control and self-management effort as telecommuters adapt to the option of working outside the traditional office and away from personal supervision. Telecommuting also changes the nature of the feedback environment in which individuals work. Third, telecommuting affects boundaries among roles, particularly with regard to work and family (Ashforth, Kreiner & Fugate, 2000). Indeed, telecommuting’s importance is largely attributed to helping individuals balance work and family responsibilities, although some research identifies telecommuting as a trend that may actually increase the potential for work-family interface and conflict (Eagle, Miles & Icenogle, 1997). The unique nature of telecommuting is such that it would seem to simultaneously provide greater control over the extent to which work and family concerns “spillover” into each other while also increasing the extent to which these two spheres actually overlap (cf. Zedeck, 1992). Finally, we argue that considering telecommuting’s effects on these mechanisms in terms of the design facets describe earlier will aid researchers in answering questions about the effect of telecommuting on outcomes such as attitudes and performance. It is likely that telecommuting could have positive or negative effects on job satisfaction, organizational commitment, and job performance depending on the implementation and the circumstances. Thus, our goal is to

136

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

provide a framework to begin to guide research seeking to clarify how, why, and when telecommuting is likely to influence these important outcomes.

Social Inclusion and Relationships A widely expressed concern of employees regarding telecommuting is that they will feel isolated from the workplace and from social contacts at work (Handy & Mokhtarian, 1995). Rousseau (1989) suggests that telecommuters might feel less inclusion in the workplace, with inclusion defined as the degree to which the individual is physically and psychologically integrated into the organization. Lack of inclusion can affect employee investment of time and energy, willingness to make long-term commitments, and how role competition is resolved (Rousseau, 1989). The isolated employee is likely to miss out on the informational benefits of social contact, and to be less likely to take on exceptional levels of responsibility (Pratt, 1984). Because telecommuting physically removes the telecommuter from the workplace, it likely affects the telecommuter’s social context in several ways. One is through the extent to which the individual identifies with the organization. Another is through changes in perceptions of social support the individual receives from supervisors, co-workers, and the organization. Social Identity Social identity theory (SIT) asserts that individuals identify who they are through the groups to which they belong, and that individuals infuse their group memberships with special meanings. Further, SIT holds that social identity processes are motivated by the enhancement of self-esteem and the reduction of uncertainty that follows from belonging to distinctive groups. According to Hogg and Terry (2000), social identity processes stem from individuals’ needs to “reduce subjective uncertainty about one’s perceptions, attitudes, feelings, and behavior and, ultimately one’s self-concept and place within the social world” (p. 124). Ashforth and Kreiner (1999) also state “through social interaction and the internalization of collective values, meanings, and standards, individuals come to see themselves through the eyes of others and construct more or less stable self-definitions and a sense of self-esteem” (p. 417). As the degree of separation between a telecommuter and his/her company increases, telecommuters will likely spend less time with their work groups and have less exposure to organizational values, meanings and standards. Less exposure to co-workers and organization culture may cause telecommuters to lose sight of their position in work groups, departments, and the larger social domain of the organization (Hogg & Mullin, 1999). This would be expected to

The Impact of Telecommuting Design

137

negatively affect telecommuters’ self-esteem and produce greater uncertainty about self-identity, psychological processes hypothesized to motivate social identity processes or, more specifically, self-categorization processes whereby individuals seek new groups with whom to identify, to assimilate themselves, and to derive self-esteem (Hogg & Terry, 2000). Similarly, telecommuters may find it more difficult to identify with their organization because of their lack of access to tokens and symbols of organizational membership and to organizational rituals and ceremonies (Wiesenfeld, Raghuram & Garud, 2001). Vega and Brennan (2000) recently argued that telecommuting should increase feelings of isolation because telecommuters lack artifacts of status, have greater difficulty in identifying with group norms and values, and have diminished opportunities for shared experiences. Thus, we can make several propositions regarding the impact of telcommuting design on identification with the organization. The more employees are away from the office and physically separated from co-workers and the organizational environment, the more difficult it may be to develop or maintain strong identification with the organization. Therefore, telecommuting frequency will be negatively related to identification with the organization. At the same time, though, different telecommuting locations provide varying amounts of exposure to the organization. Telecommuting from home or from neighborhood locations inhabited by multiple employers may provide less exposure to the organization than a satellite center that provides more organizational culture and contact with organization members, and so may make identification more difficult. These effects are important because loss of social identity from separation may affect organizational commitment, an attitude or orientation that attaches the identity of a person to an organization (Sheldon, 1971). As opposed to an employee who works on company premises, a telecommuter’s ability to identify with the organization may be weakened by the physical and psychological barriers that separate the individual from the organization and its groups. Benkhoff (1997) suggests that the absence of clear expectations resulting from exposure to a firm’s social network may disrupt the social identity process (i.e. defining and identifying oneself as a member of a given group or employer). In other words, due to lack of exposure to an organization’s context and its groups, telecommuters may become victims of a blurred social identity with specific groups within the organization and, thus, struggle to maintain organizational commitment. Therefore, greater percentage of time telecommuting, especially from home, may be associated with less commitment to the organization as a result of a loss of meaningfulness and sense of connection with the organization (Albert, Asforth & Dutton, 2000; Mowday, Porter & Steers, 1982). Telecommuting designs that increase physical and psychological separation from the organization may be associated with

138

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

greater subjective uncertainty and reduced organization-based self-esteem that motivate social identity processes. These processes are then associated with a loss of connection with the organization manifested in less identification with the organization and lower organizational commitment. Social Support Social isolation is a major factor inhibiting the success of telecommuting (Chapman, Sheehy, Heywood, Dooley & Collins, 1995). Because telecommuters spend less time in close physical proximity to others at work, they experience less interaction and potentially less rich communication (e.g. e-mail versus face-to-face). Cooper and Kurland (2002) recently argued that telecommuters are likely susceptible to greater social and professional isolation. Feldman and Gainey (1997) argued that telecommuting would decrease interaction with supervisors and co-workers, and Lowery (1996) found some evidence that employees who were physically distant from their supervisors perceived less communication with them. Communication that does occur may be delivered via leaner communication media, providing less context and information. In addition, research on the spatial configuration of work space indicates that individuals whose work setting has fewer barriers to interaction will have more expected and unexpected interactions (e.g. Oldham, Cummings & Zhou, 1995). The physical separation of telecommuters would be expected to act as a barrier to interaction, thus reducing the amount of potential social support from supervisors and co-workers. The frequency and location of telecommuting are likely to affect social support. In interviews with telecommuters, their supervisors, and non-telecommuters, Cooper and Kurland (2002) found that all groups agreed that lower telecommuting frequency would be associated with less isolation. Individuals who work outside the traditional office more often are likely to feel less social support from both supervisors and co-workers. Further, though, individuals who telecommute from a satellite center are likely to have more interaction with co-workers and perhaps experience greater inclusion in workplace norms and culture than individuals who telecommute from home. There is some evidence that extensive computer and Internet use may displace social activities and the maintenance of strong social ties (Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay & Scherlis, 1998). Similar effects may also occur from extensive telecommuting. Cooper and Kurland’s (2002) interviews also revealed that formalized policies were thought to potentially protect against isolation, however. The underlying adoption motivation may also influence perceptions of support, particularly supervisor support. Individuals who perceive that the telecommuting arrangement was implemented primarily or exclusively to benefit the organization may perceive even lower supervisory support. For example, if the employee

The Impact of Telecommuting Design

139

believes they are being required to telecommute to free up office space, they are likely to perceive their supervisors as less supportive. This may also be the case if the organization utilizes extensive close performance monitoring. On the other hand, individuals who perceive that the telecommuting arrangement was implemented primarily or exclusively to benefit them are likely to perceive greater supervisory support. Thus, telecommuting frequency, telecommuting from locations not associated with the organization (home or neighborhood locations), and extensive performance monitoring should be negatively related to perceptions of supervisor and co-worker support. Formalization of policies and the extent to which the telecommuting adoption motivation is perceived as intended to benefit the telecommuter should be positively related to perceptions of supervisor support. These relationships are important because social support from supervisors and co-workers can influence work-related attitudes such as satisfaction and commitment, and may also impact job performance because of their effects on the employee’s ability and willingness to cope with and overcome obstacles.

Perceived Organizational Support Rooted in social exchange theory (e.g. Gouldner, 1960), perceived organizational support (POS) represents employees’ “global beliefs about the extent to which the organization values their contributions and cares about their well-being” (Eisenberger, Huntington, Hutchinson & Sowa, 1986, p. 501). Shore and Shore (1995) posit that discretionary human resource practices that communicate concern for employees’ future should enhance POS. Thus, the adoption motivation underlying the decision to telecommute is likely related to perceptions of support, especially given the emphasis on employer discretion in developing POS (Shore & Shore, 1995). That is, telecommuters’ POS may vary with their perceptions of the forces motivating organizations to adopt this new work arrangement. Employees who perceive the driving forces behind telecommuting as primarily serving the organization (i.e. lower overhead costs, legal compliance, etc.) may interpret telecommuting as a sign that the organization does not care as much about their well-being as when the organization provided them a pleasant work space. In addition, employees who are forced to telecommute may interpret the decision as an indication that the organization doesn’t care about their future careers with the organization. The negative impact that telecommuting may have on career advancement is a primary concern of telecommuters (Perlow, 1997). On the other hand, employees who believe that the forces motivating telecommuting stem from the desire to help employees avoid stressful commutes, provide greater

140

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

schedule flexibility and allow more time to meet non-work responsibilities may surmise that telecommuting is an action by an organization that cares about their well-being. At the same time, telecommuters who are given a great deal of discretion in managing their work processes will likely perceive that the organization both trusts them and cares about them. Allowing telecommuters to design their own schedules and not shackling them with extremely restrictive rules and monitoring should increase perceptions of organizational support. Extremely limited flexibility or restrictive rules, policies, and monitoring, on the other hand, might reduce these perceptions. Thus, flexibility of scheduling and the extent to which the telecommuting adoption motivation is perceived as intended to benefit the telecommuter should be positively related to perceived organizational support. Formalization of policies and extensive performance monitoring should be negatively related to perceived organizational support. These relationships are important because of the potential effects on organizational commitment and other outcomes through perceived organizational support (POS). When employees perceive high levels of support, they feel obligated to reciprocate with commitment to their organization and by engaging in behaviors that support organizational goals (Wayne, Shore & Liden, 1997). Research indicates that POS is positively related to employee commitment to an organization and organizational citizenship behavior and negatively related to intentions to quit and turnover (Allen, Shore & Griffeth, 2003; Eisenberger et al., 1986; Moorman, Blakely & Niehoff, 1998; Shore & Tetrick, 1991; Wayne, Shore & Liden, 1997).

Self-Regulation Opportunities and Requirements Telecommuting purportedly provides enhanced freedom and flexibility because of the ability to eliminate or shift commutes and to determine the scheduling of when and from where to work. Telecommuters may also experience less direct supervision and monitoring from supervisors and peers because of their physical separation from the workplace. Telecommuting, then, may increase perceptions of autonomy and control. At the same time, it places additional responsibility on the telecommuter to manage their own behavior. Thus, telecommuting increases both the opportunities and the requirements for self-regulating behavior. Several aspects of telecommuting design are intended to influence how empowered employees feel to control their work life. This work arrangement also has implications for how employees self-manage their work behavior. Telecommuting also affects the

The Impact of Telecommuting Design

141

feedback environment in which the employee works, an important element of self-regulation. Autonomy and Personal Control Technological change that increases worker discretion over the pace and flow of work and that allows more control over establishing priorities, time management, and work methods increase autonomy (Rousseau, 1977; Slocum & Sims, 1980). A primary component of empowerment is self-determination to plan and schedule work and identify and remove obstacles (Spreitzer, 1995; Thomas & Velthouse, 1990). Because telecommuters’ work behaviors and methods are less likely to be closely monitored by supervisors or peers, they should perceive more discretion over their workflow and control over how they manage their work than in a traditional office setting. Therefore, increased percentage of work time spent telecommuting should enhance autonomy perceptions. Similarly, whether work occurs at home, a satellite office, or a corporate office should also affect perceptions of the amount of control one has over his or her work. Researchers suggest that working at home, and telecommuting in particular, will increase autonomy because of less direct supervision (Feldman & Gainey, 1997; Parker & Wall, 1998; Shamir & Salomon, 1985). A telecommuter who works from home should experience greater discretion and control than one who works from a satellite office, who should experience more discretion and control than one who works out of a corporate office. Thus, telecommuting from home should enhance autonomy perceptions more than telecommuting from a satellite office. Telecommuters also differ in their degree of flexibility in deciding when and from where they will telecommute. Hackman, Oldham, Janson and Purdy (1975) propose that flexibility of scheduling work is an important aspect of vertical loading that would be positively related to this job dimension. Individuals with greater scheduling flexibility should perceive greater discretion and control over their work. Thus, the degree to which telecommuting schedules are flexible should also be positively related to autonomy perceptions. Finally, formalization of telecommuting policies and extensive electronic performance monitoring may also be related to autonomy, although in a different manner. As rules, policies, and procedures for telecommuting increase, a telecommuter’s perceptions of control over work may decrease, as such procedures constrain freedom, independence, and discretion. In this regard, Parker and Wall (1998) suggest that telecommuting might not increase autonomy in the case of electronic performance monitoring or tightly controlled deadlines. Thus, increased monitoring and formalization of telecommuting policies should reduce perceptions of autonomy.

142

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

To understand the relationship between telecommuting and outcomes such as job performance and job satisfaction, it is useful to take into account direct effects on autonomy and personal control perceptions. Having control over work methods and greater influence over work-related decisions enhances feelings of personal responsibility and accountability that, in turn, lead to improvements in job performance and job satisfaction (Hackman & Oldham, 1980). Over two decades of research indicate that job autonomy is positively associated with job performance and satisfaction (Fried & Ferris, 1987; Loher, Noe, Moeller & Fitzgerald, 1985). A meta-analysis of perceived control, autonomy, and participation found consistent evidence of positive relationships with work satisfaction, job performance, and reduced stress (Spector, 1986). According to social cognitive theory, personal control perceptions can enhance performance through self-efficacy judgments (Bandura, 1986; Greenberger & Strasser, 1986). Self-efficacy refers to the beliefs an individual has over his her ability to perform a specific task. When individuals believe they can control factors that affect their work outcomes, they tend to have higher self-efficacy and set higher personal goals (Bandura & Wood, 1989). By contrast, when individuals perceive an inability to exercise control over the environment (i.e. lack autonomy), they experience a variety of adverse outcomes, including lower self-efficacy, performance decrements, and decreased satisfaction (Bazerman, 1982; Gist & Mitchell, 1992). Two longitudinal field studies indicate that personal control perceptions are positively related to job performance and job satisfaction (Greenberger, Strasser, Cummings & Dunham, 1989). Thus, telecommuting arrangements that enhance perceptions of job autonomy and personal control would be expected to enhance telecommuters’ experienced responsibility, accountability, and self-efficacy judgments and, hence, establish conditions for improvements in job performance and job satisfaction. Self-Leadership Opportunities Self-leadership is a process of using learning principles to manage one’s own behavior for the purpose of achieving desired results (Manz & Neck, 1999). It involves three general activities: (1) observing one’s own behavior to identify performance objectives; (2) establishing personal goals and plans to attain desired performance objectives; and (3) monitoring and reacting to one’s progress in working towards personal objectives (Bandura, 1991). Considering the impact of telecommuting on self-leadership opportunities is important because studies suggest that how well telecommuters adapt to internal versus external control will likely affect their job performance and job satisfaction (Frayne & Geringer, 2000; Uhl-Bien & Graen, 1998).

The Impact of Telecommuting Design

143

Broadly speaking, telecommuting’s effect on self-leadership opportunities is based on the principle that psychosocial functioning occurs within a triadic reciprocal causal framework consisting of behavior, cognitive and personal factors and the external environment (Wood & Bandura, 1989). In essence, this framework suggests that behavior is the result of interactions between the person and the external environment. As Wood and Bandura (1989) state, “individuals are both products and producers of their environment” (p. 362). Thus, self-leadership is not only a product of an individual influencing the external environment to suit his or her self-regulation needs but also is a result of the external environment exerting influence over an individual’s perceptions and attempts at self-control (Zeidner, Boekaerts & Pintrich, 2000). In this vein, telecommuting changes the environment in which work occurs and thus is likely to affect employees’ perceptions of self-leadership opportunities. Telecommuting affects self-leadership opportunities by placing employees in work environments that provide them with autonomy and that enhance their beliefs that they control their own work processes and outcomes. The job characteristics model (JCM) proposes that work designs that increase employees’ control over their own work (i.e. autonomy) cause employees to experience more personal responsibility for their work processes and outcomes that, in turn, improve intrinsic motivation, job satisfaction and job performance (Hackman & Oldham, 1980). Experiencing greater personal responsibility for one’s own work processes and outcomes is also a core psychological state of self-leadership (Manz & Neck, 1999; Renn & Vandenberg, 1995). Self-leadership shifts responsibility for influencing individuals to achieve organizational goals from an external agent to the individual employee (Manz & Neck, 1999). And, as individuals realize that they control critical work activities once performed by a manager, they experience feelings of heightened personal responsibility for the quantity and quality of their work. In addition, social cognitive theory suggests that work designs that increase autonomy or control enhances individuals’ beliefs that they can organize and execute the courses of action necessary to accomplish desired results (i.e. self-efficacy). Further, as individuals’ efficacy beliefs rise so does their self-regulatory efforts (i.e. self-leadership efforts) (Bandura, 1997). Thus, the same telecommuting dimensions that enhance autonomy and personal control perceptions ought to promote perceptions of self-leadership opportunities as well. To be more specific, telecommuting removes individuals from organizational settings comprised of structural properties that can restrict self-management opportunities (Mills, 1983). In the broadest sense, telecommuting places individuals in a more organic context than does working in a traditional work environment. In a traditional work environment, employees conduct work within

144

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

a network of physical and psychosocial structures that can frustrate self-regulation efforts. Physical structures such as tools and equipment, material and supplies, technologies, job-related information, departmental designs, procedures and policies, decision hierarchies, work designs, and even the architectural layout of a building can constrain employees’ ability to control their own work processes and outcomes (Peters & O’Connor, 1980). As one example, employees’ work within organizations may be more tightly coupled with and thus more dependent upon the work and/or resources of other individuals, work groups and departments than when it is performed in a telecommuting arrangement (Thompson, 1967; Tsui & Ashford, 1994). Consequently, employees within traditional organizations have comparatively less freedom than most telecommuters who work at home or satellite offices to decide when, where and how to work. Psychosocial properties of organizations can also interfere with self-leadership efforts. Formal leaders, such as first-line supervisors, whose main purpose is to control and monitor employee work can signal employees that they are not ultimately responsible for managing their own work processes and outcomes (Manz & Sims, 1987; Niehoff & Moorman, 1993). When employees believe that supervisors control and monitor work processes and outcomes, they are likely to experience personal inefficacy and are less likely to exercise self-regulation (Wood & Bandura, 1989). Besides formal leaders, the climate, roles, norms and social information processes within traditional work settings may affect employees’ self-leadership beliefs and actions (Feldman, 1984; Salancik & Pfeffer, 1978; Schneider, 1990). Role expectations and information acquired from the social environment, for instance, may “communicate” to employees that they are not supposed to take an active posture towards their work. Such expectations and social information would likely reduce employees’ perceived responsibility for work outcomes, lower their self-efficacy beliefs and thus dampen their self-regulatory activities (Bandura, 1997; Griffin, 1983). In contrast, telecommuting places relatively fewer physical and psychosocial constraints on employees and even allows them to design some or all aspects of their own work surroundings. Instead of waiting to perform a task until another individual, group or department provides needed resources (e.g. money or information), telecommuters may be provided with their own resources (e.g. travel budget or supply budget) and empowered to acquire the resources when they need it and from whomever can provide resources the most quickly and economically. Telecommuters also perform work in a context where they are not surrounded by a climate, role expectations, norms, leaders and co-workers who could hinder self-control attempts at work. The most obvious example is the absence of direct supervision that, in turn, reduces the frequency of direct external control and monitoring of work. As a result of minimal physical and

The Impact of Telecommuting Design

145

psychosocial constraints, telecommuters should experience a heightened sense of personal responsibility, self-efficacy and thus self-leadership opportunities. Thus, frequency of telecommuting, telecommuting from locations outside of the organization, and flexibility of scheduling should be positively associated with self-management opportunities. Formalization of policies and performance monitoring should be negatively related to self-management opportunities. Performance Feedback In work organizations, performance feedback is information about the correctness or adequacy of one’s work behavior (Ilgen, Fisher & Taylor, 1979). Understanding telecommuting’s effects on performance feedback is worthy of attention because of its relationship with important work attitudes and behaviors. Individuals need information about the quality of their work behavior to modify personal goals, work strategies and improve task performance (Locke & Latham, 1990; Taylor, Fisher & Ilgen, 1984). Indeed a recent meta-analysis indicates that feedback interventions are positively associated with task performance as long as the feedback directs attention to the task and not the self (Kluger & DeNisi, 1996). Performance feedback also affects self-efficacy judgments and training performance, especially when recipients attribute the feedback to controllable factors (Martocchio & Dulebohn, 1994). In addition, it can enhance intrinsic work motivation and job satisfaction, work-related attitudes that are important for explaining performance in a variety of settings, and for controlling absenteeism, turnover and organizational citizenship behavior (Bandura, 1997; Hackman & Oldham, 1976; Organ, 1990). Telecommuting is likely to affect feedback because to some extent this work arrangement removes the employee from the feedback environment. Revealing telecommuting’s effects on performance feedback requires conceptualizing the organization as a feedback environment and the recipient as an active rather than passive participant in the feedback process. From a feedback perspective, the organizational context is a complex and rich network of information that provides multiple cues for assessing individuals’ task performance. The office environment provides individuals performance feedback from multiple sources, including the formal organization, immediate supervisor, co-workers, customers, the task and self (Greller & Herold, 1975). While the traditional work setting offers performance feedback from multiple sources, the quality of the feedback individuals receive from these sources is not equivalent. Sources that are closer psychologically to the individual (e.g. task, self) seem to be the most informative and useful for enhancing task performance (Herold & Parsons, 1985). For instance, over twenty years of job characteristics research indicates that task feedback is consistently and positively related to job performance and negatively related to absenteeism and turnover (Fried & Ferris, 1987). In addition, more recent

146

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

individual studies find that task feedback has positive effects on task performance independent of external feedback and especially when individuals are committed to task goals (Goodman, 1998). Explaining telecommuting’s effects on feedback also requires viewing individuals as proactive participants in acquiring, generating, processing and responding to feedback messages. That is, individuals are not passive thermostats waiting idly to receive feedback to be used to reduce discrepancies between standards and actual performance (Ashford & Cummings, 1983; Herold & Fedor, 1998). Instead individuals proactively seek, create and shape their own feedback (Ashford, 1986). In Herold and Fedor’s (1998) words, “. . . the performer acts as both sender and receiver of performance feedback, generates feedback through interactions with, and manipulation of the external environment, is the receiver of various feedback messages from other sources, as well as the processor of this wide array of performance cues” (p. 221). Telecommuting thus removes individuals from an organizational setting in which they shaped and acquired performance feedback from multiple sources. By removing individuals from the organizational feedback environment, telecommuting eliminates or reduces access to some sources of feedback, and changes the nature of the relationship between the telecommuter and other sources of feedback. For example, as the frequency of telecommuting increases, telecommuters will have less exposure to and interaction with the formal organization, co-workers and managers, all which are valuable and distinct sources of positive and negative feedback (Herold & Parsons, 1985). Because they have less casual and physical access to these sources, telecommuters generally would be expected to acquire and receive less feedback from them. In addition, the feedback they do receive from these sources will be based on a narrower range of performance behaviors, as managers and co-workers have fewer opportunities to observe the full range of a telecommuter’s in-role and extra-role work behaviors. Feldman and Gainey (1997) argue that telecommuting decreases interactions with supervisors and co-workers, and Shamir and Salomon (1985) posit that working from home can decrease feedback from both these sources. Lowery (1996) also reports that employees who are physically distant from their supervisors perceive less communication with them. When supervisors do communicate with telecommuters, they may use “leaner” communication media that conveys less information, such as e-mail and telephone that do not contain facial expressions or body language (Cascio, 2000; Daft & Lengel, 1984). As telecommuters spend more time working at locations that isolate them from the organization and its members (e.g. home), they may reconfigure their sources of feedback and become more active in seeking feedback (Ashford & Cummings, 1983). Telecommuters, for instance, may become more attuned to task feedback,

The Impact of Telecommuting Design

147

generate more feedback themselves (i.e. self-generated feedback) and/or seek more feedback from customers and telecommuting peers in an effort to fill a feedback void caused by acquiring and receiving less feedback from the organization and its members. Whether telecommuters rely on the task or self-generated feedback versus seeking feedback from customers or telecommuting peers likely will be affected by their feedback propensities. Feedback propensity refers to a predisposition to desire feedback from internal or external sources (Herold, Parsons & Rensvold, 1996). High external propensity is a strong preference for feedback from external agents (e.g. supervisors, co-workers, customers); high internal propensity refers to a penchant for internally-generated feedback. High external feedback propensity individuals report that the task provides less feedback than do those with high internal propensities (Herold et al., 1996). Consequently, telecommuters with a high external feedback propensity may rely less on the task and self for feedback and seek feedback more often from external agents (e.g. customers, telecommuting peers, managers) than their high internal propensity counterparts (Renn & Fedor, 2001). By contrast, high internal propensity telecommuters would be expected to view the task and self as preferable sources of feedback and thus rely more on these sources than feedback from others. Thus, telecommuting frequency will be negatively related to receiving performance feedback from the formal organization and its members, and as telecommuting frequency increases, telecommuters with high external propensities will seek more performance feedback from external agents and rely less on feedback from the task and self than will their high internal propensity counterparts.

Role Properties and Boundaries Because telecommuting moves the work role at least partially out of the office and often into the home, it likely changes the nature of both the work role itself and relationships between work and non-work roles. Since the nature and frequency of interactions with supervisors may change, telecommuting may influence the important role property of role ambiguity. Also, since telecommuting may blur distinctions between work and non-work roles, its effects on conflict among roles is important. Given the purported benefits of telecommuting for balancing work and family obligations, we focus on its likely effects on conflict between work and family roles. Role Ambiguity Role ambiguity occurs when there is a lack of information concerning the proper definition of a job, its goals, and the permissible means for accomplishing them

148

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

(Katz & Kahn, 1978). Organizations are comprised of systems of roles that function together to affect organizational outcomes. In this system of roles, members communicate expectations, directly and indirectly, for each role. When this communication is non-existent, confusing, or inefficient, role ambiguity can occur (Tubre & Collins, 2000, p. 157). Because it represents a lack of information about how to perform a job, role ambiguity has been hypothesized to reduce effortto-performance and performance-to-reward expectancies, which, in turn, would be expected to lower motivation and job performance (Jackson & Schuler, 1985). To the extent telecommuters are removed from the workplace, they may receive less information and fewer cues from supervisors and co-workers about role requirements. They would also be expected to have fewer opportunities to seek feedback from others and may receive less frequent performance feedback from supervisors (Ashford & Cummings, 1983). In this vein, Jackson and Schuler (1985) found that leader initiating structure and feedback from others were negatively related to role ambiguity. Thus, as telecommuters become more physically and psychologically removed from the organization, they may experience greater role ambiguity. However, more thorough and clear telecommuting policies will likely reduce feelings of uncertainty about role expectations. Jackson and Schuler (1985) also found that formalization was negatively related to role ambiguity. Therefore, telecommuting arrangements that reduce the opportunity for self-management are likely to be associated with less role ambiguity. Telecommuting frequency and telecommuting from locations not associated with the organization (home or neighborhood locations) should be positively related to perceived role ambiguity, while formalization of policies should be negatively related to perceived role ambiguity. Work and Family Role Conflict Role conflict exists when an occupant’s expectations for a role are in conflict with the expectations of a role sender (Katz & Kahn, 1978). One of the important purported benefits of telecommuting is to reduce conflict between work and family roles, and at least one survey reports that telecommuters and home-based workers found it easier to separate the demands of home and office than office workers (Trent, Smith & Wood, 1994). Yet, even a brief consideration of telecommuting arrangements suggests that in some instances telecommuters may be subject to greater work – family conflict because of the close proximity of family members, especially dependent ones, and the blurring of the distinction between work space and time as well as family space and time. Similarly, Shamir (1992) argues that it is na¨ve to believe that individuals ı can achieve a better balance between work and family simply by eliminating temporal and physical boundaries between the two spheres of life. Indeed studies

The Impact of Telecommuting Design

149

indicate that individuals who work at home report that one of the most frustrating aspects of working at home is interruptions from children (Costello, 1988). Ashforth, Kreiner and Fugate (2000) recently proposed a framework of micro role transitions that may elucidate individuals’ cognitive/affective experience of telecommuting and telecommuting outcomes. According to the framework, a pair of roles can be placed on a segmentation-integration continuum based on the degree of flexibility and permeability of their boundaries and the contrast of their identities (Ashforth et al., 2000). Highly segmented roles are separated by inflexible and impermeable boundaries and have high-contrast role identities, or identities that do not share core and peripheral features and, thus, are easily distinguished from one another. Highly integrated roles have flexible and permeable boundaries and have low-contrast role identities. Because of their inflexible, impermeable boundaries and high-contrast identities, highly segmented roles benefit from less blurring but suffer from the difficulty of transitioning from one role to another. In contrast, highly integrated roles suffer from blurring and benefit from the ease of transitioning between roles. Ashforth et al. (2000) suggest that high segmentation is more evident in work and home domains compared to telecommuting because each role is associated with specific settings and times. They further argue that segmentation is less evident in the at work domain because of shared organizational contexts and weaker differentiation of role identities (p. 476). Similarly, Rau and Hyland (2002) suggest that telecommuting arrangements are closer to the integration end of this spectrum. Following the Ashforth et al. (2000) framework, telecommuting may shift work and family roles along this continuum from high segmentation to higher integration which, in turn, would be expected to produce easier role transitions but greater role blurring. For example, a telecommuter working at home should be able to transition more easily from employee to parent when a child is crying in the room next to a home office than trying to do so from a corporate office. While telecommuting may lower the magnitude of the transition between work and home roles, it may also “create confusion and anxiety about which role identity is or should be most salient and produce greater interruptions without warnings (sic)” (Ashforth et al., 2000, p. 481). An example is a telecommuter working at home whose neighbor enjoys visiting and chatting about neighborhood activities in the middle of the workday. The telecommuter likely will experience conflict over which role should take precedent (i.e. employee versus neighbor), especially if he/she values the neighbor’s friendship. So, telecommuting frequency and telecommuting from home will increase the integration of work and family roles, which, in turn, creates greater inter-role conflict. At the same time, spillover theory stresses the importance of the amount of control individuals have over the permeable boundary between subsystems

150

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

such as work and family (Zedeck, 1992). If an individual has little control over interactions between subsystems, then spillover effects tend to be negative. For individuals with the power to control these interactions, spillover effects tend to be positive. Following this logic, the increased flexibility and control provided by some telecommuting designs may result in positive spillover. More specifically, individuals who have greater flexibility in determining when and from where they will work would seem to have greater control over the interactions between their work and family subsystems and, thus, are likely to experience less work-family conflict (Pleck, Staines & Lang, 1980; Thomas & Ganster, 1995). So, flexibility of scheduling will reduce work-family conflict. These role properties are important because many studies have examined their relationships with work-related outcomes, and at least three meta-analyses have found that both are negatively related to job performance, job satisfaction, and organizational commitment (Fisher & Gitelson, 1983; Jackson & Schuler, 1985; Tubre & Collins, 2000). Further, there is evidence that when employees experience conflict between work and family roles, they also report experiencing greater job dissatisfaction and fatigue, and lower life satisfaction (Edwards & Rothbard, 2000; Frone, Russell & Cooper, 1992).

Moderators Given the complexity of these relationships, it becomes important to identify boundary conditions that specify when, for whom, and under what conditions the relationships are likely to hold. Doing so may help clarify, for example, whether there are some individuals who are less susceptible to the negative effects of physical and psychological separation from the organization. Thus, future research will need to consider moderators of the proposed relationships. Although a host of variables may affect these relationships, we discuss demographic and individual difference variables as well as some situational variables we believe to be especially important in a telecommuting context. Demographics and Individual Differences With respect to demographics and individual differences, Blegen, Mueller and Price (1988) describe kinship responsibilities as a complex constellation of family obligations based, in part, on marital status, number of children and their age. Hall (1990) argues that telecommuters with children should experience more stress than those without children, and Hill, Hawkins and Miller (1996) found evidence that telecommuting did not ease work-family conflict for all respondents, depending on the presence of young children. Thus, increased kinship responsibilities

The Impact of Telecommuting Design

151

could mitigate effects of increased self-management with work-family conflict. Organizational tenure may serve as a boundary condition of relationships between separation and social relationships. Rousseau (1989) suggests that telecommuters who have previously worked full-time in the office should feel greater inclusion and be more socialized than those who have not. Hence, longer organizational tenure may mitigate the negative effects telecommuting is proposed to have on social support. Similar effects might be found for telecommuters who already have extensive experience working in a virtual environment. Need for affiliation is an individual difference that researchers should also consider (McClelland, 1965). Individuals differ in terms of the amount and types of interpersonal interaction and affiliation they need. Telecommuters who require a great deal of interpersonal interaction (high nAff) will likely have more difficulty dealing with increased physical and psychological separation from peers and supervisors at work (Frolick, Wilkes & Urwiler, 1993). Thus, high nAff telecommuters may be more prone to seek out other groups with whom to identify and be more likely to identify less with the organization (i.e. organizational commitment) as a result of separation than low nAff telecommuters. Situational Moderators Situational moderators include training, communication, and boundary management. Telecommuters may need to learn certain skills and develop abilities to effectively manage their alternative work arrangement. For example, telecommuters and their supervisors may benefit from training in communication and feedback because telecommuting will likely reduce impromptu and planned faceto-face communication that are essential for reducing role ambiguity, planning, and adjusting goals and work-related strategies. Communication richness would be expected to assume a critical role in understanding proposed relationships with agent feedback and role ambiguity (Fulk & Boyd, 1991). Supervisors who frequently use the telephone to communicate with telecommuting subordinates, for example, may be perceived as providing work-related information and sufficient feedback, even though face-to-face discussions are rare (Daft & Lengel, 1984; Trevino, Daft & Lengel, 1987). And, there is some empirical evidence that telecommuters tend to use less rich communication (Raghuram, 1996). Venkatesh and Johnson (2002) recently suggested that organizations that invest in a virtual office environment for their telecommuters might improve communication through the use of media with greater social richness and telepresence. Lastly, to manage potential interruptions and reduce the potential for conflict, telecommuters may benefit from maintaining greater segmentation between work and family roles by erecting, maintaining, and defending boundaries between these roles. Ashforth et al. (2000) cite research on at-home workers who attained

152

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

greater segmentation by “creating a physical workspace, marking the territory with equipment and furniture, restricting access of others, rescheduling domestic tasks, and setting time when people can call or talk to them” (p. 482).

DISCUSSION
Organizational research lags far behind telecommuting practice. Consequently, we proposed a framework for investigating telecommuting that should facilitate closing the gap between research and practice on an alternative work arrangement that will likely continue to spread. We proposed a parsimonious set of telecommuting dimensions and tied those dimensions to social systems, selfmanagement opportunities and requirements, and the work–non-work interface. We suggested that the effects of telecommuting on attitudes and behaviors such as job performance, job satisfaction and organizational commitment are too complex for simple relationships (e.g. telecommuting increases job satisfaction). However, it is possible to analyze specific telecommuting designs and analyze their likely impact on the proposed mediators and thus begin to understand the likely effects on outcomes. This analysis should aid researchers in clarifying the complex and potentially contradictory effects of telecommuting. It is not practical here to identify every aspect and combination of telecommuting design and attempt to trace the impact. We can, though, examine examples of aspects of telecommuting design in order to see what our analysis suggests in terms of impact on outcomes. Consider, for example, the two arrangements identified earlier in the manuscript: an employee required to work full-time at a satellite center under close electronic monitoring, and an employee allowed to telecommute from any location at their discretion. Both arrangements are common, but our framework suggests that these two telecommuters might have very different experiences and responses. Benefits to the first employee would include increased feelings of autonomy and control and increased opportunities for self-management stemming from the high frequency of telecommuting. However, there may be a number of concerns as well. The high frequency may be associated with receiving less feedback, weaker identification with the organization, lower perceived social support from organizational members, and increased role ambiguity. Some of these relationships may be mitigated by the telecommuting location. For example, telecommuting from a satellite center, where the employee will have occasion to interact with agents and symbols of the organization, would probably not have as negative an impact on identification and social support as doing so from home. At the same time, the lack of flexibility, tight performance monitoring, and perceptions that the arrangement is intended almost entirely to benefit the organization could lead to lower POS

The Impact of Telecommuting Design

153

and perceptions of supervisory support. Overall, it is difficult to see how this arrangement could be expected to result in increased satisfaction, commitment, or performance for this employee. Organizational policies regarding telecommuting were left unspecified in this example, but formal policies on communication and feedback could serve to mitigate some of the effects on isolation and ambiguity. The second employee, allowed to telecommute at their discretion, would be expected to have different experiences. The high flexibility of scheduling the timing and location of work, along with perceptions that the arrangement is intended at least partially to benefit the employee, should lead to increased perceptions of autonomy and social support, increased POS, and potentially lower work–non-work conflict because of the increased control over the boundaries between these two spheres. These expectations suggest the potential for more positive work attitudes and perhaps increased performance as well. Several aspects of telecommuting design were not specified in this example, but it might be important to consider managing the frequency of telecommuting or policies regarding communication because of their potential effects on identification and social support. Of course, these are only two examples that do not begin to specify the possible combinations of telecommuting designs and their effects on outcomes. Our goal was to point out how a more detailed analysis of the variations in how telecommuting arrangements are designed and their direct effects on important mechanisms might lead to an increased understanding of how telecommuting can influence attitudes and performance. At the same time, these proposed relationships are subject to empirical verification.

Future Research Clearly, the effects of telecommuting on attitudes and behavior are complex. Research is needed to evaluate the propositions and the processes through which telecommuting affects job performance, job satisfaction and organizational commitment. For example, research is needed on the design dimensions. Do individuals perceive telecommuting along the proposed dimensions? We derived the present dimensions inductively and from previous conceptual work (e.g. Feldman & Gainey, 1997). However, we are unaware of qualitative or empirical studies that have systematically defined and measured individuals’ perceptions of telecommuting arrangements. Other design facets may be important, such as technical requirements and support. Similarly, to what extent do the individual design facets operate together? Will the potentially positive effect on autonomy of providing greater scheduling flexibility be negated by very formal and restrictive policies? To answer these and similar questions, researchers may need to wrestle with how to weigh or combine the design dimensions.

154

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Research is also needed on the proposed mediating mechanisms. Are the proposed design facets related to these mechanisms as suggested? To what extent do these mechanisms account for relationships with outcomes? And, what are the total effects on outcomes of competing processes? For example, if increased self-management opportunities do increase autonomy and perceived organizational support and decrease work – family conflict, but at the same time increase role ambiguity, what will be the net effect on attitudes and behaviors? The role of social identity processes may also lead to several important research questions. With which other groups will a telecommuter seek to identify? Will a telecommuter, for instance, begin identifying with independent consultants who, in turn, influence the telecommuter to quit full-time work? How long will it take for individuals to begin searching for other groups with whom to identify? Finally, other outcomes may be important. For example, although we discussed the likely effects of telecommuting on performance per se, we did not address the possible effects on contextual performance or organizational citizenship behaviors (e.g. Motowidlo & Van Scotter, 1994; Organ, 1987). It seems plausible that the greater autonomy and reduced social contact with others at the work place may reduce or inhibit such prosocial unrewarded activities. The effects on work-family conflict may also depend on the type of conflict. Some research suggests three types of work-family conflict: time-based (the demands of one role make it difficult to be available to the other role), strain-based (the strain from one role interferes with the other role, like separating from a hard day at work when you get home), and behavior based (behavioral expectations in one role are incompatible with those in another role) (Greenhaus & Beutell, 1985). Telecommuting may reduce time-based work-family conflict, but increase strain-based and behavior-based work-family conflict (McCloskey & Igbaria, 2003). Voluntary turnover may be another outcome of interest. We have argued that telecommuting affects both job satisfaction and organizational commitment, typically the two most important attitudinal variables in turnover research (cf. Hom & Griffeth, 1995). Telecommuting might also influence turn‘over by providing a work option for specific populations who might otherwise quit such as individuals close to retirement, who are unwilling to relocate, or with new family responsibilities. Current employees may also demand telecommuting as an option and, if not offered, they may leave for alternatives that do offer a telecommuting option.

Practical Implications Telecommuters are the fastest growing segment of home-based workers and managers are puzzled about supervising from a distance and the effect of telecommuting

The Impact of Telecommuting Design

155

on teamwork and corporate identity (Becker & Steele, 1995; Dunkin, 1995). What are the best telecommuting arrangements for employees? Are there key dimensions of telecommuting that can be diagnosed and manipulated to increase the likelihood that telecommuting will have a positive effect on job performance and organization identification (i.e. commitment)? How will a particular individual appraise telecommuting, and are certain employees better suited for telecommuting than others? Our framework provides specific telecommuting design facets that can be measured and manipulated by managers in response to these questions and others. Our analysis also suggests ways that organizations can minimize the potential negative effects of telecommuting. Although telecommuting allows organizations and employees to overcome time and distance constraints (Fulk & DeSanctis, 1995; O’Mahony & Barley, 1999), organizations are concerned with the long-term effects of telecommuting on employees’ identification with the organization (Elsbach, 1999). Our analysis does suggest that over time, physical and psychological separation from the organization’s context and work groups may diminish telecommuters’ organization identification. To overcome feelings of isolation, managers may need to plan activities/assignments that require telecommuters to spend time on company premises and to interact with work groups. Becker and Steele (1995) argue that individuals who have been working at home all day do not want to come to the company office to do the same thing. Rather, while on company premises, they want to interact with others, and discuss customers and goals. Managers, thus, may consider redesigning corporate offices and telecenters to accommodate such needs. Managers may also organize extracurricular activities where corporate employees and telecommuters can work together to achieve common goals (e.g. community service projects) (Elsbach & Glynn, 1996). Doing so may allow telecommuters to continue to internalize the organization’s collective values, meanings, and standards and, thus, maintain their attachment to the organization (Ashforth & Kreiner, 1999; Benkhoff, 1997). Our analysis also suggests that telecommuting promotes greater integration of home and work roles (Ashforth et al., 2000). Thus, organizations must be sensitive to the potential blurring of work and family roles created by telecommuting. Shifting work and family roles from high segmentation to higher integration may cause greater work – family conflict unless a telecommuter is able to erect and maintain boundaries between these roles. Hence, training telecommuters in designing physical workspaces for optimal separation seems appropriate. Finally, managers should be sensitive to telecommuters’ needs to have a life outside of work. Telecommuters may work at home, but their home is not a twenty-four hour workplace. In sum, the question of whether telecommuting is a good or bad way of designing work is probably not as important as questions of how to design and implement

156

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

telecommuting arrangements. Societal pressures to reduce traffic congestion, air pollution, and fuel consumption are likely to continue to push organizations to utilize telecommuting. At the same time, pressures from employees demanding greater flexibility and work-family balance, especially in tight labor markets, may also lead to increased telecommuting. And, employers who see opportunities to reduce overhead costs or tap underutilized labor sources will continue to experiment with telecommuting. Thus, the important question becomes twofold: what are the effects of designing work to include telecommuting; and how can we design telecommuting to take advantage of its potential benefits while minimizing its potential drawbacks? This framework should assist managers and researchers in systematically answering those questions.

ACKNOWLEDGMENTS
The authors would like to acknowledge the helpful comments of Rabi Bhagat. Portions of this research were supported by a grant from the Fogelman College of Business and Economics at the University of Memphis.

REFERENCES
Albert, S., Ashforth, B. E., & Dutton, J. E. (2000). Organizational identity and identification: Charting new waters and building new bridges. Academy of Management Review, 25, 13–17. Allen, D. G., Shore, L. M., & Griffeth, R. W. (2003). The role of perceived organizational support and supportive human resource practices in the turnover process. Journal of Management, 29, 99–118. Armstrong, L. (1993). The office is a terrible place to work. Business Week, 46. Ashford, S. J. (1986). Feedback-seeking in individual adaptation: A resource perspective. Academy of Management Journal, 29, 465–487. Ashford, S. J., & Cummings, L. L. (1983). Feedback as an individual resource: Personal strategies of creating information. Organizational Behavior and Human Performance, 32, 370–398. Ashforth, B. E., & Kreiner, G. E. (1999). How can you do it? Dirty work and the challenge of constructing a positive identity. Academy of Management Review, 24, 413–434. Ashforth, B. E., Kreiner, G. E., & Fugate, M. (2000). All in a day’s work: Boundaries and micro role transitions. Academy of Management Review, 25, 472–491. AT&T (1998). AT&T telework guide: A telework success story. http://www.att.com/telework/ chap05.html Bailey, D. E., & Kurland, N. B. (2002). A review of telework research: Findings, new directions, and lessons for the study of modern work. Journal of Organizational Behavior, 23, 383–400. Bandura, A. (1986). Social foundations of thought and action: A social cognitive view. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman Press.

The Impact of Telecommuting Design

157

Bandura, A., & Wood, R. E. (1989). Effect of perceived controllability and performance standards on self regulation of complex decision making. Journal of Personality and Social Psychology, 56, 805–814. Bartol, K. M., & Liu, W. (2002). Information technology and human resources management: Harnessing the power and potential of netcentricity. In: G. Ferris (Ed.), Research in Personnel and Human Resources Management (Vol. 21, pp. 215–242). Greenwich, CT: JAI Press. Bazerman, M. H. (1982). Impact of personal control on performance: Is added control always beneficial? Journal of Applied Psychology, 67, 472–479. Becker, F., & Steele, F. (1995). Workplace by design: Mapping the high performance workscape. San Francisco, CA: Jossey-Bass. Benkhoff, B. (1997). A test of the HRM model: Good for employers and employees. Human Resource Management Journal, 7, 44–60. Blegen, M. A., Mueller, C. W., & Price, J. L. (1988). Measurement of kinship responsibility for organizational research. Journal of Applied Psychology, 73, 402–409. Cascio, W. F. (2000). Managing a virtual workplace. Academy of Management Executive, 14, 81–90. Chapman, A. J., Sheehy, N. P., Heywood, S., Dooley, B., & Collins, S. C. (1995). The organizational implications of teleworking. In: C. L. Cooper & I. T. Robertson (Eds), International Review of Industrial and Organizational Psychology (Vol. 10, pp. 229–248). Chrobot, D. L., Janat, E. A., & Gerstner, C. R. (1997). The impact of problem – focused coping on quality of work life in the virtual office environment. Paper presented at the meeting of the Society for Industrial and Organizational Psychology, St. Louis, MO. Clancy, T. (1994). The virtual corporation, telecommuting, and the concept of team. Academy of Management Executive, 8, 7–10. Cooper, C. D., & Kurland, N. B. (2002). Telecommuting, professional isolation, and employee development in public and private organizations. Journal of Organizational Behavior, 23, 511–532. Costello, C. B. (1988). Clerical home-based work: A case study of work and family. In: K. E. Christensen (Ed.), The New Era of Home-Based Work (pp. 135–145). Boulder, CO: Westview Press. Daft, R. L., & Lengel, R. H. (1984). Information richness: A new approach to managerial behavior and organizational design. Research in Organizational Behavior, 6, 191–233. Dunkin, A. (1995). Taking care of business – without leaving your house. Business Week, 106–107. Duxbury, L., & Neufeld, D. (1999). An empirical evaluation of the impacts of telecommuting on intra-organizational communication. Journal of Engineering and Technology Management, 16, 1–28. Eagle, B. W., Miles, E. W., & Icenogle, M. L. (1997). Interrole conflicts and the permeability of work and family domains: Are there gender differences? Journal of Vocational Behavior, 50, 168–184. Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work and family: Clarifying the relationship between work and family constructs. Academy of Management Review, 25, 178–199. Eisenberger, R., Huntington, R., Hutchinson, S., & Sowa, D. (1986). Perceived organizational support. Journal of Applied Psychology, 71, 500–507. Elder, A. E., & Smith, C. S. (1999). Telework: Implications for the individual. Paper presented at the meeting Society for Industrial and Organizational Psychology, Atlanta, GA. Elsbach, K. D. (1999). An expanded model of organizational identification. In: R. Sutton & B. Staw (Eds), Research in Organizational Behavior (Vol. 21, pp. 163–200). Greenwich, CT: JAI Press. Elsbach, K. D., & Glynn, M. A. (1996). Believing your own PR: Embedding identification in strategic reputation. In: J. Baum & J. Dutton (Eds), Advances in Strategic Management (Vol. 13, pp. 65–90). Greenwich, CT: JAI Press.

158

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Emery, F. E., & Trist, E. L. (1960). Socio-technical systems theory. In: C. H. Churchman & M. Verhulst (Eds), Management Science, Models and Techniques (Vol. 2, pp. 83–97). New York: Pergamon. Fairweather, N. B. (1999). Surveillance in employment: The case of teleworking. Journal of Business Ethics, 22, 39–49. Farrah, B. J., & Dagen, C. D. (1993). Telecommuting policies that work. HR Magazine, 64–71. Feldman, D. C. (1984). The development and enforcement of group norms. Academy of Management Review, 9, 47–53. Feldman, D. C., & Gainey, T. W. (1997). Patterns of telecommuting and their consequences: Framing the research agenda. Human Resource Management Review, 7, 369–388. Fisher, C. D., & Gitelson, R. A. (1983). A meta-analysis of the correlates of role conflict and ambiguity. Journal of Applied Psychology, 68, 320–333. Frayne, C., & Geringer, M. (2000). Self-management training for improving job performance: A field experiment involving salespeople. Journal of Applied Psychology, 85, 361–372. Fried, Y., & Ferris, G. R. (1987). The validity of the job characteristics model: A review and metaanalysis. Personnel Psychology, 40, 287–322. Frolick, M. N., Wilkes, R. B., & Urwiler, R. (1993). Telecommuting as a workplace alternative: An identification of significant factors in American firms’ determination of work-at-home policies. Journal of Strategic Information Systems, 2, 206–222. Frone, M. R., Russell, M., & Cooper, M. L. (1992). Antecedents and outcomes of work – family conflict: Testing a model of the work-family interface. Journal of Applied Psychology, 77, 65–78. Fulk, J., & Boyd, B. (1991). Emerging theories of communication in organizations. Journal of Management, 17, 407–446. Fulk, J., & DeSanctis, G. (1995). Electronic communication and changing organizational forms. Organization Science, 6, 337–349. Gist, M. E., & Mitchell, T. R. (1992). Self-efficacy: A theoretical analysis of its determinants and malleability. Academy of Management Review, 17, 183–211. Glass, J. L., & Finley, A. (2002). Coverage and effectiveness of family – responsive workplace policies. Human Resource Management Review, 12, 313–337. Goodman, J. S. (1998). The interactive effects of task and external feedback on practice performance and learning. Organizational Behavior and Human Decision Processes, 76, 223–252. Goodrich, J. N. (1990). Telecommuting in America. Business Horizons, 31–37. Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement. American Sociological Review, 25, 161–178. Greenberger, D. B., & Strasser, S. (1986). Development and application of a model of personal control in organizations. Academy of Management Review, 11, 164–177. Greenberger, D. B., Strasser, S., Cummings, L. L., & Dunham, R. B. (1989). The impact of personal control on performance and satisfaction. Organizational Behavior and Human Performance, 43, 29–51. Greenhaus, J. H., & Beutell, N. J. (1985). Sources of conflict between work and family roles. Academy of Management Review, 10, 76–88. Greller, M. M., & Herold, D. M. (1975). Sources of feedback: A preliminary investigation. Organizational Behavior and Human Performance, 13, 244–256. Griffin, R. W. (1983). Objective and social sources of information in task redesign: A field experiment. Administrative Science Quarterly, 28, 184–200. Hackman, J. R., & Oldham, G. (1976). Motivation through the design of work: Test of a theory. Organizational Behavior and Human Performance, 16, 250–279. Hackman, J. R., & Oldham, G. (1980). Work redesign. Reading, MA: Addison-Wesley.

The Impact of Telecommuting Design

159

Hackman, J. R., Oldham, G., Janson, R., & Purdy, K. (1975, Summer). A new strategy for job enrichment. California Management Review, 57–71. Hall, D. T. (1990). Telecommuting and the management of work – home boundaries. In Paradigms revised: Annual Review of Communications and Society (pp. 177–208). Institute for Information Studies. Nashville, TN: Northern Telecom & Queenstown, MD: Aspen Institute. Handy, S. L., & Mokhtarian, P. L. (1995). Planning for telecommuting: Measurement and policy issues. Journal of the American Planning Association, 61, 99–111. Herold, D. M., & Fedor, D. B. (1998). Individuals interaction with their feedback environment: The role of domain specific individual differences. In: G. R. Ferris (Ed.), Research in Personnel and Human Resource Management (Vol. 16, pp. 215–254). Greenwich, CT: JAI Press. Herold, D. M., & Parsons, C. K. (1985). Assessing the feedback environment in work organizations: Development of the job feedback survey. Journal of Applied Psychology, 70, 290–305. Herold, D. M., Parsons, C. K., & Rensvold, R. B. (1996). Individual propensities in the generation and processing of performance feedback (Working Paper). Georgia Institute of Technology. Higo, K., Sheng, O. R., Shin, B., & Figueredo, A. J. (2000). Understanding relationships among teleworkers’ e-mail usage, e-mail richness perceptions, and e-mail productivity perceptions under a software engineering environment. IEEE Transactions on Engineering Management, 47, 163–173. Hill, E. J., Hawkins, A. J., & Miller, B. C. (1996). Work and family in the virtual office: Perceived influences of mobile telework. Family Relations, 15, 293–301. Hill, E. J., Miller, B. C., Weiner, S. P., & Colihan, J. (1998). Influences of the virtual office on aspects of work and work/life balance. Personnel Psychology, 51, 667–683. Hogg, M. A., & Mullin, B. A. (1999). Joining groups to reduce uncertainty: Subjective uncertainty reduction and group identification. In: D. Abrams & M. A. Hogg (Eds), Social Identity and Social Cognition (pp. 249–279). Oxford: Blackwell. Hogg, M. A., & Terry, D. J. (2000). Social identity and self-categorization processes in organizational contexts. Academy of Management Review, 25, 121–140. Hom, P. W., & Griffeth, R. W. (1995). Employee turnover. Cincinnati, OH: South-Western College Publishing. Igbaria, M., & Guimaraes, T. (1999). Exploring differences in employee turnover intentions and its determinants among telecommuters and non-telecommuters. Journal of Management Information Systems, 16, 147–164. Ilgen, D. R., Fisher, C. D., & Taylor, M. S. (1979). Consequences of individual feedback on behavior in organizations. Journal of Applied Psychology, 64, 349–371. International Telework Association (2000). http://www.telecommute.org/ Jackson, S. E., & Schuler, R. S. (1985). A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work settings. Organizational Behavior and Human Decision Processes, 36, 16–78. Katz, D., & Kahn, R. L. (1978). The social psychology of organizations. New York: Wiley. Kirkman, B. L., Rosen, B., Gibson, C. B., Tesluk, P. E., & McPherson, S. O. (2002). Five challenge to virtual team success: Lessons from Sabre, Inc. Academy of Management Executive, 16, 67–79. Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119, 254–284. Kraut, R., Patterson, M., Lundmark, V., Kiesler, S., Mukopadhyay, T., & Scherlis, W. (1998). Internet paradox: A social technology that reduces social involvement and psychological well-being? American Psychologist, 53, 1017–1031.

160

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Kurland, N. B., & Bailey, D. E. (1999, Autumn). The advantages and challenges of working here, there, anywhere, and anytime. Organizational Dynamics, 53–68. Ladda, C. (1995). Telecommuting raises productivity. HRFocus, 6. Lamond, D., Standen, P., & Daniels, K. (2000). The psychosocial impact of telework: A theoretical and research framework. Paper presented at the Academy of Management Meeting, Toronto, Canada. Locke, E. A., & Latham, G. P. (1990). A Theory of Goal Setting and Task Performance. Englewood Cliffs, NJ: Prentice-Hall. Loher, B. T., Noe, R. A., Moeller, N. L., & Fitzgerald, M. P. (1985). A meta-analysis of the relations of job characteristics to job satisfaction. Journal of Applied Psychology, 70, 280–289. Lowery, T. (1996). Alternative work arrangements: The effects of distance and media use on the supervisor-subordinate relationship. Unpublished doctoral dissertation. Rice University. Manz, C. C., & Neck, C. P. (1999). Mastering self-leadership: Empowering yourself for personal excellence. Englewood Cliffs, NJ: Prentice-Hall. Manz, C. C., & Sims, H. P. (1987). Leading workers to lead themselves: The external leadership of self-managed working teams. Administrative Science Quarterly, 32, 106–128. Martocchio, J. J., & Dulebohn, J. (1994). Performance feedback effects in training: The role of perceived controllability. Personnel Psychology, 47(2), 357–373. Maynard, R. (1994). The growing appeal of telecommuting. Nation’s Business, 61–62. McClelland, D. C. (1965). Toward a theory of motive acquisition. American Psychologist, 23, 321–333. McCloskey, D. W., & Igbaria, M. (2003). Does ‘Out of Sight’ mean ‘Out of Mind’? An empirical investigation of the career advancement prospects of telecommuters. Information Resources Management Journal, 16, 19–34. McCune, J. C. (1998). Telecommunicating revisited. Management Review, 87(10). Mills, P. K. (1983). Self-management: Its control and relationship to other organizational properties. Academy of Management Review, 8, 445–453. Moorman, R. H., Blakely, G. L., & Niehoff, B. P. (1998). Does perceived organizational support mediate the relationship between procedural justice and organizational citizenship behavior? Academy of Management Journal, 41, 351–368. Mowday, R. T., Porter, L. W., & Steers, R. M. (1982). Employee-organization linkages: The psychology of commitment, absenteeism, and turnover. New York: Academic Press. Niehoff, B. P., & Moorman, R. H. (1993). Justice as a mediator of the relationship between methods of monitoring and organizational citizenship behavior. Academy of Management Journal, 36, 527–556. Oldham, G. R., Cummings, A., & Zhou, J. (1995). The spatial configuration of organizations. Research in Personnel and Human Resources Management, 13, 1–37. Olson, M. H. (1989). Work at home for computer professionals: Current attitudes and future prospects. ACM Transactions on Information Systems, 7, 317–338. Olszewski, P., & Mokhtarian, P. (1994). Telecommuting frequency and impacts for state of California employees. Technological Forecasting and Social Change, 45, 275–286. O’Mahony, S., & Barley, S. R. (1999). Do digital communications affect work and organization? The state of our knowledge. In: R. Sutton & B. Staw (Eds), Research in Organizational Behavior (Vol. 21, pp. 125–162). Greenwich, CT: JAI Press. Organ, D. W. (1990). The motivational basis of organizational citizenship behavior. In: B. M. Staw & L. L. Cummings (Eds), Research in Organizational Behavior (Vol. 12, pp. 43–72). Greenwich, CT: JAI Press. Parker, S., & Wall, T. (1998). Job and work design. Thousand Oaks, CA: Sage Publications.

The Impact of Telecommuting Design

161

Passmore, W., & Sherwood, J. (1978). Sociotechnical systems: A sourcebook. San Diego, CA: University Associates. Perlow, L. (1997). Finding time: How corporations, individuals, and families can benefit from new work practices. New York: Cornell University. Peters, L. H., & O’Connor, E. J. (1980). Situational constraints and work outcomes: The influences of a frequently overlooked construct. Academy of Management Review, 5, 391–397. Pleck, J. H., Staines, G. L., & Lang, L. (1980). Conflict between work and family life. Monthly Labor Review, 103, 29–32. Pratt, J. H. (1984). Home teleworking: A study of its pioneers. Technological Forecasting and Social Change, 25, 1–14. Raghuram, S. (1996). Knowledge creation in the telework context. International Journal of Technology Management, 11, 859–870. Raghuram, S., Garud, R., Wiesenfeld, B., & Gupta, V. (2001). Factors contributing to virtual work adjustment. Journal of Management, 27, 383–405. Raghuram, S., Wiesenfeld, B., & Garud, R. (2000). Adjustment to telecommuting: Role of selfefficacy and structuring. Paper presented at the Academy of Management Meeting, Toronto, Canada. Rau, B. L., & Hyland, M. M. (2002). Role conflict and flexible work arrangements: The effects on applicant attraction. Personnel Psychology, 55, 111–136. Renn, R. W., & Fedor, D. B. (2001). Development and field test of a feedback seeking, self-efficacy, and goal setting model of work performance. Journal of Management, 27, 563–584. Renn, R. W., & Vandenberg, R. J. (1995). The critical psychological states: An underrepresented component in job characteristics model research. Journal of Management, 21, 279–304. Roberts, K. H., & Grabowski, M. (1996). Organizations, technology and structuring. In: S. Clegg, C. Hardy & W. Nord (Eds), Handbook of Organization Studies (pp. 409–423). Thousand Oaks, CA: Sage Publications. Rousseau, D. M. (1977). Technological differences in job characteristics, employee satisfaction, and motivation: A synthesis of job design research and sociotechnical systems theory. Organizational Behavior and Human Performance, 19, 18–42. Rousseau, D. M. (1989). Human resource planning for the future. Annual Review of Psychology, 245–266. Salancik, G. R., & Pfeffer, J. (1978). A social information processing approach to job attitudes and task design. Administrative Science Quarterly, 23, 224–253. Schneider, B. (1990). The climate for service: An application of the climate construct. In: B. Schneider (Ed.), Organizational Climate and Culture (pp. 383–412). San Francisco: Jossey-Bass. Scott, C. R., & Timmerman, C. E. (1999). Communication technology use and multiple workplace identifications among organizational teleworkers with varied degrees of virtuality. IEE Transactions on Professional Communication, 42, 240–260. Shamir, B. (1992). Home: The perfect workplace? In: S. Zedeck (Ed.), Work, Families, and Organizations (pp. 272–311). San Francisco: Jossey-Bass Publishers. Shamir, B., & Salomon, I. (1985). Work-at-home and the quality of working life. Academy of Management Review, 10, 455–464. Sheldon, M. E. (1971). Investments and involvements as mechanisms as producing commitment to the organization. Administrative Science Quarterly, 16, 142–150. Shore, L. M., & Shore, T. M. (1995). Perceived organizational support and organizational justice. In: R. S. Cropanzano & M. Kacmar (Eds), Organizational Politics, Justice, and Support: Managing the Social Climate of the Workplace (pp. 149–164).

162

DAVID G. ALLEN, ROBERT W. RENN AND RODGER W. GRIFFETH

Shore, L. M., & Tetrick, L. E. (1991). A construct validity study of the survey of perceived organizational support. Journal of Applied Psychology, 76, 637–643. Slocum, J. W., & Sims, H. P. (1980). A typology for integrating technology, organization, and job design. Human Relations, 33, 193–212. Sparrow, P. R., & Daniels, K. (1999). Human resource management and the virtual organization: Mapping the future research issues. In: C. L. Cooper & D. M. Rousseau (Eds), Trends in Organizational Behavior (pp. 44–61). John Wiley & Sons. Spector, P. E. (1986). Perceived control by employees: A meta-analysis of studies concerning autonomy and participation at work. Human Relations, 39, 1005–1016. Spreitzer, G. M. (1995). Psychological empowerment in the workplace: Dimensions, measurement, and validation. Academy of Management Journal, 38, 1442–1465. Sturgill, A. F. C. (1998). Relationship of telecommuting to organizational communication: A preliminary study of group process and communication satisfaction. Unpublished doctoral dissertation. Cornell University. Taylor, M. S., Fisher, C. D., & Ilgen, D. R. (1984). Individuals’ reactions to performance feedback in organizations: A control theory perspective. In: G. Ferris & K. Rowland (Eds), Research in Personnel and Human Resource Management (Vol. 2, pp. 81–124). Greenwich, CT: JAI Press. Teo, T. S. H., Lim, V. K. G., & Wai, S. H. (1998). An empirical study of attitudes towards teleworking among information technology personnel. International Journal of Information Management, 18, 329–343. Thomas, L. T., & Ganster, D. C. (1995). Impact of family-supportive work variables on work-family conflict and strain: A control perspective. Journal of Applied Psychology, 80, 6–15. Thomas, K. W., & Velthouse, B. A. (1990). Cognitive elements of empowerment. Academy of Management Review, 15, 666–681. Thompson, J. D. (1967). Organizations in action. New York: McGraw-Hill. Tomaskovic-Devey, D., & Risman, B. J. (1993). Telecommuting innovation and organization: A contingency theory of labor process change. Social Science Quarterly, 74, 367–385. Trent, J. T., Smith, A. L., & Wood, D. L. (1994). Telecommuting: Stress and social support. Psychological Reports, 74, 1312–1314. Trevino, L. K., Lengel, R. H., & Daft, R. L. (1987). Media symbolism, media richness, and media choice in organizations: A symbolic interactionist perspective. Communication Research, 14, 553–574. Trist, E. L., & Bamforth, K. W. (1954). Some social and psychological consequences of the long-wall method of coal-getting. Human Relations, 30, 3–38. Tsui, A. S., & Ashford, S. J. (1994). Adaptive self-regulation: A process view of managerial effectiveness. Journal of Management, 20, 93–121. Tubre, T. C., & Collins, J. M. (2000). Jackson and Schuler (1985) revisited: A meta-analysis of the relationships between role ambiguity, role conflict, and job performance. Journal of Management, 26, 155–169. Uhl-Bien, M., & Graen, G. B. (1998). Individual self-management: Analysis of professionals’ selfmanaging activities in functional and cross-functional work teams. Academy of Management Journal, 41, 340–350. Vega, G., & Brennan, L. (2000). Isolation and technology: The human disconnect. Journal of Organizational Change Management, 13, 468–481. Venkatesh, V., & Johnson, P. (2002). Telecommuting technology implementations: A within and between subjects longitudinal field study. Personnel Psychology, 55, 661–687.

The Impact of Telecommuting Design

163

Wayne, S. J., Shore, L. M., & Liden, R. C. (1997). Perceived organizational support and leader-member exchange: A social exchange perspective. Academy of Management Journal, 40, 82–111. Wells, S. J. (2001, October). Making telecommuting work. HR Magazine, 35–45. Wiesenfeld, B. M., Raghuram, S., & Garud, R. (1998). Communication patterns as determinants of organizational identification in a virtual organization. Journal of Computer Mediated Communication, 3. Wiesenfeld, B. M., Raghuram, S., & Garud, R. (2001). Organizational identification among virtual workers: The role of need for affiliation and perceived work-based social support. Journal of Management, 27, 213–229. Wood, R. E., & Bandura, A. (1989). Social cognitive theory of organizational management. Academy of Management Review, 14, 361–384. Zedeck, S. (1992). Work, Family, and Organizations. San Francisco: Jossey-Bass. Zeidner, M., Boekaerts, M., & Pintrich, P. R. (2000). Self-regulation: Directions and challenges for future research. In: M. Boekaerts, P. R. Pintrich & M. Zeidner (Eds), Handbook of Self-Regulation (pp. 750–768). San Diego: Academic Press.

RESEARCH ON EMPLOYEE CREATIVITY: A CRITICAL REVIEW AND DIRECTIONS FOR FUTURE RESEARCH
Jing Zhou and Christina E. Shalley
ABSTRACT
The examination of contextual factors that enhance or stifle employees’ creative performance is a new but rapidly growing research area. Theory and research in this area have focused on antecedents of employee creativity. In this paper, we review and discuss the major theoretical frameworks that have served as conceptual foundations for empirical studies. We then provide a review and critical appraisal of these empirical studies. Based on this review, we propose exciting possibilities for future research directions. Finally, we discuss implications of this body of work for human resource management.

INTRODUCTION
Research on employee creativity, as a sub-area in the field of micro organizational behavior, has had a relatively short history. The foundation for this research began in the late 1980s through the middle of the 1990s. Building on her seminal work on

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 165–217 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22004-1

165

166

JING ZHOU AND CHRISTINA E. SHALLEY

the social psychology of creativity (Amabile, 1983), Amabile (1988) was perhaps the first person to propose a theory-based and empirically supported componential framework to understand the personal (e.g. domain-relevant knowledge and skills, and creativity-relevant skills and strategies) and environmental factors (e.g. contracted – for rewards) that could facilitate or inhibit employee creativity. Shalley (1991) then demonstrated the differential effects of two factors that have substantial implications for creativity in the workplace – productivity goals and creativity goals on creativity. Woodman, Sawyer and Griffin (1993) made the explicit declaration that creativity is affected by the interaction of personal and organizational or environmental factors, and reviewed many individual, group, and organizational factors that might interact to affect creativity. Oldham and Cummings (1996) provided an important empirical test of the associations between personal factors, contextual factors, their interactions, and creativity in the workplace, and successfully demonstrated the existence of these associations. Although the history of focused scholarly research on creativity in the workplace is relatively short, during the past decade empirical research in this area has catapulted. Researchers have investigated effects of a variety of contextual or organizational factors, individual differences, as well as the interactions of contextual factors and individual differences on creativity (e.g. Amabile, 1988, 1996; Amabile & Conti, 1999; Amabile, Conti, Coon, Lazenby & Herron, 1996; Eisenberger & Armeli, 1997; Eisenberger & Selbst, 1994; George & Zhou, 2001, 2002; Madjar, Oldham & Pratt, 2002; Oldham & Cummings, 1996; Perry-Smith & Shalley, 2003; Shalley, 1991, 1995; Shalley & Oldham, 1997; Shalley & Perry-Smith, 2001; Tierney & Farmer, 2002; Tierney, Farmer & Graen, 1999; Zhou, 1998a, 2003; Zhou & George, 2001; Zhou & Oldham, 2001). While this body of work has made significant contributions to our knowledge of what factors enhance or impair creative activities in the workplace, substantial challenges, unanswered questions, and exciting research opportunities remain ahead of us. The purpose of this paper is to critically review research on employee creativity, and propose avenues for future work. The paper unfolds as follows. We first review: (a) existing major theoretical frameworks; (b) dominant research methods that have been used to research individuals’ creative performance; and (c) results of major empirical studies. Next, we propose several directions for future research. Finally, we discuss implications of this body of research for human resource management. We emphasize extant theories and empirical research that promote an understanding of employee creativity in the workplace. We also emphasize empirical studies that were either conducted in organizational settings or provide clear implications for an understanding of creativity in the workplace.

Research on Employee Creativity

167

THEORETICAL FRAMEWORKS
Creativity is defined as the production of new and useful ideas concerning products, services, processes and procedures (e.g. Amabile, 1996; Oldham & Cummings, 1996; Shalley, 1991; Zhou, 1998a). This definition can include creative solutions to business problems, creative business strategies, or creative changes in job processes. Currently, there are two influential models concerning creativity in the workplace (Amabile, 1983, 1988, 1996; Woodman et al., 1993). While there are similarities and differences between these models, together they complement each other and provide a solid framework that guides research in the area of employee creativity. Notably, in addition to individual characteristics, both models discuss effects of the work environment or organizational context, describing a variety of potentially relevant contextual factors that either enhance or restrict employee creativity. Each of these models is briefly reviewed below. Additionally, we briefly discuss some related conceptual work on creativity that has been developed.

The Componential Model of Creativity Amabile (1988) published a chapter proposing a componential theory of creativity in the workplace, which was partially based on the componential model of a social psychology of creativity she formulated and tested earlier (Amabile, 1983). It represented one of the first comprehensive and grounded theories of employee creativity. This theory has been further tested, revised and updated since then (e.g. Amabile, 1996). Essentially, the theory posits that there are three key components of creativity or creative performance: domain-relevant skills, creativity-relevant processes, and task motivation. According to Amabile, the first component, domain-relevant skills, refers to factual knowledge and expertise in a given domain. Amabile has argued that domain-relevant skills tend to be affected by formal and informal education, and individuals’ perceptual, cognitive, and motor abilities. The second component, which was initially called creativity-relevant skills but recently has been changed to creativity-relevant processes, includes explicit or tacit knowledge concerning the appropriate strategies for producing creative ideas, appropriate cognitive styles and work styles for creative idea production. Amabile reasoned that training in creative skills and strategies, experiences in creative activities, and possessing certain personality characteristics are likely to positively influence creativity-relevant processes. Subsequent empirical research on training for creative problem solving has indicated that training can help enhance employee’s level of creativity (e.g. Basudur, Wakabayashi & Graen, 1990).

168

JING ZHOU AND CHRISTINA E. SHALLEY

The third component, task motivation, includes individuals’ attitudes toward a task and their perceptions of his or her own motivation for working on the task. In general, an individual’s motivation can be intrinsic or extrinsic in nature. Amabile proposed that intrinsic motivation should be defined as “any motivation that arises from the individual’s positive reaction to qualities of the task itself; this reaction can be experienced as interest, involvement, curiosity, satisfaction, or positive challenge” (1996, p. 115). On the other hand, extrinsic motivation can be defined as “any motivation that arises from sources outside of the task itself.” This model premised that intrinsic motivation rather than extrinsic motivation was critical for creativity to occur, especially at the stage of discovering or defining a problem for which creative ideas or solutions need to be produced, and at the stage of actually coming up with creative ideas or solutions. There are at least two significant implications of the fact that task motivation is featured prominently in the componential model. First, it makes a powerful and convincing statement that individuals with great creative potential (e.g. domain-relevant knowledge and skills) may not actually produce creative ideas; they need to be willing to engage in creative activities in a intense and persistent manner. This emphasis on motivation represents a significant advancement from prior theorizing of the determinants of individual creativity (see Barron & Harrington, 1981 for a review of personality factors and creativity). Second, it sets the stage for researchers to identify contextual factors that enhance or constrain creative performance via promoting or diminishing intrinsic motivation. As such, the componential model of creativity is often considered an intrinsic motivation perspective of creativity. Much of the prior research using the intrinsic motivation perspective has identified creativity-relevant contextual factors that have been theorized to affect intrinsic motivation and subsequently creativity, and they have been premised on cognitive evaluation theory (Deci & Ryan, 1980, 1985). Essentially, this line of research has argued that specific contextual factors positively or negatively influence individuals’ intrinsic motivation, which, in turn, influence an individual’s creativity. Thus, it is necessary to outline the key elements of cognitive evaluation theory here. According to cognitive evaluation theory, individuals will experience high levels of intrinsic motivation toward a task when they feel competent and self-determining on a given task. All contextual factors or conditions have two potential functions: informational or controlling. The relative salience of these two functions determines whether a contextual factor will have a positive or negative effect on individuals’ levels of intrinsic motivation. When the informational aspect is salient, individuals perceive that they are being supported and encouraged to take initiatives and to try new things, with little external pressure to achieve certain things in prescribed ways. As a result, their intrinsic motivation tends to be maintained or enhanced and they are likely to exhibit high levels of creativity.

Research on Employee Creativity

169

In contrast, when the controlling aspect is salient, individuals perceive that their thoughts, feelings, and actions are being constrained toward doing specific things in certain ways. Thus, they feel that they are no longer the origin of their own actions. Rather, they sense that they are under the tight control of external forces. As a result, their intrinsic motivation tends to diminish, and they would be expected to exhibit relatively low levels of creativity. Cognitive evaluation theory has guided much of the recent research on creative performance (e.g. Shalley, 1995; Shalley & Perry-Smith, 2001; Zhou, 1998a; Zhou & Oldham, 2001). However, even though these models and empirical studies discuss the importance of intrinsic motivation as the psychological process that accounts for employee creativity, little research has unequivocally demonstrated that intrinsic motivation mediates effects of contextual factors on creativity. In fact, recently the importance of intrinsic motivation rather than motivation in general or some other underlying mechanisms, such as focus of attention and affect, has been questioned (Shalley & Perry-Smith, 2001). This issue will be more fully discussed later in the paper.

An Interactionist Approach to Organizational Creativity Woodman, Sawyer and Griffin’s (1993) interactionist perspective of organizational creativity is premised on the idea that creativity is an individual level phenomenon that can be affected by both dispositional and situational variables. In fact, it is the interaction of individual’s disposition and contextual factors that more fully predicts creative performance. While Amabile’s and the Woodman et al.’s models both stress the role of contextual factors at the individual, group and organizational levels, Woodman and colleagues’ model explicitly stresses the importance of the interaction between the person and the situation, and is based on the strong theoretical base of interactional psychology (Schneider, 1983; Terborg, 1981). Finally, an important part of this model is its propositions concerning influences across levels of analysis. These researchers argued that the cross-level influences are essential in identifying and understanding group and organizational characteristics that can enhance or constrain creative behavior in a complex social system. According to Woodman and colleagues (1993), creative performance in organizations is a function of a host of individual, group and organizational characteristics that interact to enhance or constrain creativity. They called for a systematic investigation of the social and contextual influences at all levels that affect creativity. Specifically, for individual characteristics they discuss how cognitive abilities or style, personality, intrinsic motivation and knowledge are important. The group characteristics that they discuss include norms, cohesiveness,

170

JING ZHOU AND CHRISTINA E. SHALLEY

size, diversity, roles, task and problem solving approaches. Finally, organizational characteristics such as culture, resources, rewards, strategy, structure and technology are highlighted. In the model they propose that creative persons, groups, and organization are inputs that are transformed in some way by the creative process and the creative situation, which includes enhancers and constraints for creative activities. The potential outcome of this transformation of the inputs is a creative product.

Related Conceptual Work on Creativity Ford (1996) proposed a theory of individual creative action in multiple social domains, such as, group, organizational, institutional, and market domains. He argued that creative action by an individual is a result of the joint influence of sensemaking, motivation, and knowledge and ability. Ford proposed that creative and habitual actions are competing behavioral options for an individual. He argued that as long as habitual actions remain more attractive, even when the context may be favorable for creative action, an individual would tend to choose the habitual action. Therefore, creative actions, though extremely important, will remain as more uncommon instances in organizations. Drazin, Glynn and Kazanjian (1999) also proposed a multi-level model of organizational creativity, in which they defined creativity not using the more typical outcome based definition, but as an individual’s psychological engagement in creative activity regardless of whether the outcomes are creative or not. Their process definition of creativity focuses on individuals attempting creative activities. They proposed that individuals form frames of reference that mediate their engagement in creative activities. Their model is focused more on larger scale organizational projects that run for long durations, and they suggest several ways to test their ideas. Three very recent conceptual pieces take a different approach than presenting multi-level models. In the first piece, Mainemelis (2001) proposed a model that describes the way individuals experience timelessness by being totally engaged in their work activities. This model looks at contextual conditions that facilitate or impair this engrossment process, and the effect of timelessness on the creativity of employees. He suggests that timelessness is composed of four experiences: a feeling of immersion, a recognition of time distortion, a sense of mastery, and a sense of transcendence. Mainemelis’s approach is highly focused on the experience of working on a specific activity and the resulting task related affective processes that may increase the likelihood of creativity in organizations. In another recent conceptual piece, Unsworth (2001) questioned whether creativity was really a unitary construct, as much of the literature has treated it. Essentially, she argued that the common definition used for creativity (e.g. novel

Research on Employee Creativity

171

ideas that are appropriate for the situation) implies that creativity is really only one construct without looking at the type of idea, why it was generated, or how the process began. She called for a more fine-grained analysis of the processes involved in creativity. Unsworth developed a matrix of four creativity types that varied on the following two dimensions: what was the driver for the engagement (external or internal), and what was the problem type (open or closed). Open ideas are those ideas that are discovered by the individual, while closed ideas are presented to the individual. The four creativity types are: responsive (closed, external), expected (open, external), contributory (closed, internal), and proactive (open, internal). The third and most recent conceptual piece (Perry-Smith & Shalley, 2003) highlights the importance of others for generating creative ideas by focusing more on the social side of creativity. Using concepts from social network theory, Perry-Smith and Shalley (2003) explored the association between the context of social relationships and individual creativity from a more macro perspective. Taken together, their propositions can be summarized by describing an individual’s journey from the fringe of the network to the center. They argued that weaker ties are generally more beneficial than stronger ties for creativity. Specifically, they proposed that a peripheral position with many connections outside of the network is likely to be associated with more creative insights and potentially more groundbreaking advancements. However, once this high level of creativity has been achieved, they suggest that the peripheral individual would start becoming relatively more central. The exposure to diverse people and information brought about by this centrality will spark new ideas and aid in the generation of additional creative insights. This reciprocal, spiraling process of increased centrality and increasing creativity will continue until a point of diminishing returns. At this point, the person may become too entrenched or immersed in the status quo, ultimately, constraining their creativity unless they maintain connections outside of their social field. While the ideas presented in each of these conceptual pieces are intriguing and merit further exploration, as of yet, little empirical work has specifically examined the issues laid out by these researchers. This is not surprising given that each of the articles are relatively recent, we would expect to see more work based on these frameworks in the future.

RESEARCH METHODS
Definitional Issues As mentioned earlier, creativity refers to the production of new and useful ideas concerning products, services, processes and procedures (e.g. Amabile, 1996;

172

JING ZHOU AND CHRISTINA E. SHALLEY

Oldham & Cummings, 1996; Shalley, 1991; Zhou, 1998a). Operationally, especially in studies conducted in the behavioral laboratory, creativity is defined as “the production of responses or works that are reliably assessed as creative by appropriate judges” (Amabile, 1996, p. 83). Appropriate judges are usually experts or those with advanced knowledge within the domain studied. Essentially, because creativity is considered by many to be historically, culturally, and socially bound (cf., Amabile, 1996), it is important to obtain assessment from multiple, appropriate judges and these judges’ assessments need to reach an acceptable level of agreement concerning the extent to which an idea or product is creative. Furthermore, creativity is generally considered to be a continuous concept with a focus on how relatively creative something is, rather than a discrete decision that something is or is not creative (Amabile, 1996; Perry-Smith & Shalley, 2003; Shalley, 1995). Therefore, using both continua of the spectrum, a creative solution may be a minor adaptation or a major breakthrough. For instance, Mumford and Gustafson (1988) discuss the differences between minor and major contributions, with minor contributions reflecting adjustments, recombinations and extensions of existing ideas whereas major contributions are more groundbreaking and radical in their approach. Thus, the entire spectrum of creativity can be important because at times organizations may desire more incremental creative solutions, while at other times it may be more desirable to have employees attempt breakthroughs considered more monumental.

Measurement of Creativity The Consensual Assessment Technique The majority of research in this area has used Amabile’s consensual assessment technique, which is derived from her operational definition of creativity described above (Amabile, 1983, 1996). Typically, two or more raters serve as expert judges. The judges normally have relevant educational degrees and a number of years of relevant work experience. They are usually asked to independently rate the overall creativity of each solution or product on a scale ranging from not at all creative to extremely creative. Some studies have used a 7-point scale, others have used an 11-point scale for the ratings. As done in Shalley (1991, 1995), Shalley and Perry-Smith (2001), Zhou (1998a), and Zhou and Oldham (2001), if the interrater reliability for the judges ratings is acceptable, then a creativity score is computed as an average of the creativity ratings for each individual across the generated solutions. Before computing the creativity score, it is important to check the acceptability of both: (a) inter-judge reliability (the extent to which the ratings were consistent)

Research on Employee Creativity

173

using an intraclass correlation approach (Shrout & Fleiss, 1979); and (b) inter-judge agreement (the extent to which the judges assigned the same rating to each solution) (Tinsley & Weiss, 1975). In earlier creativity research, researchers often checked interrater reliability by using Spearman-Brown coefficients and Cronbach’s coefficient alpha. Increasingly, as the field becomes more sophisticated, so does the variety of techniques used. For example, in addition to the aforementioned formulas proposed by Shrout and Fleiss (1979) and Tinsley and Weiss (1975), another alternative is to calculate the rwg (James, Demaree & Wolf, 1984). Importantly, each of these methods for calculating interrater reliability and agreement has its assumptions. Researchers need to select a method whose assumptions and applicability are most suitable to the research design and nature of the data. Most of the studies conducted have taken an overall creativity rating approach in which raters were first given the definition of creativity as including both originality or novelty and usefulness, and then they rated the extent to which each idea or solution generated by each participant is creative. After checking the reliability and agreement of the judges’ ratings, a creativity score is computed for each participant by taking the average of the overall creativity ratings across all ideas generated and averaging across the judges’ ratings as well. A somewhat different approach was taken in Zhou and Oldham (2001), where two judges first separately evaluated both the originality and usefulness of the solutions to Shalley’s (1991) memos. Once the judges’ originality and usefulness ratings were deemed reliable and in agreement, an originality score and a usefulness score were created for each participant by taking averages of originality and usefulness ratings, respectively. Next, a creativity index was formed for each participant by multiplying the originality and usefulness scores for a composite measure. Which of the above reviewed two approaches is better remains an empirical question. Furthermore, if originality and usefulness are rated separately and then combined into a composite measure, whether it would be best to multiply, add, or differentially weight each sub-factor in some way is unclear and needs to be explored. For the time being, both the overall creativity rating approach (e.g. Shalley, 1991; Shalley & Perry-Smith, 2001; Zhou, 1998a) and the multiplicative approach used in Zhou and Oldham (2001) appear to be consistent with the definition of creativity (i.e. both originality and usefulness are necessary conditions for an idea to be judged as creative). Supervisor Ratings Having supervisors rate the creativity of their employees is most commonly used in field studies (George & Zhou, 2001, 2002; Oldham & Cummings, 1996; Scott & Bruce, 1994; Shin & Zhou, in press; Tierney et al., 1999; Zhou, 2003; Zhou &

174

JING ZHOU AND CHRISTINA E. SHALLEY

George, 2001). Currently in the literature, we have identified four different rating scales that have been used to have supervisors rate the level of creativity of their employees. These four scales are: (a) George and Zhou’s 13-item scale (3 of the 13 items were adapted from Scott & Bruce, 1994) (George & Zhou, 2001, 2002; Shin & Zhou, in press; Zhou, 2003; Zhou & George, 2001); (b) Oldham and Cummings’ (1996) 3-item scale; (c) Scott and Bruce’s (1994) 6-item scale; and (d) Tierney and colleagues’ 9-item scale (Tierney et al., 1999, 4 of these nine items were adapted from Ettlie and O’Keefe, 1982. Tierney and Farmer, 2002, used six items of the Tierney et al., 9-item scale). In looking at their items, many of these scales share similarities and yet they have different emphases in notable ways. In particular, some scales may be more appropriate for studying certain populations (e.g. R&D employees) than others (e.g. employees whose jobs do not revolve around creativity). In addition, some scales tend to emphasize innovation (i.e. the implementation of creative ideas) instead of producing creative ideas and approaches per se, whereas other scales tend to be more balanced with regard to producing and implementing ideas. Systematic investigations are needed to examine the extent to which these four scales converge, and whether different scales are more suitable for studies with different jobs, types of industries and employees, and at different stages of the creative process. Objective Measures Some field studies have also used objective measures such as the number of patents, patent disclosures, research papers and technical reports, and ideas submitted to employee suggestion programs (e.g. Oldham & Cummings, 1996; Pelz & Andrews, 1966; Scott & Bruce, 1994; Tierney et al., 1999). Articles reporting these studies usually have included some discussion of whether these objective measures are the same as creativity. Obviously, whether to use these objective measures depends on whether it makes sense in studying different populations. For example, number of patents or patent disclosures, research papers and technical reports may be relevant indicators of creative performance for employees working in research and development. However, they are not relevant indicators of creativity for employees working on jobs that potentially involve some level of creativity (e.g. nursing) but do not normally involve producing patents and research papers. In addition, although many studies have found similar results between objective measures and supervisory ratings of employee creativity, some studies have found varying results, and still other studies have found varying results between different objective measures. For instance, Scott and Bruce (1994) reported that supervisor ratings were significantly correlated with number of invention disclosures. Likewise, Tierney et al. (1999) found that supervisor ratings were significantly correlated with invention disclosures and research reports, and that patterns of

Research on Employee Creativity

175

results were similar across these three subjective and objective measures. However, Oldham and Cummings (1996) found different results for patent disclosures written and contributions to company’s suggestions program, and yet results involving patents and supervisor rated creativity were similar. It is possible that for certain jobs and in certain organizations, different measures converge in tapping creativity. It is also possible that for other kinds of jobs and in other organizations, different measures are tapping different types or dimensions of creativity. More research is needed to examine utilities of subjective versus objective measures of creativity, and among different objective measures of creativity.

Research Design Laboratory Studies In the laboratory studies on creativity, an important issue has been the task that participants work on. Essentially, it has been argued that to allow creativity to manifest, the type of task that research participants work on is critical. More specifically, participants should be working on a complex, open-ended task where a heuristic rather than algorithmic solution or outcome is possible. Complexheuristic tasks are open-ended and ill-structured, and lack a clear, straightforward path to a solution (Amabile, 1983; McGraw, 1978). Thus, it is necessary to discuss the tasks used in these studies. The tasks that are particularly relevant given the purposes of this paper include the tasks developed and used by Amabile and her colleagues (1996), the task developed by Shalley (1991), and the task developed by Torrance (1974). Over the years, Amabile and colleagues have developed many tasks used in their programmatic research on creativity. Amabile (1996) grouped these tasks into three categories: artistic (e.g. collages and paintings), problem solving (e.g. ideas for high-tech products), and verbal (e.g. poems and stories) tasks. Notably, Amabile (1996) reported that these tasks had been used with a variety of populations (e.g. school children, artists, and undergraduate or MBA students), and yielded satisfactory interrater reliabilities when the researchers used the consensual assessment technique. Shalley’s (1991) memo task is similar in many ways to an in-basket exercise in that research participants are asked to read through a number of memos and respond to the issues presented in the memos. However, the typical in-basket exercise is also focused on being able to get the memo off one’s desk quickly, or to decide to delegate the issue or respond to it in the future. In Shalley’s task, the participants are encouraged to respond to each memo presented in a thoughtful, complete manner and not to worry about completing all the memos

176

JING ZHOU AND CHRISTINA E. SHALLEY

provided. The task involves responding to a series of problems presented to the personnel director of a large steel manufacturing company. The participants are given information concerning their role as the personnel director, background on the company, and a packet of memos that deal with a variety of management related issues (e.g. incidents of potential sexual harassment, employee theft, and productivity issues). Individuals are normally told to try to craft a response to deal with the issue presented in each of the memos they are given. The original task included a set of twenty-two different memos (Shalley, 1991) so that there was a range of different types of issues that participants could respond to and also so that participants could be assigned productivity goals that varied in their difficulty level. Since most of the subsequent studies using this task have not been concerned with the use of productivity goals, typically the studies have used a sub-set of the original twenty-two memos since it is extremely difficult for any individual to respond to all twenty-two issues in the time usually allotted for the task. Torrance’s (1974) unusual uses task involves asking research participants to generate unusual uses for a variety of different objects. The objects commonly used include a brick, newspaper, tire, junk auto, and hanger. On the basis of Guilford’s (1956, 1967) theory concerning components of creativity and divergent thinking, the participants’ creativity is assessed by considering four dimensions: fluency (i.e. the total number of new uses generated for the five objects), flexibility (i.e. the extent to which each participant used different categories of ideas or approaches to a problem), elaboration (i.e. the extent to which the ideas are well developed), and originality (i.e. the extent to which the ideas are unusual or statistically infrequent). Many prior studies used variations of the Guilford and Torrance unusual uses task to investigate correlates between individual differences and creativity, and the task is considered to be appropriate for research on contextual influences of creativity. However, Amabile (1996) cautioned that the task is somewhat removed from real-world scenarios of creative performance and its scoring may represent quite narrow definitions of creative achievements (e.g. verbal fluency). Thus far, we have reviewed three commonly used tasks in laboratory research on creativity. Laboratory studies have a number of potential strengths. In particular, with laboratory studies researchers are more likely to be able to establish causality, and by their nature they have internal validity. With laboratory studies the researchers can also isolate different constructs and cleanly manipulate the constructs of interest while controlling conditions that are not the focus of the study, and instead may just represent noise. These strengths are particularly attractive for creativity research, which is a relatively new topic compared with many other topics in the field of organizational behavior and human resource management.

Research on Employee Creativity

177

However, as with any research method, laboratory studies have potential limitations. A nature of the beast with laboratory studies is the issue of external validity. While many of these studies have tried to select contextual factors to examine that are important for employees in organizations and have tried to structure the laboratory environment so that it is as close as feasible to a typical work environment, the laboratory is not a real work environment and participants are often not full time employees. As such, while laboratory studies have inherent strengths in terms of control over the situation, cleaner manipulation of constructs and getting rid of noise, they do err on the side of less external validity, potentially making their results not be as generalizable to an organization as a typical field study. As always, selecting any single research method involves balancing trade-offs, and by the use of triangulation of methods, consistent results across methods can be gained to inform the field. Field Studies To date, most field studies on employee creativity have used quantitative data collected by questionnaire measures and, when available, archival data (e.g. number of patents obtained) collected from participating companies’ records. And, not different from many field studies in organizational behavior and human resource management, many field studies on creativity are primarily cross-sectional, with measures of independent variables and dependent variables collected in close proximity of each other. While prior field studies have been largely quantitative in nature, qualitative case studies may compliment quantitative studies to provide fuller and richer descriptions on how creative performance unfolds in the organizational context. There are abundant anecdotes about creative individuals, however, well-executed and rigorous qualitative studies on creativity have been rare. An exception is a study by Hargadon and Sutton (1997). In an ethnographic study that lasted approximately two and a half years, the authors conducted observations, interviews, informal conversations and gathered archival data at a product design firm. On the basis of the ethnography, the authors developed a technology brokering model of product innovation. This model describes how the organization acted as a technology broker, using its network position of a product design consulting firm working with many clients in at least forty industries to broker, transfer and adapt existing knowledge from one industry to another or combining preexisting ideas, products and knowledge into an original product. This qualitative study provides implications for theory development and research on how organizations might use its external network positions and develop internal processes to gather, store, and retrieve knowledge in order to routinely develop innovative products. As creativity is a new and rapidly growing research

178

JING ZHOU AND CHRISTINA E. SHALLEY

area, qualitative studies might be particularly suitable for grounded theory building: Such studies have the potential to provide framebreaking, convincing, and testable insights, which are especially needed in newer research areas (Eisenhardt, 1989). There are a number of strengths of conducting field studies to investigate creativity. In particular, field studies use employees in their work organizations, and they have the potential to be more generalizable and have greater external validity compared with studies conducted in the behavioral laboratory. However, conducting a study in an organizational setting in itself does not guarantee greater internal and external validity, and hence is not necessarily a superior method relative to lab studies. This is because while field studies have strengths, at the same time they also have several potential limitations. For example, as mentioned above, because of resource and access constraints, to date field studies on creativity are largely cross-sectional in nature. Hence, they cannot unequivocally determine direction of causality, even if hypotheses derived from theory receive empirical support. In addition, in field studies researchers often cannot definitively control extraneous factors and noise. While carrying out longitudinal studies might lessen some of the limitations associated with field studies, they themselves pose tremendous challenges to researchers. In addition to the typical challenges in all longitudinal field studies such as cost and participant attrition, there are unique challenges for creativity research. For example, creative endeavors in organizations may unfold over an extended period of time. In addition, unless these endeavors are R&D projects that have a clear starting point and ending point, many creative activities initiated by employees with non R&D jobs may not have clear-cut beginning and ending points. Moreover, different people may use different strategies and go through different cognitive and behavioral processes to produce creative outputs, which makes continual tracking of creativity-related behaviors a real challenge for researchers. These and other challenges notwithstanding, a longitudinal study reported by Amabile and Conti (1999) provided an example on how some of the challenges might be addressed. More specifically, they investigated creativity enhancing or inhibiting work environments before, during, and after a major downsizing. They were not able to follow the same group of employees to collect all waves of data. Instead, they randomly sampled a different group of employees for each wave of data collection. Finally, while qualitative case studies can provide rich information compared with typical quantitative survey studies, and they are particularly useful for theory building, they do have the inherent limitation of not being generalizable to other organizations and situations.

Research on Employee Creativity

179

A REVIEW OF EMPIRICAL STUDIES
This review is divided into two parts: the first part focuses on contextual factors as antecedents of creativity, and the second part focuses on individual differences as antecedents or moderators of creativity.

Antecedents of Creativity – Contextual Factors Goals and Expectations for Creative Activity One potential way that managers may be able to influence the occurrence of creative activity is through goal setting. Goals influence motivation through their impact on self-regulatory mechanisms (Kanfer & Ackerman, 1989). Research on goal setting has indicated that it is an incredibly effective motivational technique (Locke & Latham, 1990). The basic motivational assumption of goal setting is that goals increase attention and effort by providing clear targets toward which individuals can direct their energies. Goals serve to motivate and direct attention to important facets of a task or project and facilitate information acquisition. Goals regulate action directly by affecting what people pay attention to, how hard they work, and how long they persist on a task. Goals affect action indirectly by motivating people to discover and use task strategies that will facilitate goal achievement. Goals are more likely to be attained when people are strongly committed to the goals and are given feedback concerning their progress in relation to their goals. A robust finding across numerous studies concerning the relation between goal setting and performance has been that setting difficult and specific goals improves productivity (Locke, Shaw, Saari & Latham, 1981). Presumably, a difficult and specific goal can enhance performance because it would direct individuals’ attention and energy toward achieving certain aspects or levels of performance. Interestingly, however, studies that demonstrated this performance-enhancing effect of goal-setting usually focused on the sheer productivity aspect (i.e. the number of task units completed) of individual task performance. Because individuals have limited cognitive resources and energy, it is possible that goals that direct individuals’ attention toward completing more task units would simultaneously divert their attention away from coming up with creative ideas about their work. However, it is also possible that a goal could either help or hinder task engagement depending on whether the goal focuses individuals on aspects of the task that may facilitate or stifle creative performance. For instance, since goals are directive they could focus individuals on aspects of the task that would facilitate their performance. On the other hand, it is possible that the goal could inhibit engagement by narrowing an individual’s focus of attention and

180

JING ZHOU AND CHRISTINA E. SHALLEY

information acquisition to particular aspects of the task that may not facilitate creative performance. In addition, a goal may be perceived as an extrinsic constraint, and thereby, reduce individuals’ intrinsic motivation to perform a task (Shalley & Oldham, 1985). In other words, while productivity goals can enhance productivity, they may have neutral or even negative effects on creativity. Furthermore, it also is possible that goals could be set for individuals to be creative. These ideas are consistent with the distinction between performance- and mastery-oriented goals (Dweck & Leggett, 1988; Elliott & Dweck, 1988). Individuals motivated by a mastery goal orientation focus on task mastery and skill improvement. In contrast, individuals motivated by a performance orientation focus on the assessment of the level of their abilities relative to others. Thus, a performance orientation directs individuals’ attention to themselves, away from concentrating on the task, and makes the individuals feel pressured and anxious to show their competence. The lack of task interest and feelings of anxiety often result in poor performance on complex tasks (Utman, 1997). Shalley (1991) conducted a study to empirically test some of the issues discussed above. She examined the effects of two types of goals (i.e. a creativity goal and a productivity goal) on individuals’ creative performance. More specifically, participants were assigned one of three productivity goals (difficult, do your best, or no goal) and one of three creativity goals (difficult, do your best, or no goal) while they worked on the memo task described earlier in this paper. Participants assigned a difficult productivity goal were asked to generate solutions for at least 14 of 22 memos presented to them, a do your best goal was to generate solutions for as many memos as possible, and those with no productivity goal were told to work at their own speed. As for the creativity goals, participants assigned a difficult creativity goal were told that 90% of the solution generated needed to be creative, those with a do your best creativity goal were told to try to generate highly creative solutions, and those with no creativity goal heard no mention of creativity. An interactive effect for goal type on creative performance was found. Specifically, creativity was low if individuals were given a do-your-best or difficult productivity goal without also assigning a creativity goal. However, individuals exhibited high levels of both creativity and productivity if they were assigned a do-your-best or difficult creativity goal and a difficult productivity goal. Shalley (1991) interpreted these findings as indicating that once a creativity goal was assigned, individuals were primed to focus their attention and effort on this objective, and therefore, exhibited overall higher levels of creativity than individuals who were not assigned a creativity goal. Furthermore, in the absence of a creativity goal, a productivity goal appeared to cause individuals to narrow their attention to goal-relevant stimuli alone (e.g. being productive), and to ignore subtle aspects of the task that could have been helpful in their being creative. Therefore,

Research on Employee Creativity

181

in this case, productivity goals served as attentional controls, causing performance on other dimensions not addressed by the goal to suffer while the presence of a creativity goal served to boost individuals’ level of creativity on the task. In a similar study as above, Carson and Carson (1993) found that students who were assigned only a creativity goal or a creativity goal along with a quantity goal performed more creatively on a verbal fluency task (i.e. Torrance’s task described earlier) than individuals assigned a quantity goal only. Furthermore, in another study, Shalley (1995, Study 2) found that assigning a creativity goal had a significant, positive effect on individuals’ creativity. Additionally, in this study, individuals assigned a creativity goal had lower productivity than those with no creativity goal, presumably because they were focusing their attention and effort on creativity rather than productivity. Therefore, in general, the presence of a creativity goal appears to enhance creativity. However, it may do so at the expense of other aspects of performance such as productivity, as measured by sheer quantity of output (Shalley, 1995, Study 2). In summary, creativity goals seem to serve as a mechanism for inducing individuals to exhibit certain behaviors, perform at a desired level, and also as a standard against which task behavior can be evaluated. A number of studies also have found that when individuals know that creativity is important, but an explicit goal has not been set, they are more likely to actually be creative. For instance, Speller and Schumacher (1975) found that individuals’ scores on creativity tests improve if they are told that they are taking a creativity test. Additionally, it has been found that when individuals were given test instructions to “be creative” they were significantly more creative on a divergent thinking test than individuals given instructions that did not emphasize a desire for creativity (Manske & Davis, 1968). The results of these studies could be explained by the fact that the mention of a creativity test or instructions to be creative serves as a goal, priming attention and effort on trying to be creative. Priming is a process in which particular information is activated for recall. Thus, it would be expected that the notion that creativity is expected or even required in order to perform a job effectively should facilitate creative performance by causing individuals to focus all of their attention and effort on generating novel, appropriate responses. Research has indicated that employees can accurately identify when creative activity is required by their job. For instance, Shalley, Gilson and Blum (2000) found that at least 75% of their sample of 2,870 working individuals (i.e. 2,200 people) accurately reported the level of creativity that was required in order to perform their job when their responses were compared to the creativity rating for those same jobs by the U.S. Department of Labor in their Dictionary of Occupational Titles (DOT). The researchers speculated that for the remaining quarter of the sample, there may have been some misperceptions, but also factors such as the

182

JING ZHOU AND CHRISTINA E. SHALLEY

job incumbent, organizational culture, and manager’s expectations that may have caused an individual’s job to have a higher or lower level of required creativity than the matching DOT job. Recently, Shalley (in press) asserted that if managers would like their employees to be more creative then they need to explicitly or implicitly stress the importance of role requirements for creative performance. Therefore, creating role expectations by assigning goals for creative performance, making creative activity a job requirement, and developing a culture that is supportive of creative activity, is expected to lead to higher levels of creative performance. The Effect of Evaluation and Feedback on Creativity Since performance evaluation is an integral part of any job, it is important to examine its effect on various aspects of performance, including creativity. In the area of employee creativity, there has been a relatively large amount of research focused on different facets of performance evaluation for employees’ creativity. These studies range from examining the expectation of evaluation, to the provision of feedback, to the type of feedback actually given (e.g. developmental assessments). Each of these areas will be discussed in detail below. Expected evaluation. Early research on the effects of expected evaluation for intrinsic motivation and creativity had inconsistent results. On one hand, some research suggested that evaluation can have dysfunctional effects on intrinsic motivation and creativity (e.g. Amabile, 1979; Amabile, Goldfarb & Brackfield, 1990; Shalley & Oldham, 1985). For example, Amabile (1979) found that individuals who expected his or her task performance to be evaluated exhibited lower intrinsic motivation than those who did not expect an evaluation. This result may have occurred because expected evaluation undermined intrinsic motivation by focusing the participants on extrinsic reasons for engaging in the task (e.g. receiving a positive evaluation) rather than an intrinsic reason such as enjoyment of the task. Furthermore, in this study she found that individuals expecting an evaluation had significantly lower levels of creativity on an artistic task than individuals not expecting an evaluation. On the other hand, other studies have found that evaluation can positively affect intrinsic motivation and creativity (e.g. Harackeiwicz & Elliott, 1993; Jussim, Soffin, Brown, Ley & Kohlhepp, 1992). For instance, Shalley (1995) conducted two studies looking at the effect of expected evaluation on creativity. The first found no significant effect for expected evaluation (Shalley, 1995, Study 1). The second study (Shalley, 1995, Study 2) found that individuals who worked alone, had a creativity goal, and expected an external evaluation demonstrated the highest level of creativity. Taken together, the results of these two studies indicate that expected evaluation is not necessarily harmful to people’s creativity and can actually be beneficial to creativity in certain situations.

Research on Employee Creativity

183

Shalley (1995) suggested that cognitive evaluation theory could help in explaining the unexpected positive effect of expected evaluation. Specifically, the relative salience of the informational and controlling aspects of expected evaluation may have mediated the effects of evaluation on creativity through its effect on intrinsic motivation. She argued that the informational component of expected evaluation would have been expected to be salient in this study, given the nature of the task and participants. Most participants were business majors who should have had some familiarity with this type of task and thus, would have been interested in receiving an evaluation of their creativity in generating solutions to managerial problems. Thus, they should have perceived expert judgments on how well they performed as primarily informational. Therefore, she argued that their creativity could have been enhanced since they anticipated getting useful information from the evaluation. However, it was not possible to empirically test this argument since intrinsic motivation had not been measured in this study. Revisiting this issue, Shalley and Perry-Smith (2001) specifically examined the differential effects of anticipating an informational versus controlling evaluation on individuals’ intrinsic motivation and creativity. They argued that the actual manner in which evaluation was manipulated in past studies appeared to be different based on the instructions given, which could have caused some evaluations to be perceived as more controlling and some as more informational. For instance, in Shalley’s (1995) study participants were told that experts in the area would perform a detailed review of the content of their solutions and that they would receive a copy of this review. This approach was expected to have focused individuals on expecting an evaluation that could enhance their competency and boost his or her learning. On the other hand, Bartis, Szymanski and Harkins (1988) found that evaluation decreased creativity. Shalley and Perry-Smith (2001) argued that in Bartis and colleagues’ (1988) study, participants were given instructions that were more controlling, in that they should have felt pressured to perform. Therefore, Shalley and Perry-Smith (2001) hypothesized and found that when individuals expect an informational evaluation they have significantly higher levels of creativity than when expecting a controlling evaluation. Perhaps more importantly, while many prior studies relied on cognitive evaluation theory to predict how the informational aspect of a given contextual factor would promote creativity via enhancing intrinsic motivation, whereas the controlling aspect of a given contextual factor would inhibit creativity via reducing intrinsic motivation, studies have not directly tested the mediating effect of intrinsic motivation. In Shalley and Perry Smith (2001), they first directly tested the role of cognitive evaluations theory’s primary intervening cognitive process of locus of causality (or self-determination) in mediating the relationship between

184

JING ZHOU AND CHRISTINA E. SHALLEY

evaluation and intrinsic motivation. Locus of causality is the psychological process that cognitive evaluation theory posits is responsible for the relationship between informational and controlling aspects of the context and individuals’ intrinsic motivation. They found that locus of causality was indeed the mediator linking expected evaluation and intrinsic motivation. Furthermore, in this study they also empirically tested whether creativity is most likely to occur when individuals are intrinsically motivated. Although never tested empirically before, much of the creativity research had been based on this association (e.g. Amabile, 1983). Interestingly, intrinsic motivation was not found to mediate the relationship between evaluation and creativity. The researchers went on to suggest other possible intervening mechanisms in addition to or besides intrinsic motivation that may mediate the effect of external factors on individuals’ level of creativity. For instance, they suggested that focus of attention, affect, domain relevant skills, and creativity relevant skills may play some role in the relationship between external factors and their effects on employee creativity. Feedback valence and style. Zhou (1998a) examined the effect of feedback valence and feedback style on individual’s creativity. Zhou (1998a) defined feedback valence as the positive or negative outcome of the comparison between an individual’s creativity and normative or situational criteria. If the comparison indicates that the individual’s ideas are more creative than the criteria, the feedback valence would be positive. If the comparison shows that the ideas are less creative than the criteria, the feedback valence would be negative. In addition to feedback valence, Zhou (1998a) argued that creativity-related feedback has another distinct component – feedback style. The concept of feedback style is important because people do not necessarily perceive and interpret the feedback they receive in a straightforward manner. Rather, depending on the style in which feedback is delivered, the recipient may or may not choose to listen to and respond to feedback. Zhou’s argument concerning the joint effects of feedback valence and style is based on the intrinsic motivation framework (Amabile, 1996) mentioned earlier in this paper. More specifically, Zhou (1998a) defined feedback style as the manner in which feedback is delivered. Feedback style can be either informational or controlling. Feedback delivered in an informational style is not restrictive or constraining nor does it impose the feedback giver’s will or wishes on the feedback recipient. Instead, when it is used to deliver feedback, the informational style suggests that the feedback recipient is in control of his or her own behaviors and the external feedback provider’s goal is simply to help the feedback recipient to learn and develop his or her creative capabilities and creative performance. Because of the nature of informational feedback, the recipient tends to feel that the feedback

Research on Employee Creativity

185

communicated in this type of style is informative and supportive. As a result, he or she is most likely to respond to the feedback in a positive way. On the other hand, supervisors sometimes intentionally or unintentionally deliver feedback in a controlling style. According to Zhou (1998a), feedback delivered in a controlling style makes external constraints salient by emphasizing certain types of ideas or outcomes that the feedback recipient must obtain or certain levels of creativity the recipient must achieve. While used in delivering feedback, a controlling style emphasizes the demands of an external force to obtain certain behavioral outcomes from the feedback recipient. By making external control salient, the controlling style instills feelings of external causality to the feedback recipient. That is, the recipient is led to feel that his or her behaviors and actions did not originate from him- or herself but from controlling others. Taken together, employees should respond most positively to positive feedback when it is delivered in an informational style. In this condition, employees’ perceived competence and self-determination will be enhanced and supported, and as a result, high intrinsic motivation should be experienced (Deci & Ryan, 1985) and high creativity should be exhibited. However, employees should exhibit relatively moderate creativity when they receive positive feedback delivered in a controlling style, or negative feedback delivered in an informational style. Finally, employees should respond negatively when negative feedback is delivered in a controlling style. In this condition, employees’ perceived competence and self-determination will be diminished, low intrinsic motivation will be experienced, and low levels of creativity will be exhibited. Zhou (1998a) examined these issues in a laboratory study that tested the hypotheses that feedback valence and style would jointly affect individuals’ creative performance. Participants in this study worked on Shalley’s (1991) memo task. Participants were randomly assigned to different feedback valence-by-feedback style conditions: positive feedback and informational style, positive feedback and controlling style, negative feedback and informational style, and negative feedback and controlling style. Results showed that of the four experimental conditions, individuals in the positive feedback and informational style condition exhibited the greatest creativity, whereas individuals in the negative feedback and controlling style condition exhibited the least creativity, thereby supporting the hypotheses. After reading the above discussion, one might be inclined to conclude that in order to promote creativity, we need to give people positive feedback delivered in an informational style. This approach is appropriate if the feedback recipient indeed has come up with ideas that are more creative than a standard. If, however, the recipient has come up with ideas that are less creative, should we still give him or her fake positive feedback for the sake of boosting creativity? Little research has examined this issue in depth. Yet, to the extent that individuals need

186

JING ZHOU AND CHRISTINA E. SHALLEY

to learn standards for creativity and learn appropriate creativity-relevant skills and strategies to produce truly creative work (Amabile, 1996; Shalley & Perry-Smith, 2001; Zhou, 2003), giving fake positive feedback would jeopardize this important learning process and hence should be avoided. When one must give negative feedback, one should do so but deliver it in an informational rather than controlling style (Zhou, 1998a). Developmental feedback. Feedback may also vary on its developmental orientation. Feedback messages that are high on developmental orientation provide employees with helpful and valuable information that enables the employees to learn, develop, and make improvements on the job (Zhou, 2003). Such feedback implies that individuals can constantly become better (e.g. more skilled, more creative, and generally more in control of their work) by learning and improving on the job. Feedback messages that are low on developmental orientation contain little information to help employees learn to be creative and grow on the job. Prior research has shown that developmental feedback, in combination with other personal characteristics and contextual variables, facilitated employees’ creative performance. For example, Zhou and George (2001) found that when office employees at a petroleum drilling equipment company experienced high levels of job dissatisfaction and continuance commitment (i.e. employees are committed to their organizations because of necessity instead of affective attachment to or identification with the organizational values and goals; Allen & Meyer, 1996), the more they received feedback high on developmental orientation from coworkers, the higher their creative performance. The researchers reasoned that when an employee is dissatisfied with the job yet perceives the cost of quitting to be too high or alternatives are not available, he or she may either choose to be passive about the job and only work to meet the minimum standard, or choose to proactively change the unsatisfactory and unattractive work situation into something that is more interesting and attractive. If the employee receives feedback from coworkers that focuses on job improvement, the employee’s attention is likely to be directed toward learning and making improvements on the job, in the process of which he or she may be stimulated to see things from different perspectives and come up with new and useful ways of doing things. In another study with hospital employees (e.g. nurses, pharmacists, and so forth), Zhou (2003) hypothesized that when creative coworkers were present, the more supervisors provided developmental feedback, the greater the likelihood that employees would exhibit creative performance. Using the intrinsic motivation perspective described earlier in this paper, Zhou reasoned that developmental feedback is likely to boost intrinsic motivation. This enhanced interest in the task itself and an orientation toward learning and improvement should allow the

Research on Employee Creativity

187

employees to benefit from the presence of creative coworkers. Under this joint condition, the employees are likely to be immersed in the task and focused on skill mastery (Dweck & Leggett, 1988; Elliott & Dweck, 1988; Utman, 1997), and hence they may find the presence of creative coworkers a particularly welcoming opportunity for them to learn and improve creativity-relevant skills. In addition, their enjoyment of the task and orientation toward learning and improvement enable them to seek challenges, to be persistent, and not to be afraid of making mistakes (Dweck, 1986), which are essential for their being able to truly master and utilize creativity skills and strategies. As such, when supervisors provide developmental feedback and creative coworkers are present, employees are likely to properly acquire and use creativity-relevant skills and strategies to search for new and better ways of doing things, which increases the chance of exhibiting high levels of creativity. Results of the study supported this hypothesis. Expected developmental assessment strategies. Zhou and Oldham (2001) argued that much prior research using cognitive evaluation theory to examine effects of contextual factors or practices on creativity had shown that practices restricted creativity by increasing the salience of the controlling aspect. They noted that, however, less was known about the practices that enhance creativity by increasing the relative salience of the informational aspect. They conducted a laboratory study to examine a practice that might have this effect-providing individuals with the expectation of a non-critical, developmental assessment of their creativityrelevant skills. They also examined the effects of two sources for administering the assessment: other actors and the individual him- or herself. According to Zhou and Oldham (2001), a developmental assessment involves: (a) reviewing or assessing individuals’ task responses and; (b) providing them with a non-critical report of this assessment that is intended to help develop their relevant skills and competencies (Farh, Cannella & Bedeian, 1991; London, Larsen & Thisted, 1999). Thus, different from a traditional evaluation approach in which a person’s creativity was critiqued and contrasted against some standard, a developmental assessment approach emphasizes providing individuals with learning and mastery experiences. The goal of this latter approach is to help the individuals to develop creativity-relevant skills and competencies (Nickerson, 1999). Zhou and Oldham defined other-administered assessment as a method in which individuals expected external actors to assess their task responses. Because it was developmental in nature, the outcome of the assessment was to be used for the purpose of developing the focal individuals creativity-relevant skills and to be made available to the focal individuals only. Alternatively, in a self-administered assessment, individuals expected an opportunity to conduct a personal analysis

188

JING ZHOU AND CHRISTINA E. SHALLEY

of their task responses that would help them in developing creativity skills. The focal individual was in charge of the assessment – no external assessments were to be provided, and other actors are involved only to the extent that they provided assistance that the individual requested for conducting the self-assessment. Taking the intrinsic motivation perspective of creativity, Zhou and Oldham predicted higher creativity in a self-administered developmental assessment condition than in other-administered or control (no developmental assessment) conditions. In the former, the controlling aspect should be relatively nonsalient and the informational aspect should be relatively salient. In the other-administered condition, individuals expected the opportunity to develop their personal competencies, which should contribute to the informational aspect. However, because external actors conducted the assessment, the relative salience of the controlling aspect should also be boosted, counteracting the effects of the informational aspect, and resulting in lower creativity. Finally, in a control condition, no external actors were involved to increase the salience of the controlling aspect. At the same time, individuals in this condition did not expect the opportunity to develop their creativity skills, which reduced the relative salience of the informational aspect and should result in relatively low creativity. In a laboratory study, Zhou and Oldham manipulated the expected developmental assessment variable at three levels (self-administered, other-administered, or control). In addition, they also measured individuals’ creative personality (see the section on individual differences below for more detailed results of this part of the study). Sixty-eight participants worked on the memo task developed by Shalley (1991). Results showed that, as predicted, individuals who expected a self-administered developmental assessment exhibited higher creativity than those who expected either no assessment or an other-administered assessment. In summary, taken together these studies researching different aspects of performance evaluation seem to be clearly indicating that while evaluation of employees work is a necessary part of any job, how an evaluation is given or how it is expected can have a dramatic effect on employees’ creativity. Thus, it is not the presence or absence of an external factor per se that positively or negatively affects creative performance. Rather, the way in which people perceive the contextual factors matters for creativity. In other words, the informational component of the evaluation needs to be stressed and the controlling component should be downplayed or not mentioned regardless of whether we are discussing the expectation of evaluation, the provision of feedback, developmental assessments, and so on. Thus, when the informational component of performance evaluation is salient, it appears to enhance individuals’ creative performance while the salience of the controlling aspect of evaluation should have detrimental effects for employees’ creativity.

Research on Employee Creativity

189

Social Influences on Individual’s Creativity – The Role of Others There have been a series of studies that have focused on the effect of others on individual’s creativity. In general, these studies have focused on the level of creative performance exhibited by focal individuals when the presence of others in some way could have a positive or negative effect on focal individuals’ creative performance. For instance, some studies have looked at the effect of the presence of competitive others while others have looked at the effect of the presence of others working independently and noncompetitively in the same room (i.e. coactors) on individuals’ creativity. Finally, other studies have focused on creative role models and how they may impact individuals’ creative performance. The presence of coactors. There has not been a great deal of research on how working in the presence of others may affect individuals’ overall performance and, in particular, their creativity. The research that has examined this has yielded inconsistent results. For instance, Amabile and her colleagues (1990) examined the effects of coaction and audience surveillance on creativity and found no effect for coaction and a nonsignificant negative effect for surveillance. Conversely, Matlin and Zajonc (1968) found that surveillance had a significant, negative effect on originality, which is one key component of creativity. Given that employees are increasingly working in open offices in which their work stations are in view of others’ work stations (Oldham, 1988), it is important to determine how working in the presence of others, either as coactors or as an audience watching their behavior, affects various performance outcomes. There has not been any recent work on audience surveillance but there has been some work on the effect of coactors. In two laboratory studies discussed earlier in the expected evaluation section, Shalley (1995) examined the effects of coaction (i.e. individuals working at a table with five others working on the same task) versus working alone in a private room on individual creative performance. Based on the distraction/conflict theory of social facilitation (Baron, 1986), the presence of others could serve as a distraction and can either energize people or disrupt their performance, depending on whether their attention is focused on dominant or subordinate responses. The researcher argued that distraction could lead to attentional conflict, which may mediate social facilitation effects by increasing drive and causing attentional overload (Baron, 1986). Attentional overload was expected to cause individuals to restrict his or her cognitive focusing, causing them to attend to cues that are central to the task and to ignore peripheral cues. Attentional overload has also been suggested to cause individuals to use cognitive shortcuts to avoid taxing their attentional cues, such as the use of pre-existing algorithms. In these studies it was hypothesized that attentional overload would be dysfunctional for creativity, given the cognitive activities commonly proposed

190

JING ZHOU AND CHRISTINA E. SHALLEY

to be necessary for creativity. Specifically, individuals need to explore a number of potential solutions before choosing the solution deemed highest in novelty and appropriateness. Thus, individuals must attend to a wide range of environmental stimuli in attempting to activate different knowledge structures. Given this, the presence of coactors was expected to have a social impairment effect on individuals’ creativity performance by disrupting their cognitive processing. Using the memo task described earlier in this paper, Shalley (1995) conducted two studies. In the first study, she found that participants who worked alone produced more creative solutions for the memo problems than those working at a table with five others. However, in the second study, she did not find an independent effect of coaction on creativity. Rather, individuals with a creativity goal who worked alone in a private room under the expectation of evaluation had the highest level of creativity. Therefore, it appears that under certain conditions, it may be better to have individuals work alone rather than in the presence of coactors, but under other conditions coactors may have a neutral or negative effect. Presence of creative role models. The use of role models is common in daily work life and in organizational training modules (Decker & Nathan, 1985; Robinson, 1980). Modeling is believed to help clarify role expectations and enhance employees’ skill acquisition. However, little research has looked at whether the use of modeling has an effect on creativity. There is some related developmental research that has examined the presence of role models in a renowned individuals life. For instance, Bloom and Sosniak (1981) looked at talented internationally recognized individuals across different fields and found that most of these individuals had at least one model of achievement in their life during childhood (oftentimes, a parent). Zuckerman (1977) studied ninety Nobel laureates and found a critical role for modeling in at least some types of creativity. And Simonton (1975, 1984) found that those who had been exposed to creative role models were more likely to be recognized for their creativity. In the area of cognitive modeling, Gist (1989) found that a cognitive modeling training program for managers focusing on techniques to enhance innovative problem solving increased the quantity and divergence of ideas generated, and Meichenbaum (1975) found that cognitive modeling led to an increase in flexibility and originality on tests of divergent thinking. Finally, Harris and Evans (1974) did find that college students who were given creative responses to read subsequently exhibited higher creativity on a similar task than those given a more standard response to read. However, their task was a fairly simple, algorithmic task, so it remains unclear whether these results would hold for a more complex-heuristic task. Therefore, Shalley and Perry-Smith (2001) set out to examine whether modeling has an effect on subsequent creativity performance, and if so, whether the

Research on Employee Creativity

191

type of model makes a difference. The results in their study concerning expected evaluation have already been discussed in the performance evaluation section of this paper, so here we will focus on their arguments and results for modeling. Their anticipated reasoning for the effects of modeling on creativity was based on social learning theory (Bandura, 1969). According to this theory (Bandura, 1969, 1986), learning can take place vicariously by modeling and self-control processes. If individuals are capable of performing a behavior, but do not, they are more likely to exhibit it after a visual demonstration of the behavior (i.e. behavioral modeling) or through the transmission of examples of appropriate rules and thought processes (i.e. cognitive modeling). Through vicarious learning, individuals do not merely observe an example or behavior but actively pay attention to and store a symbolic representation in their mind of how to behave in certain situations (Bandura, 1969). Thus, modeling can provide cognitive and behavioral tools for innovation by providing standards for organizing thought, strengthening or weakening social inhibitions concerning appropriate behaviors, and encouraging versatility (Bandura, 1986). Shalley and Perry-Smith (2001) argued that individuals who were given a creative work model should be able to learn what is considered creative from this model, and this would actually result in them exhibiting more creative behavior. On the other hand, if someone was given a more standard or routine model to review, this would cause them to have lower creativity as compared to those given a creative model or those given no model. The researchers expected this effect also when the individuals were specifically told that the standard model was given to them as an example of what is not desired. Results of their laboratory study showed that individuals exhibited greater creativity when they were given a creative model than individuals not given such a model. Individuals who were given no model or a standard model did not differ significantly on their level of creative performance. Finally, individuals working in the expected controlling evaluation and standard solution condition exhibited the lowest levels of creativity. Therefore, providing work models can serve as a contextual factor for creativity. In addition to providing individuals with models of creative work, we also can surround individuals with creative role models – other employees. In the two field studies in the section titled “developmental feedback” described above, Zhou (2003) did not predict direct or main effects of the presence of creative co-workers on employee creativity. Consistent with Shalley and Perry-Smith (2001), Zhou developed the hypotheses by integrating the intrinsic motivation framework of creativity and social learning theory, and she added an interactional perspective of creativity to this study. More specifically, Zhou theorized that the presence of creative coworkers would provide employees with the opportunity to learn and acquire creativity – relevant

192

JING ZHOU AND CHRISTINA E. SHALLEY

skills and strategies. In addition, employees also needed to experience intrinsic motivation in order to take advantage of this opportunity. She argued that supervisors might maintain or enhance employees’ intrinsic motivation by either providing high levels of developmental feedback or low levels of close monitoring (i.e. the extent to which supervisors keep close tabs on their employees to ensure that the employees do exactly what they are told, perform tasks in expected ways, and do not do things that supervisors might disapprove of). Converging results from the two field studies demonstrated that, as hypothesized, when creative coworkers were present, the less supervisors engaged in close monitoring, the greater the creativity exhibited by the employees. Study 2 also found that the contribution of this joint condition was stronger for employees with less creative personalities. In addition, Study 2 found that when creative coworkers were present, the more supervisors provided developmental feedback, the greater the employees’ creativity. Presence of competitive others. Another creativity-related contextual factor that has intrigued researchers is competition. While anecdotes have suggested a powerful and positive impact of competition on innovation and technological advancement in industries, both of which contain elements of creativity, little systematic research has been conducted to determine how competition affects individual creativity. Shalley and Oldham (1997) conducted one of the few empirical studies devoted to examining the impact of competition on creativity. On the basis of the intrinsic motivation framework of creativity, these researchers argued that the effects of competition on creativity were complex; designed appropriately, competition might enhance instead of minimize creativity. Thus, it was necessary to distinguish two aspects of competition: informational and controlling. Depending on which aspect is made more salient, competition was hypothesized to either facilitate or inhibit creativity. When the informational aspect is relatively more salient, competition provides individuals with valuable information concerning how well they are performing the task. Individuals may welcome and indeed desire such information. Hence, their intrinsic motivation is maintained or enhanced, and subsequently their creativity is enhanced. In contrast, when the controlling aspect of competition is relatively more salient, the individuals are likely to feel that they are performing the task because an external factor forces or coerces them to do so, and their intrinsic motivation and creativity are likely to be diminished. On the basis of the above conceptual analysis, Shalley and Oldham (1997) made the informational or the controlling aspect of competition more salient by manipulating two experimental factors: the presence and visibility of competitors. They had three competitor presence conditions (i.e. no competition, competing with present others, and competing with absent others) with two visibility

Research on Employee Creativity

193

conditions (i.e. visible or non-visible to others while working). Participants worked on Torrance’s (1974) creativity task in which they generated unusual uses for five objects. The researchers used three creativity measures: fluency (i.e. the total number of new uses generated for the five objects), flexibility (i.e. the extent to which each participant used different categories of ideas or approaches to a problem), and overall creativity (i.e. the extent to which the uses generated were both novel and practical). Results indicated that participants exhibited higher levels of creativity in two informational conditions (i.e. when participants were in competition with others present but not visible, and when they were in competition with absent others and were visible to noncompetitive others) than in a controlling condition (i.e. when participants were in competition with present others and visible to them), thereby supporting the informational versus controlling competition hypothesis. However, other results failed to support the hypothesis. In particular, participants who competed with absent others and were not visible to noncompetitive others exhibited low levels of creativity. Taken together, results of this study provided mixed support for the informational versus controlling competition hypotheses. The above study underscores the complexity of the competition – creativity relationship. More systematic research is needed to untangle this complexity. We may need to abandon the hope for finding a straightforward positive or negative impact of competition on creativity. In fact, a host of personal (e.g. gender) and social factors (intra-group versus inter-group competition) might need to be taken into consideration in order to develop a fuller understanding of how and why competition affects creativity (Amabile, 1996). In summary, the role of others in affecting an individual’s level of creativity appears to largely depend on whether others have exerted influences that the focal individual perceives as more controlling or informational. Leadership and Supervisory Behaviors Theory and research also indicate that leadership and supervisory behaviors may play an important role in the creative process. Notably, Oldham and Cummings (1996) suggested that supportive, non-controlling supervisors create a work environment that fosters creativity. Similarly, Andrews and Farris (1967) found that scientists’ creativity was higher when their organizations were supportive and when managers listened to their employees’ concerns and asked for their input into decisions affecting them. A number of studies have shown that open interactions with supervisors, encouragement, and support enhance employees’ creativity and innovation (e.g. Kimberly, 1981; Oldham & Cummings, 1996; Tierney et al., 1999). Therefore, managers should work on personally encouraging employees and developing nurturing relationships to promote creativity. If they

194

JING ZHOU AND CHRISTINA E. SHALLEY

are supportive and provide adequate resources for employees to be creative (e.g. Amabile et al., 1996), creative activity should be more likely to occur. A number of field studies examined the relationship between supervisor behaviors and employee creativity. For example, Oldham and Cummings (1996) identified two types of creativity-relevant supervisor behaviors, namely supportive supervision (e.g. encouraging employees to voice their concerns) and controlling supervision (e.g. pressuring employees to think, feel, or behave in certain ways). On the basis of the intrinsic motivation perspective, they argued that supportive supervision would relate positively to creativity, and controlling supervision would relate negatively to creativity. They collected data on three indicators of creativity from a sample of manufacturing employees and their supervisors. The first indicator was supervisor ratings of employee creativity. The second measure was patent disclosures written, referring to the total number of internal patent disclosures written by the employee over a two-year period. The third indicator was whether employees’ contributions to their organizations’ suggestion programs were accepted (1 = accepted and 0 = unaccepted). The prediction concerning the main effect of supportive supervision on creativity did not receive support. Also, the prediction concerning the main effect of controlling supervision received partial support (i.e. controlling supervision negatively related to supervisor ratings of creativity). Finally, and consistent with the interactional approach of creativity, results of this study showed that personal and contextual factors interacted to affect creativity. These latter results will be reviewed in more detail in the section on individual differences later in this paper. Using the interactional approach in understanding creativity, George and Zhou (2001) examined the role that one specific supervisor behavior – close monitoring – played in creating a context or situation in which conscientiousness was related to individuals’ creative behavior. Conscientiousness is one of five traits in the Five-Factor Model of personality (Costa & McCrae, 1992). It refers to individual differences in organization, determination, impulse control, and conformity (Costa & McCrae, 1992; Hogan & Ones, 1997). Individuals who are high on this trait tend to have a strong sense of purpose and self-control. Conscientious individuals are achievement-oriented and hard working, they obey rules and conform to norms, and they are dependable, reliable and thorough (Costa & McCrae, 1992; Goldberg, 1992; Hogan & Blake, 1996). While prior research has shown that conscientiousness was strongly and positively related to job performance (Barrick & Mount, 1991; Tett, Jackson & Rothstein, 1991), Costa and McCrae (1992) suggest that high conscientiousness may result in excessive meticulousness, orderliness, or workaholic tendencies. George and Zhou (2001) argued that these tendencies might not contribute positively to creative performance. They argued that conforming, controlling

Research on Employee Creativity

195

one’s impulses, following rules, and striving to achieve predetermined goals could go counter to changing the status quo and coming up with new and better ways of doing things. Thus, according to the interactional perspective of creativity, under certain situations individuals high on conscientiousness may actually exhibit low levels of creativity (Feist, 1998; Walker, Koestner & Hum, 1995). Indeed, using a sample of office employees from a manufacturing company, George and Zhou (2001) found no main effect for conscientiousness but there was an interaction effect. Specially, conscientiousness resulted in low levels of creativity when supervisors engaged in close monitoring and coworkers were unsupportive (i.e. engaging in inaccurate communication, being unhelpful, and creating a negative work environment). Redmond, Mumford and Teach (1993) conducted a laboratory experiment in which they manipulated different types of leader behaviors, such as contributing to high or low subordinate self-efficacy, setting learning or performance goals, and providing or not providing problem construction. Ninety-six students participated in the experiment. They worked on a marketing task in which they played the role of a marketing intern who was asked to develop an advertising campaign for a novel fictional product, and one of three male confederates played the role of their manager. Consistent with the consensual assessment technique (Amabile, 1983), 20 marketing executives served as judges and rated the originality and quality of the participants’ solutions. Results showed that leader behaviors that contributed to problem construction and feelings of high self-efficacy led to greater subordinate creativity. In a recent study of Korean companies, Shin and Zhou (in press) examined the role of transformational leadership and conservation value on follower creativity in a sample of 290 employees and their supervisors from 46 companies. Based on Bass’ (1985) definition of transformational leadership, it is comprised of four dimensions: intellectual stimulation, individualized consideration, charisma, and inspirational motivation. Transformational leadership was expected to positively relate to follower creativity by boosting their intrinsic motivation. In addition, the relationship between transformational leadership and creativity was expected to vary for employees with different levels of the value of conservation. Conservation value involves tradition, conformity, and security (Schwartz, 1992). Shin and Zhou hypothesized that conservation value would moderate the relationship between transformational leadership and creativity such that, for those followers with higher levels of conservation, transformational leadership would have a stronger positive relationship with creativity than for followers lower on conservation. Further, they proposed that intrinsic motivation would mediate the moderated relationship between transformational leadership, conservation, and creativity. Indeed, results showed that transformational leadership was positively related

196

JING ZHOU AND CHRISTINA E. SHALLEY

to follower creativity. Also, followers’ conservation value did moderate the relationship between transformational leadership and creativity. Finally, intrinsic motivation partially mediated the contribution of transformational leadership to creativity and fully mediated the contribution of the transformational leadership by conservation interaction to creativity. Therefore, this study lends some empirical support to the intrinsic motivation perspective of creativity. While most of the aforementioned studies examined specific supervisory behaviors based on the intrinsic motivation perspective of creativity, and the Shin and Zhou (in press) study integrated transformational leadership theory and the intrinsic motivation perspective, two other studies – the Scott and Bruce (1994) study and the Tierney, Farmer and Graen (1999) study – used leader-member exchange (LMX) theory (Dansereau, Graen & Haga, 1975; Graen & Scandura, 1987). Scott and Bruce (1994) argued that LMX theory suggested that the quality of the exchange or relationship between a supervisor and his or her subordinate should be related to the subordinate’s innovativeness. The relationship between the supervisor and different subordinates may differ in its quality. Over time, the supervisor and some of his or her subordinates may develop a high-quality relationship in which the subordinates gain trust, autonomy, and discretion in task performance and decision-making, all of which would allow the subordinates to have the opportunity to be innovative. On the other hand, the supervisor and other subordinates may develop low-quality relationships in which the subordinates do not have the supervisor’s trust and liking, and do not have autonomy or discretion to come up with creative ideas. Hence, Scott and Bruce predicted that the quality of leader-member exchange between an employee and his or her supervisor would be positively related to his or her innovative behavior. In addition, the researchers predicted that the quality of leader-member exchange between an employee and his or her supervisor would be positively related to the employee’s perception of an innovation-supportive climate. To test their hypotheses, Scott and Bruce collected data from a sample of 172 engineers, scientists, and technicians and their managers from a research and development (R&D) facility of a large company. They used the 6-item supervisor rating scale discussed earlier under the measurement section of this paper to measure innovative behavior. Results supported the hypotheses. Tierney, Farmer and Graen (1999) examined the role of leadership for creativity using an interactionist model of employee characteristics, leader characteristics and LMX. They collected data from 191 R&D employees of a large chemical company. In the study, they measured creative performance by having supervisors rate their employees, and also by counting the number of invention disclosure forms submitted and the number of research reports published by each employee. Results suggested that employees’ intrinsic motivation orientation (a trait-like

Research on Employee Creativity

197

characteristic; see Amabile, 1996 for a discussion concerning state and trait intrinsic motivation), cognitive style, and LMX significantly affected their creative performance. Moreover, there were significant interactions between employees’ intrinsic motivation orientation and leaders’ intrinsic motivation orientation, and between LMX and employee cognitive style that were related to employee creative performance. Other Characteristics of the Work Environment Researchers have examined the effect of other characteristics of the work environment on employees’ creativity. Although there has been less research on each of these factors than the ones we reviewed above, it is informative to highlight their preliminary findings. First, some research has examined the effect of autonomy or discretion on employee creativity. Both Shalley (1991) and Zhou (1998a) investigated how personal discretion in working on a task may influence individuals’ creative performance. Neither study found a direct or main effect of task autonomy or personal discretion on individual creativity. However, they both found that autonomy or discretion interacted with other contextual factors to influence creativity. More specifically, in Shalley’s (1991) study, participants with high personal discretion had complete freedom in how they worked on the task and participants with low discretion were told that they had to follow a fixed work procedure. She found that individuals exhibited low levels of creativity when they had no creativity goal and a productivity goal or no creativity goal and low personal discretion. Further, this study suggested that the presence of a creativity goal appeared to compensate for the negative effects of low task discretion on creativity. On the basis of the intrinsic motivation perspective of creativity, Zhou (1998a) found that task autonomy interacted with feedback valence and feedback style in such a way that when individuals received positive feedback delivered in an informational style, and when simultaneously working in a high task autonomy condition, they exhibited the highest level of creativity than working in all other conditions involving feedback valence, style, and autonomy. One important aspect of performance management that has received less research attention concerns with effects of the provision of rewards on employee creativity. Little field research has directly examined the effects of rewards on employee creativity in the workplace, although a few studies using schoolchildren as research participants have been conducted in the behavioral laboratory (Amabile, Hennessey & Grossman, 1986; Eisenberger & Armeli, 1997; Eisenberger & Selbst, 1994). If creativity is one component of an employee’s role expectations, one would expect it to be rewarded appropriately. It has been argued that

198

JING ZHOU AND CHRISTINA E. SHALLEY

employees need to know that creative behavior is acceptable, that they may actually take risks within certain bounds and possibly fail without punishment, and that they should be aware that creative performance will be rewarded (e.g. Shalley, in press). For instance, Abbey and Dickson (1983) found that the climate of innovative R&D units included rewards for recognition of performance and the willingness to experiment with ideas. Rewards can be monetary or non-monetary recognition or praise. Consistent with the intrinsic motivation perspective for creativity, Eisenberger and Armeli (1997) have argued for the informational value of a reward for creativity. Specifically, rewards should be seen as something given in recognition of an individual’s competence, attempts to engage in creative activity, and their actual creative accomplishments. When seen this way, rewards may be perceived as informational and are likely to have a positive effect on employees’ creativity. Most creativity research has examined effects of contextual factors on creativity within an organization’s boundary. Few studies have examined the impact of factors outside of the organization’s boundary. A notable exception is a study conducted by Madjar, Oldham and Pratt (2002) on work and non-work support for creativity. These researchers define support for creativity as the extent to which other individuals aid and encourage employees’ creative performance. They expected that two sources of support – support from co-workers and supervisors and support from family and friends – would each enhance employees’ creativity. In addition, the researchers hypothesized that the mood states (positive or negative) experienced by the employees would mediate the relationship between support for creativity from work and non-work sources and employee creativity. They tested the hypotheses with 265 employees and their supervisors from three organizations in the Bulgarian knitwear industry. Results showed that support from both the work and non-work sources were each positively correlated with employees’ creativity rated by their supervisors. Interestingly, hierarchical regression results showed that support from family and friends contributed to employees’ creativity over and above the contribution made by coworkers and supervisors. In addition, results demonstrated that positive mood states mediated the relationship between work and non-work support and employees’ creativity. Whereas many prior studies focused on a subset of contextual factors in organizations, Amabile, Conti, Coon, Lazenby and Herron (1996) examined effects of the total work environment on creativity. Using Amabile’s componential model of creativity as a point of departure, these researchers reviewed a diverse set of literature and identified aspects of the work environment that stimulate or inhibit creativity in the workplace. Amabile and colleagues developed and validated a measurement instrument, which they called KEYS, to measure the aspects of the

Research on Employee Creativity

199

work environment identified above. Through various rigorous tests, they found that KEYS had satisfactory psychometric properties, and could readily distinguish the work environments for high-creativity projects and low-creativity projects. The aspects of stimulating environments include organizational encouragement (e.g. an organizational culture that supports creativity), supervisor encouragement (e.g. a supervisor who encourages creativity), work group supports (e.g. a diverse and open work group), sufficient resources (e.g. sources necessary for creativity), challenging work (e.g. work tasks are challenging and important), and freedom (e.g. freedom in deciding what to work on or how to go about it). The aspects of inhibiting environments include organizational impediments (e.g. internal politics) and workload pressure (e.g. excessive time pressures). Conclusion There has been quite a bit of research conducted to examine the effects of contextual factors on creativity. Some areas are more clearly defined than others, but overall, there is still need for more work. Also, much of the prior research on the effects of contextual conditions on creativity has focused on motivation as the underlying explanatory factor, yet few studies have directly examined whether this relationship actually exists. Further, the studies that have empirically examined this have not yielded definitive results. For example, Shin and Zhou (in press) found that intrinsic motivation partially mediated the contribution of transformational leadership for individuals’ creativity. On the other hand, Shalley and Perry-Smith (2001) found no significant mediation for intrinsic motivation in the relationship between expected evaluation and creative performance. Therefore, more research identifying and examining underlying or intervening psychological processes and how they affect employee creative performance is warranted.

Antecedents of Creativity – Individual Differences Creative Personality Early research on creativity emphasized the importance of individual personality and problem-solving ability or style that result in creativity (Barron, 1965; Mackinnon, 1965). These researchers have argued that individuals differ in their potential to come up with creative ideas or to produce creative outputs, although identifying a stable set of personality traits that can be used to reliably and potently predict actual creative performance over time and across situations, especially in the workplace, continues to remain elusive (e.g. Barron & Harrington, 1981; Martindale, 1989). One notable effort in isolating a set of creativity-relevant personality characteristics was exhibited by Gough (1979). Using a set of adjectives

200

JING ZHOU AND CHRISTINA E. SHALLEY

that are said to be descriptive of creative (e.g. confident) or non-creative (e.g. interest narrow) personalities, Gough developed and validated a measurement instrument called Creative Personality Scale (CPS). A study previously mentioned in this review (Oldham & Cummings, 1996) used Gough’s (1979) CPS to examine the direct and moderated relationship between creative personality and creativity. Results showed that CPS was positively correlated with just one of the three indicators of creativity-patent disclosures written. In addition, using the interactional approach, Oldham and Cummings hypothesized that personal factors (i.e. creativity-relevant personal characteristics) and contextual factors (e.g. job complexity, supportive supervision and non-controlling supervision) would interact such that employees would exhibit the highest creativity when they scored high on the CPS, worked on complex jobs as indicated by a motivating potential score (Hackman & Oldham, 1980), and were supervised in a supportive and non-controlling fashion. Results involving two of the three indicators of employee creativity-patent disclosures written and supervisor ratings of creativity – supported the researchers’ predictions. Zhou and Oldham (2001) examined the main and interactive effects of creative personalities and expected developmental assessment strategies on creativity. They found that individuals with more creative personalities, which were measured by the CPS, exhibited higher levels of creative performance. In addition, individuals exhibited the highest level of creative performance when they expected a self-administered assessment (i.e. an opportunity to assess their own work in order to develop their creativity-relevant skills) and possessed creative personalities. In a more recent study, Zhou (2003) noted that prior research focused on conditions that maximize the creativity of individuals with creative personalities (Oldham & Cummings, 1996; Zhou & Oldham, 2001). She raised an interesting and practical question concerning how to promote creativity in employees with less creative personalities (George & Zhou, 2001). Indeed, an unexpected finding in the Madjar et al. (2002) study was that individuals with less creative personalities exhibited greater creativity when they received support from family and friends. Thus, Zhou (2003) set out to investigate the possibility that individuals with relatively less creative personalities would benefit more from the joint conditions of low supervisor close monitoring (or high supervisor developmental feedback) and the presence of creative coworkers than individuals with creative personalities. She reviewed the literature that indicated that individuals with relatively low selfesteem or less confidence in their own capabilities and behavior were more likely to pay attention to competent role models than individuals with high self-esteem (Bandura, 1986; Weiss, 1977, 1978). She noted that low self-confidence is a hallmark of individuals with less creative personalities (Feist, 1998; Gough, 1979).

Research on Employee Creativity

201

In addition, this literature showed that individuals with less prior experiences on a given task tended to benefit more from models than individuals with more experiences (Rakestraw & Weiss, 1981). On the basis of these diverse sets of literature, Zhou concluded that individuals with less prior experience, less confidence or lower self-esteem were more uncertain about appropriate behaviors, approaches, or standards. These individuals would be more likely to look for and be influenced by role models because the models’ behaviors and strategies could serve as behavioral standards, guidance or sources of inspiration. Following the same logic, Zhou argued that employees with less creative personalities would be more likely to seek out and benefit substantially from the presence of creative models under conditions such as supervisors’ low close monitoring or high developmental feedback. Under both of these two supervisor conditions, individuals with less creative personalities, who were relatively less self-confident and have less prior experiences or success in creative activities, should experience enhanced interest in the task. This enhanced interest should allow them to find creative models stimulating and instructive. Under the joint condition of low close monitoring (or high developmental feedback) and the presence of creative coworkers, then, they would not only learn a great deal of creativity-relevant skills and strategies from coworkers, but also be motivated to experiment and use these strategies and engage in creative endeavors. Consequently, they should become more creative. Results of Zhou’s (2003) study partially supported her predictions. Results showed that creative personality moderated the joint condition of close monitoring and the presence of creative coworkers. Specifically, employees with relatively less creative personalities exhibited greater creativity when creative co-workers were present and close monitoring was low than when creative coworkers were present and close monitoring was high. Big-Five Personality Factors As previously discussed, the CPS, the creative personality measure, taps a set of specific personality characteristics said to be related to creative performance. Some researchers have argued that examining relations between creativity and more global personality factors that are based on an agreed-upon personality theory holds the promise of facilitating theory development and comparison across studies (e.g. Feist, 1998, George & Zhou, 2001). In recent years, some personality researchers have come to a consensus that the Five-Factor Model of personality is a comprehensive and parsimonious representation of personality (McCrae, 1989; Wiggins & Trapnell, 1997; see Block, 1995, however, for limitations of this model). According to the Five-Factor Model, personality traits are hierarchically organized with five global factors sitting on top of this hierarchy

202

JING ZHOU AND CHRISTINA E. SHALLEY

and more specific traits nested under each of the five factors. The five factors are called conscientiousness, openness to experience, extraversion, neuroticism, and agreeableness. An example of a facet of the broad conscientiousness construct is dutifulness. Another is achievement orientation. Of the five factors, conscientiousness and openness to experience are considered to be most conceptually relevant to creative performance (Costa & McCrae, 1992; McCrae, 1987). Indeed, in a meta-analytic review of studies based on samples of artists and scientists, Feist (1998) concluded that creative artists and scientists tended to be more open to experiences and less conscientious than their less creative counterparts. The Feist review did not speak directly about how conscientiousness and openness to experience are related to creative performance exhibited by employees whose jobs do not revolve around creativity. In fact, the personality and psychological processes underlying this latter everyday creativity in business organizations might be different from the creativity of artists and scientists (Feist, 1998; George & Zhou, 2001). With this in mind, George and Zhou (2001) investigated when conscientiousness and openness to experience were related to creativity in employees. As reviewed previously, conscientiousness was related to low levels of creativity when supervisors engaged in close monitoring and coworkers were unsupportive (i.e. engaging in inaccurate communication, being unhelpful, and creating a negative work environment). Individuals with higher openness to experience were more curious, flexible, imaginative, independent, sensitive to aesthetics, sophisticated, and they were more open to change, new ideas, experiences, and unconventional perspectives than individuals scoring lower on openness. In contrast, individuals with lower openness were more conservative, conventional, rigid and socialized (Costa & McCrae, 1992, 1995; Feist, 1998). Taking an interactional perspective, George and Zhou (2001) argued that although high levels of openness to experience predispose individual employees to be creative, this predisposition may not manifest in organizations because organizations sometimes present people with strong situations that shape their behavior. In order for employees high on openness to experience to actually exhibit creative performance, organizational contexts need to allow for and encourage the manifestation of their predisposition to be creative. George and Zhou (2001) identified two aspects of the organizational context that are especially helpful for employees with high openness: positive feedback valence and heuristic tasks (i.e. tasks with unclear ends, unclear means, or multiple means). Results showed no main effect of openness to experience, but employees with high levels of openness to experience exhibited greater creativity when feedback valence was positive and tasks were heuristics (i.e. ends or means of the tasks were unclear. That is, they were not predetermined for or imposed on the employees).

Research on Employee Creativity

203

Self-Efficacy and Role Identity An emerging area is employees’ beliefs about whether they can be creative or not. One approach to address this issue is to examine creative role identity. Creative role identity refers to whether an individual views him- or herself as a creative person. Farmer, Tierney and Kung-McIntyre (in press) investigated how creative role identity is associated with employees’ creative performance. They surveyed a sample of engineers, software developers, research scientists, doctors and pharmacists from eight Taiwanese organizations. Their results showed that creative role identity was predicted by perceived coworker creativity expectations, self-views of creative behaviors, and high levels of exposure to the U.S. culture. Creativity was found to be the highest when a strong creative role identity was paired with perceptions that the organization valued creative work. Another approach focuses on the idea of creative self-efficacy. Using data from two diverse firms, Tierney and Farmer (2002) tested a new construct, creative self-efficacy. Creative self-efficacy refers to the employees’ beliefs that they can be creative in their work roles. They found that job tenure, job self-efficacy, supervisor behavior, and job complexity contributed to creative efficacy beliefs. Creative self-efficacy also was associated with creative performance. Future research will need to further explore the role of creative self-efficacy and creative role identity and how these constructs may interact with various contextual factors in influencing creative performance. Research also needs to examine the links between creative role identity and creative self-efficacy. One appears to be a more global measure (e.g. creative role identity), the other more specific (e.g. creative self-efficacy), for the same kind of concept. Which construct is more important for creative performance remains a subject for future research. Conclusion As demonstrated in the field studies reviewed above, individual differences variables still cannot reliably and potently predict actual creative performance across situations in the workplace. For example, CPS predicted one of three indicators of creativity in the Oldham and Cummings (1996) study. In the George and Zhou (2001) study, neither of the zero-order correlations between openness to experience and creativity, and between conscientiousness and creativity was statistically significant. Clearly, more studies are needed before a firm conclusion can be drawn. Nevertheless, the existing literature suggests that to the extent that it is consistent with theories being tested, it is advisable that the person-situation interaction perspective be used as a guiding principle in future field studies. In doing so, however, the researchers need to elaborate and specify how exactly the person and the situation interact to affect employee creativity (George & Zhou, 2001), instead of simply making a general statement

204

JING ZHOU AND CHRISTINA E. SHALLEY

that both the person and the situation are necessary for creative performance to occur. Another fruitful avenue for future research in this area may involve exploring the relative benefit of specific versus global personality factors in predicting employee creativity. There already appears to be an advantage to using specific personality characteristics (e.g. CPS) in terms of more reliable predictive ability than more global personality factors (e.g. openness to experience). Similarly, creative self-efficacy seems to be a promising new avenue to explore and research is needed to examine whether it is more relevant than creative role identity.

DISCUSSION, IMPLICATIONS, AND FUTURE RESEARCH DIRECTIONS
Thus far, we have reviewed prior theory and research on employee creativity in the workplace. As evident in this review, employee creativity is a relatively young field of research; yet, considerable progress has been made. Looking ahead, researchers are faced with many conceptual and methodological challenges, and exciting opportunities. We now turn to a discussion of these challenges and opportunities for adding to our understanding of the creativity phenomenon. The issues discussed below are meant to stimulate future research, rather than serve as an exhaustive list of the research needed to be done.

Challenges and Unanswered Questions The Paradox of Rewards As reviewed earlier in this paper, few studies have focused on the effects of rewards or compensation programs on creativity in the workplace despite the important function and widespread use of rewards in organizations. Studies conducted in the behavioral laboratory yielded rather mixed results. On one hand, Amabile, Hennessey and Grossman (1986) demonstrated negative effects of contracted-for-rewards (i.e. participants agree to work on a task in order to receive a reward. As such, the task activity becomes a means to an end) on creativity. On the other hand, Eisenberger and colleagues (e.g. Eisenberger & Armeli, 1997; Eisenberger & Selbst, 1994) showed that rewards can have informational value, which can be used to encourage creativity. Given that practitioners tend to believe rewards should be used to promote a variety of work outcomes including creativity, and many compensation programs (e.g. stock options and gain-sharing plans) have been designed for this purpose, it is imperative to

Research on Employee Creativity

205

systematically untangle the complexity of the impact of rewards on creativity in the workplace. Additional Contextual Factors More studies are needed to identify and examine contextual factors using existing theoretical perspectives (e.g. the intrinsic motivation perspective). For instance, both of the main creativity models discussed earlier mention the importance of the availability of slack resources for creative performance, yet little research has empirically examined the effect of the availability of different levels of slack resources or different types of slack resources on creativity at work. In addition to the social context at work, we also need to examine effects of the physical work environment (Oldham, Cummings & Zhou, 1995) on employee and work group creativity. This is because it would be relatively easy to design offices or alter office configurations in ways to facilitate creative performance if we know what specific factors in the physical office space may enhance or stifle creativity. Furthermore, research has not examined the effect of different types of organizational designs (e.g. matrix, collateral group structures) on the creativity of individuals and groups at work. These, along with others, are promising avenues for future research to explore. Psychological Mechanisms We need to directly examine theorized psychological mechanisms linking contextual factors and employee creative performance. As reviewed earlier, studies conducted by Shalley and Perry-Smith (2001), Madjar et al. (2002), and Shin and Zhou (in press) represent important steps in this line of research. Clearly, more studies of this type need to be conducted in organizations where the stakes for creative performance are noteworthy. Measurement Issues As discussed earlier, research is needed to tackle unanswered questions concerning measurement of creative performance in both lab and field studies. In studies conducted in the behavioral laboratory, two approaches have been used to assess creativity: the raters rate overall creativity vs. the raters rate separately the originality and usefulness dimensions and then multiply these two scores to create a creativity index. In the latter (and newer) approach, should the originality and usefulness scores be multiplied, added, or weighted in a particular way? For instance, is originality a more important component for creativity than usefulness and hence should it be weighted more heavily, or should the two dimensions of creativity be weighted equally? If some studies use the more established method of averaging ratings of overall creativity and others use the newer method of

206

JING ZHOU AND CHRISTINA E. SHALLEY

multiplying separate ratings of originality and usefulness ratings, does this affect our ability to compare results of different studies? In addition, four types of supervisory rating scales have been used in field studies. Systematic research is needed to evaluate the relative strengths and weaknesses of these four scales in terms of their ability to measure the full range of creative performance in the workplace.

Breaking New Theoretical Grounds While we need to continue to test, refine, and expand existing theories on creativity, perhaps more exciting research opportunities lie in breaking new theoretical grounds. There are several possibilities in this regard. First, we need to formulate and test models of work group or team creativity. Although social psychologists have conducted many laboratory studies on idea generation in groups (e.g. Paulus, Brown & Ortega, 1999), the groups involved in those studies were usually ad hoc groups assembled for the sole purpose of conducting a laboratory study. Group output was often the number of original ideas generated rather than the extent to which the output was both original and appropriate, which is the definition of creativity. In addition, much of this research has focused on one group creativity technique – brainstorming (Paulus et al., 1999). More theory development and empirical research are needed to specify the group and organizational factors that may affect creative output produced by intact work teams, and how these factors affect work team creativity. A study by West and Anderson (1996) on creativity in top management teams in the health care industry represents an important step in this direction. Second, theory and research are needed for an understanding of how dissatisfaction and bad mood affect employee creativity. George and Zhou (2002) observed that research in this area is at an interesting crossroads. On one hand, researchers have shown that positive mood, induced in laboratory settings, facilitates performance on problem solving tasks that involve creative thinking (Isen, Daubman & Nowicki, 1987; Isen, Johnson, Mertz & Robinson, 1985). And a recent field study by Madjar et al. (2002) demonstrated the positive relation between positive mood and employee creativity. On the other hand, studies have found that negative mood can foster creativity and positive mood can discourage it (Hirt, McDonald & Melton, 1996; Martin & Stoner, 1996; Tighe, 1992). In a study conducted at a helicopter manufacturing facility, and using a mood-as-input model, George and Zhou (2002) showed that negative moods were positively related to employee creativity when employees perceived a high level of recognition and rewards for creative performance and their clarity of feelings was high. In contrast, positive moods were negatively related to creativity when perceived recognition and

Research on Employee Creativity

207

rewards for creativity and clarity of feelings were high. As reviewed earlier in this paper, Zhou and George (2001) showed that under certain conditions, job dissatisfaction was positively related to employee creativity. Third, contextual factors in the workplace exist at multiple levels (i.e. individual, group, and organizational), and individual employees’ creativity is often influenced by cross-level factors (Woodman et al., 1993). Field research has often relied on questionnaire measures to assess contextual factors at only one level. In reality, factors at different levels may either work in concert to promote or reduce creativity or, more interestingly, contradict each other. This possibility makes understanding creativity more complex yet intriguing, and managing creativity more challenging. Consider, for example, effects of feedback on individual employee creativity. How would an employee respond to feedback if this particular feedback provided by a co-worker contradicts aggregated feedback given by other people in the employee’s group? Zhou (in press) argues that this contradiction is especially salient when one feedback giver provides negative feedback and several other people provide positive feedback. She speculates that this cross-level problem may not be equally salient or detrimental for all individuals. It is possible that employees with high levels of achievement motivation would be more responsive to negative feedback from one co-worker even though at the same time the rest of the group provided positive feedback. In fact, employees with high levels of achievement motivation may actively seek negative feedback in order to become more creative and to accomplish more. Employees with high levels of emotional intelligence also may be more responsive to one single co-worker’s negative feedback, even though the rest of the group has provided positive feedback. No doubt, cross-level contradictions such as this one are common in organizations, and theory and research are needed to fully explain and predict their effects on employee creativity. Fourth, we need to understand more about how patterns of relationships, or social networks, affect creativity at work. Perry-Smith and Shalley (2003) developed a social network perspective highlighting the social side of creativity and suggested a variety of mechanisms through which the social context influences creativity. Their propositions suggest that organizations interested in being innovative and creative should consider facilitating interactions across work groups, departments, and other discrete units. Also, they suggest that an organization may want to support social activities that promote contact with professionals outside of the organization that could lead to more creative strategies. Therefore, it makes sense that many of the relationships they have proposed should be further explored and empirically tested. Fifth, we need to theorize and examine how the creative process unfolds over time. Producing a creative idea may involve a series of stages, from the initial

208

JING ZHOU AND CHRISTINA E. SHALLEY

identification of a problem to the actual generation of creative ideas to validation, revision, and the eventual decision on a creative idea or solution. Research should start to examine the different stages and what individual and contextual factors are most important at each stage. Last, but not least, we need to identify and investigate antecedents of employee creativity in the international context. As businesses are becoming more global, so should research on employee creativity. Traditionally, researchers doing international work tend to use a comparative research design in which data are collected from samples from two or more countries (typically a Western country such as the U.S. and countries from other parts of the world) and the researchers examine whether a conceptual model developed in the U.S. applies to another country. Although this traditional approach is valuable, indigenous research holds considerable promise in advancing our understanding of creativity in the international context. In this context, indigenous research refers to a type of design in which the researchers focus their attention on identifying and uncovering unique creativity enhancing or restraining factors that are embedded in a non-Western cultural context. For example, a relatively robust finding in U.S.-based studies is that autonomy or choice is conducive to creativity. Will this finding hold in all types of cultural contexts?

Implications for Human Resource Management Research on employee creativity has profound implications for managing companies’ most valuable resources – their employees. Below we focus on discussing implications of creativity research for human resource management. The reader interested in implications beyond human resource management is referred to two recently edited volumes. More specifically, Ford (in press) offers a broader treatment of how creativity research is related to management practices in general, and Shavinina (2003) offers a comprehensive and systematic treatment of how employee creativity might be related to organizational innovation. Performance Appraisal All too often, researchers discover that top executives consider employee creativity to be highly desirable in their organizations, but they do not have a system in place to keep track of the extent to which their non R&D employees exhibit creativity (cf. George & Zhou, 2001). If creativity is desirable and contributes to an organization’s effectiveness and competitiveness, it must be measured and appraised, the employees must be given feedback on the creative ideas they come up with, and those who contribute truly new and useful ideas must be recognized

Research on Employee Creativity

209

and rewarded. Toward this end, creative performance must be recognized in the performance appraisal process. Past research and an abundance of anecdotal information support creativity as important for an organization’s long-term survival and competitiveness. However, having supervisors evaluate employees’ creative achievements is not without difficulties and challenges. Supervisors may not be able to discover and recognize creative ideas generated by their employees (Zhou, 1998b; Zhou & Woodman, 2003). Thus, supervisors may need to be trained on a continuous basis to become and remain as experts who are knowledgeable in certain domain. Derived from the consensual assessment technique (Amabile, 1983), to the extent possible, multiple supervisors may be used to evaluate an employee’s creativity. In addition, supervisors sometimes cannot accurately perceive and evaluate employee creativity. However, perceptual inaccuracies may not stem from their being incapable of doing so, but from their belief that employee creativity threatens their power (e.g. Staw, 1995). Indeed, creative ideas often represent a desire for change and a rejection of the status quo, and managers may or may not like to accept the proposed new ways of doing things (e.g. Zhou & George, 2001). To somewhat minimize the adverse impact of this tendency, and to the extent possible, organizations should periodically collect and evaluate more objective indicators of employee creativity such as employees’ contributions to the organizations’ suggestion programs (cf. Oldham & Cummings, 1996), in addition to supervisor ratings. Reward and Compensation Programs Although we have just stated above that creative achievements need to be acknowledged and rewarded, we caution that managers be mindful about the intricacies of effects of reward on creativity discussed earlier in this paper. We suggest that managers be careful not to explicitly tie rewards to creativity (i.e. contracted-forrewards), because research has shown that contracted-for-rewards undermined creativity (Amabile et al., 1986). However, it is also possible that certain types of reward programs actually are more facilitative of creativity. For instance, organizations that use stock option plans or gain-sharing programs may find that employees are more committed to the organization for the long term and more willing to exert the kind of effort that is needed in order for them to be creative. Additionally, if they experience negative mood states, they may feel more responsible for doing something about this, and ultimately they may be more creative (George & Zhou, 2002). Employee Relations Much of the research reviewed in this paper has indicated that employees will be more creative when they are treated in a supportive manner. This does not mean

210

JING ZHOU AND CHRISTINA E. SHALLEY

that managers should feel that they have to coddle their employees. Rather, the research indicates that it is important to provide employees with information that is conveyed in a supportive, constructive manner, whether or not this information is positive or negative in nature. It makes sense that when employees are treated in a supportive manner they will be more likely to listen to the information, potentially be more willing to alter their behaviors, and learn from this information. Ultimately, employees may demonstrate greater creativity. Recruitment and Selection While managers might be tempted to recruit and select employee with creative personalities, the research reviewed in this paper suggests that it might be problematic to primarily use recruitment and selection to boost an organization’s creativity stock. Prior studies have shown that creative abilities and personalities only provide individuals with the potential to be creative. Whether or not individuals with high creative potential will actually exhibit creative performance depends partly on the organizational context (e.g. George & Zhou, 2001), and partly on whether the individuals are motivated to do so (Amabile, 1996). Therefore, while recruiting and selecting employees with the right set of abilities and personality characteristics for creative activity can be desirable, setting up the right work context for creativity to emerge appears to be most promising. Training Although it has been argued that creativity is desirable in a wide variety of jobs in the workplace, producing creative ideas is not a naturally high-probability event for most people (Staw, 1995), and training may help to offset individuals’ inability to produce truly new and useful ideas. Basudur and his colleagues showed that training in creative problem solving can be very effective (e.g. Basudur, Graen & Green, 1982; Basudur et al., 1990). Such training can be organized around Amabile’s (1996) componential model of creativity. That is, the goal of such training is to enhance all three components – domain-relevant skills, creativity-relevant processes, and task motivation. For example, organizations may train employees so that their domain-relevant skills and creativity-relevant processes can be enhanced. Organizations may also train employees to embrace what they do and hence foster their task motivation. Further, organizations may train supervisors on how to set up the nurturing and supportive environment to boost employees’ motivation to be creative. In summary, we sought to review and discuss the last 15 to 20 years of contemporary theory and research on employee creativity in the workplace. Before this time, there was little organizational research on creativity. Since then, there have been a tremendous number of advancements in the literature that both academics

Research on Employee Creativity

211

and practitioners interested in human resources need to be aware of. On the basis of this critical review and appraisal, we have proposed a number of future research directions, and discussed implications of this research literature for human resource management practices. Because employees’ creative ideas and accomplishments are increasingly becoming the cornerstone of organizations’ competitiveness in today’s knowledge-based economy, much more research is needed in this area to enhance our understanding of what, how, and why certain contextual factors and individual differences affect employees’ creative performance. It is our hope that this paper provides a solid background for the reader by highlighting the areas that have been well researched, noting emerging research areas, and specifying other areas that deserve more research attention for truly understanding how organizations can try to enhance the occurrence of employees’ creative behaviors.

REFERENCES
Abbey, A., & Dickson, J. W. (1983). R&D work climate and innovation in semiconductors. Academy of Management Journal, 26, 362–368. Allen, N. J., & Meyer, J. P. (1996). Affective, continuance, and normative commitment to the organization: An examination of construct validity. Journal of Vocational Behavior, 49, 252–276. Amabile, T. M. (1979). Effects of external evaluation on artistic creativity. Journal of Personality and Social Psychology, 37, 221–233. Amabile, T. M. (1983). The social psychology of creativity. New York: Springer-Verlag. Amabile, T. M. (1988). A model of creativity and innovation in organizations. In: B. M. Staw & L. L. Cummings (Eds), Research in Organizational Behavior (Vol. 10, pp. 123–167). Greenwich, CT: JAI Press. Amabile, T. M. (1996). Creativity in context. Boulder, CO: Westview Press. Amabile, T. M., & Conti, R. (1999). Changes in the work environment for creativity during downsizing. Academy of Management Journal, 42, 630–640. Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996). Assessing the work environment for creativity. Academy of Management Journal, 39, 1154–1184. Amabile, T. M., Goldfarb, P., & Brackfield, S. C. (1990). Social influences on creativity: Evaluation, coaction, and surveillance. Creativity Research Journal, 3, 6–21. Amabile, T. M., Hennessey, B. A., & Grossman, B. S. (1986). Social influences on creativity: The effects of contracted-for reward. Journal of Personality and Social Psychology, 50, 14–23. Andrews, F. M., & Farris, G. F. (1967). Supervisory practices and innovation in scientific teams. Personnel Psychology, 20, 497–575. Bandura, A. (1969). Principles of behavior modification. New York: Holt, Rinehart & Winston. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Baron, R. S. (1986). Distraction-conflict theory: Progress and problems. In: L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 19, pp. 1–40). New York: Academic Press.

212

JING ZHOU AND CHRISTINA E. SHALLEY

Barron, F. (1965). The psychology of creativity. In: T. Newcombe (Ed.), New Directions in Psychology (Vol. 2, pp. 1–34). New York: Holt, Rinehart & Winston. Barron, F., & Harrington, D. M. (1981). Creativity, intelligence, and personality. Annual Review of Psychology, 32, 439–476. Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimensions and job performance: A meta-analysis. Personnel Psychology, 44, 1–26. Bartis, S., Szymanski, K., & Harkins, S. G. (1988). Evaluation and performance: A two edged knife. Personality and Social Psychology Bulletin, 14, 242–251. Bass, B. M. (1985). Leadership and performance beyond expectation. New York: Free Press. Basudur, M. S., Graen, G. B., & Green, S. G. (1982). Training in creative problem solving: Effects on ideation and problem finding and solving in an industrial research organization. Organizational Behavior and Human Performance, 30, 41–70. Basudur, M. S., Wakabayashi, M., & Graen, G. B. (1990). Individual problem-solving styles and attitudes toward divergent thinking before and after training. Creativity Research Journal, 3, 22–32. Block, J. (1995). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 117, 187–215. Bloom, B. S., & Sosniak, L. A. (1981). Talent development vs. schooling. Educational Leadership, 86–94. Carson, P. P., & Carson, K. D. (1993). Managing creativity enhancement through goal setting and feedback. Journal of Creative Behavior, 27, 36–45. Costa, P. T., & McCrae, R. R. (1992). Revised NEO personality inventory (NEO PI-R) and NEO five-factor inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., & McCrae, R. R. (1995). Solid ground in the wetlands of personality: A reply to Block. Psychological Bulletin, 117, 216–220. Dansereau, F., Graen, G., & Haga, W. (1975). A vertical dyad linkage approach to leadership within formal organizations: A longitudinal investigation of the role-making process. Organizational Behavior and Human Performance, 13, 46–78. Deci, E. L., & Ryan, R. M. (1980). The empirical exploration of intrinsic motivational processes. In: L. Berkowitz (Ed.), Advances in Experimental Social Psychology (pp. 39–80). New York: Academic Press. Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York: Plenum. Decker, P. J., & Nathan, B. R. (1985). Behavior modeling training. New York: Praeger. Drazin, R., Glynn, M. A., & Kazanjian, R. K. (1999). Multilevel theorizing about creativity in organizations: A sensemaking perspective. Academy of Management Review, 24, 286–307. Dweck, C. S. (1986). Motivational processes affecting learning. American Psychologist, 41, 1040– 1048. Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and personality. Psychological Review, 95, 256–273. Eisenberger, R., & Armeli, S. (1997). Can salient reward increase creative performance without reducing intrinsic creative interest? Journal of Personality and Social Psychology, 72, 652–663. Eisenberger, R., & Selbst, M. (1994). Does reward increase or decrease creativity? Journal of Personality and Social Psychology, 66, 1116–1127. Eisenhardt, K. M. (1989). Building theories from case study research. Academy of Management Review, 14, 532–550.

Research on Employee Creativity

213

Elliott, E. S., & Dweck, D. S. (1988). Goals: An approach to motivation and achievement. Journal of Personality and Social Psychology, 54, 5–12. Ettlie, J. E., & O’Keefe, R. D. (1982). Innovative attitudes, values, and intentions in organizations. Journal of Management Studies, 19, 163–182. Farh, J. L., Cannella, A. A., & Bedeian, A. G. (1991). Peer ratings. Group and Organization Studies, 16, 367–386. Farmer, S. M., Tierney, P., & Kung-McIntyre, K. (in press). Employee creativity in Taiwan: An application of role identity theory. Academy of Management Journal. Feist, G. J. (1998). A meta-analysis of personality in scientific and artistic creativity. Personality and Social Psychology Review, 4, 290–309. Ford, C. (1996). A theory of individual creative action in multiple social domains. Academy of Management Review, 21, 1112–1142. Ford, C. (in press). Handbook of organizational creativity. Hillsdale, NJ: Lawrence Erlbaum. George, J. M., & Zhou, J. (2001). When openness to experience and conscientiousness are related to creative behavior: An interactional approach. Journal of Applied Psychology, 86, 513–524. George, J. M., & Zhou, J. (2002). Understanding when bad moods foster creativity and good ones don’t: The role of context and clarity of feelings. Journal of Applied Psychology, 87, 687–697. Gist, M. E. (1989). The influence of training method on self-efficacy and idea generation among managers. Personnel Psychology, 42, 787–805. Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26–42. Gough, H. G. (1979). A creative personality scale for the Adjective Check List. Journal of Personality and Social Psychology, 37, 1398–1405. Graen, G. B., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing. In: L. L. Cummings & B. M. Staw (Eds), Research in Organizational Behavior (Vol. 9, pp. 175–208). Guilford, J. P. (1956). The structure of intellect. Psychological Bulletin, 53, 267–293. Guilford, J. P. (1967). The nature of human intelligence. New York: McGraw-Hill. Hackman, J. R., & Oldham, G. R. (1980). Work redesign. Reading, MA: Addison-Wesley. Harackeiwicz, J. M., & Elliott, A. J. (1993). Achievement goals and intrinsic motivation. Journal of Personality and Social Psychology, 65, 904–915. Hargadon, A., & Sutton, R. I. (1997). Technology brokering and innovation in a product development firm. Administrative Science Quarterly, 42, 716–749. Harris, M. B., & Evans, R. C. (1974). The effects of modeling and instructions on creative responses. The Journal of Psychology, 86, 123–130. Hirt, E. R., McDonald, H. E., & Melton, R. J. (1996). Processing goals and the affect-performance link: Mood as main effect or mood as input? In: L. L. Martin & A. Tesser (Eds), Striving and Feeling: Interactions among Goals, Affect, and Self-Regulation (pp. 303–328). Mahwah, NJ: Erlbaum. Hogan, R., & Blake, R. J. (1996). Vocational interests: Matching self-concept with the work envi ronment. In: K. R. Murphy (Ed.), Individual Differences and Behavior in Organizations (pp. 89–144). San Francisco: Jossey-Bass. Hogan, J., & Ones, D. S. (1997). Conscientiousness and integrity at work. In: R. Hogan, J. Johnson & S. Briggs (Eds), Handbook of Personality Psychology (pp. 849–870). San Diego, CA: Academic Press. Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem solving. Journal of Personality and Social Psychology, 52, 1122–1131.

214

JING ZHOU AND CHRISTINA E. SHALLEY

Isen, A. M., Johnson, M. M. S., Mertz, E., & Robinson, G. R. (1985). The influence of positive affect on the unusualness of word associations. Journal of Personality and Social Psychology, 48, 1413–1426. James, L. R., Demaree, R. B., & Wolf, G. (1984). Estimating within-group interrater reliability with and without response bias. Journal of Applied Psychology, 69, 85–98. Jussim, L., Soffin, S., Brown, R., Ley, J., & Kohlhepp, K. (1992). Understanding reactions to feedback by integrating ideas from symbolic interactionism and cognitive evaluation theory. Journal of Personality and Social Psychology, 62, 402–421. Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive-abilities – An integrative aptitude treatment interaction approach to skill acquisition. Journal of Applied Psychology, 74, 657–690. Kimberly, J. R. (1981). Managerial innovation. In: P. C. Nystrom & W. H. Starbuck (Eds), Handbook of Organizational Design (pp. 84–104). New York: Oxford University Press. Locke, E. A., & Latham, G. P. (1990). A theory of goal setting and task performance. Englewood Cliffs, NJ: Prentice-Hall. Locke, E. A., Shaw, K. N., Saari, L., & Latham, G. P. (1981). Goal setting and task performance: 1969–1980. Psychological Bulletin, 90, 125–152. London, M., Larsen, H. H., & Thisted, L. N. (1999). Relationships between feedback and selfdevelopment. Group and Organization Management, 24, 5–27. Mackinnon, D. W. (1965). Personality and the realization of creative potential. American Psychologist, 20, 273–281. Madjar, N., Oldham, G. R., & Pratt, M. G. (2002). There’s no place like home? The contributions of work and non-work creativity support to employees’ creative performance. Academy of Management Journal, 45, 757–767. Mainemelis, C. (2001). When the muse takes it all: A model for the experience of timelessness in organizations. Academy of Management Review, 26, 548–565. Manske, M. R., & Davis, G. A. (1968). Effects of simple instructional biases upon performance in the unusual uses test. Journal of General Psychology, 79, 25–33. Martindale, C. (1989). Personality, situation, and creativity. In: J. Glover, R. Ronning & C. Reynolds (Eds), Handbook of Creativity (pp. 211–232). New York: Plenum. Martin, L. L., & Stoner, P. (1996). Mood as input: What we think about how we feel determines how we think. In: L. L. Martin & A. Tesser (Eds), Striving and Feeling: Interactions among Goals, Affect, and Self-Regulation (pp. 279–301). Mahwah, NJ: Erlbaum. Matlin, M. W., & Zajonc, R. B. (1968). Social facilitation of word associations. Journal of Personality and Social Psychology, 10, 455–460. McCrae, R. R. (1987). Creativity, divergent thinking, and openness to experience. Journal of Personality and Social Psychology, 52, 1258–1265. McCrae, R. R. (1989). Why I advocate the Five-Factor Model: Joint factor analyses of the NEO-PI with other instruments. In: D. M. Buss & N. Cantor (Eds), Personality Psychology: Recent Trends and Emerging Directions (pp. 237–245). New York: Springer-Verlag. McGraw, K. (1978). The detrimental effects of reward on performance: A literature review and a prediction model. In: M. Lepper & D. Greene (Eds), The Hidden Costs of Reward (pp. 33–60). Hillsdale, NJ: Erlbaum. Meichenbaum, D. (1975). Enhancing creativity by modifying what subjects say to themselves. American Educational Research Journal, 12, 129–145. Mumford, M. D., & Gustafson, S. B. (1988). Creativity syndrome: Integration, application, and innovation. Psychological Bulletin, 103, 27–43. Nickerson, R. S. (1999). Enhancing creativity. In: R. Sternberg (Ed.), Handbook of Creativity (pp. 392–430). New York: Cambridge University Press.

Research on Employee Creativity

215

Oldham, G. R. (1988). Effects of changes in workspace partitions and spatial density on employee reactions: A quasi-experiment. Journal of Applied Psychology, 73, 252–258. Oldham, G. R., & Cummings, A. (1996). Employee creativity: Personal and contextual factors at work. Academy of Management Journal, 39, 607–634. Oldham, G. R., Cummings, A., & Zhou, J. (1995). The spatial configuration of organizations. In: G. R. Ferris (Ed.), Research in Personnel and Human Resource Management (pp. 1–37). Greenwich, CT: JAI Press. Paulus, P. B., Brown, V., & Ortega, A. H. (1999). Group creativity. In: A. Montuori & R. E. Purser (Eds), Perspectives on Creativity. Cresskill, NJ: Hampton Press. Pelz, D. C., & Andrews, F. M. (1966). Scientists in organizations: Productive climates for research and development. New York: Wiley. Perry-Smith, J. E., & Shalley, C. E. (2003). The social side of creativity: A static and dynamic social network perspective. Academy of Management Review, 28, 89–106. Rakestraw, T. L., & Weiss, H. M. (1981). The interaction of social influences and task experience on goals, performance, and performance satisfaction. Organizational Behavior and Human Performance, 27, 326–344. Redmond, M. R., Mumford, M. D., & Teach, R. (1993). Putting creativity to work: Effects of leader behavior on subordinate creativity. Organizational Behavior and Human Decision Processes, 55, 120–151. Robinson, J. C. (1980). Will behavior modeling survive the 1980s? Training and Development Journal, 34, 22–28. Schneider, B. (1983). Interactional psychology and organizational behavior. In: L. L. Cummings & B. M. Staw (Eds), Research in Organizational Behavior (Vol. 5, pp. 1–31). Greenwich, CT: JAI Press. Schwartz, S. J. (1992). Universals in the content and structure of values: Theory and empirical tests in 20 countries. In: M. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 25, pp. 1–65). New York: Academic Press. Scott, S. G., & Bruce, R. A. (1994). Determinants of innovative behavior: A path model of individual innovation in the workplace. Academy of Management Journal, 37, 580–607. Shalley, C. E. (1991). Effects of productivity goals, creativity goals, and personal discretion on individual creativity. Journal of Applied Psychology, 76, 179–185. Shalley, C. E. (1995). Effects of coaction, expected evaluation, and goal setting on creativity and productivity. Academy of Management Journal, 38, 483–503. Shalley, C. E. (in press). What managers can do to enhance creative activity: Role expectations for creative performance. In: C. Ford (Ed.), Handbook of Organizational Creativity. Hillsdale, NJ: Lawrence Erlbaum. Shalley, C. E., Gilson, L. L., & Blum, T. C. (2000). Matching creativity requirements and the work environment: Effects on satisfaction and intention to leave. Academy of Management Journal, 43, 215–223. Shalley, C. E., & Oldham, G. R. (1985). Effects of goal difficulty and expected evaluation on intrinsic motivation: A laboratory study. Academy of Management Journal, 28, 628–640. Shalley, C. E., & Oldham, G. R. (1997). Competition and creative performance: Effects of competitor presence and visibility. Creativity Research Journal, 10, 337–345. Shalley, C. E., & Perry-Smith, J. E. (2001). Effects of social-psychological factors on creative performance: The role of informational and controlling expected evaluation and modeling experience. Organizational Behavior and Human Decision Processes, 84, 1–22. Shavinina, L. V. (2003). International handbook on innovation. Hillsdale, NJ: Lawrence Erlbaum.

216

JING ZHOU AND CHRISTINA E. SHALLEY

Shin, S. J., & Zhou, J. (in press). Transformational leadership, conservation, and creativity: Evidence from Korea. Academy of Management Journal. Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420–428. Simonton, D. K. (1975). Sociocultural context of individual creativity: A transhistorical time-series analysis. Journal of Personality and Social Psychology, 32, 1119–1133. Simonton, D. K. (1984). Artistic creativity and interpersonal relationships across and within generations. Journal of Personality and Social Psychology, 46, 1273–1286. Speller, K. G., & Schumacher, G. M. (1975). Age and set in creative test performance. Psychological Reports, 36, 447–450. Staw, B. M. (1995). Why no one really wants creativity. In: C. M. Ford & D. A. Gioia (Eds), Creative Action in Organizations (pp. 161–166). Thousand Oaks, CA: Sage. Terborg, J. R. (1981). Interactional psychology and research on human behavior in organizations. Academy of Management Review, 6, 569–576. Tett, R. P., Jackson, D. N., & Rothstein, M. (1991). Personality measures as predictors of job performance: A meta-analytic review. Personnel Psychology, 44, 703–742. Tierney, P., & Farmer, S. M. (2002). Creative self-efficacy: Potential antecedents and relationship to creative performance. Academy of Management Journal, 45, 1137–1148. Tierney, P., Farmer, S. M., & Graen, G. B. (1999). An examination of leadership and employee creativity: The relevance of traits and relationships. Personnel Psychology, 52, 591–620. Tighe, E. (1992). The motivational influences of mood on creativity. Unpublished doctoral dissertation, Brandeis University. Tinsley, H. E. A., & Weiss, D. J. (1975). Interrater reliability and agreement of subjective judgments. Journal of Counseling Psychology, 22, 358–376. Torrance, E. P. (1974). Torrance tests of creative thinking: Directions manual and scoring guide. Englewood Cliffs, NJ: Prentice-Hall. Unsworth, K. (2001). Unpacking creativity. Academy of Management Review, 26, 289–297. Utman, C. H. (1997). Performance effects of motivational state: A meta-analysis. Personality and Social Psychology Review, 1, 170–182. Walker, A. M., Koestner, R., & Hum, A. (1995). Personality correlates of depressive styles in autobiographies of creative achievers. Journal of Creative Behavior, 29, 75–94. Weiss, H. M. (1977). Subordinate imitation of supervisor behavior: The role of modeling in organizational socialization. Organizational Behavior and Human Performance, 19, 89–105. Weiss, H. M. (1978). Social learning of work values in organizations. Journal of Applied Psychology, 63, 711–718. West, M. A., & Anderson, N. R. (1996). Innovation in Top Management Teams. Journal of Applied Psychology, 81, 680–693. Wiggins, J. S., & Trapnell, P. D. (1997). Personality structure: The return of the Big Five. In: R. Hogan, J. Johnson & S. Briggs (Eds), Handbook of Personality Psychology (pp. 737–765). San Diego, CA: Academic Press. Woodman, R. W., Sawyer, J. E., & Griffin, R. W. (1993). Toward a theory of organizational creativity. Academy of Management Review, 18, 293–321. Zhou, J. (1998a). Feedback valence, feedback style, task autonomy, and achievement orientation: Interactive effects on creative performance. Journal of Applied Psychology, 83, 261–276. Zhou, J. (1998b). Managers’ recognition of employee creative ideas: A social-cognitive approach. Paper presented at “The 21st Century Change Imperative: Evolving Organizations & Emerging Networks” Conference, Center for the Study of Organizational Change, University of Missouri-Columbia, June 12th–14th.

Research on Employee Creativity

217

Zhou, J. (2003). When the presence of creative coworkers is related to creativity: Role of supervisor close monitoring, developmental feedback, and creative personality. Journal of Applied Psychology, 88, 413–422. Zhou, J. (in press). Promoting creativity through feedback. In: C. Ford (Ed.), Handbook of Organizational Creativity. Hillsdale, NJ: Lawrence Erlbaum. Zhou, J., & George, J. M. (2001). When job dissatisfaction leads to creativity: Encouraging the expression of voice. Academy of Management Journal, 44, 682–696. Zhou, J., & Oldham, G. R. (2001). Enhancing creative performance: Effects of expected developmental assessment strategies and creative personality. Journal of Creative Behavior, 35, 151–167. Zhou, J., & Woodman, R. W. (2003). Managers’ recognition of employees’ creative ideas: A socialcognitive model. In: L. V. Shavinina (Ed.), International Handbook on Innovation. Hillsdale, NJ: Lawrence Erlbaum. Zuckerman, H. (1977). Scientific elite: Nobel laureates in the U.S. New York, NY: Free Press.

THE RELATIONSHIP BETWEEN WORK-FAMILY HUMAN RESOURCE PRACTICES AND FIRM PROFITABILITY: A MULTITHEORETICAL PERSPECTIVE
Michelle M. Arthur and Alison Cook
ABSTRACT
Few studies have investigated the relationship between work-family human resource practices and firm-level outcomes. Several organizational studies have addressed the antecedents to firm adoption of work-family initiatives; however, the majority of work-family research investigates the relationship between work-family practices and individual-level outcomes. The current paper begins by providing a critical analysis and synthesis of the extant work-family literature. In addition, we integrate the organizational learning research on firm commitment to work-family policies and the human resource model. We suggest that the level of firm commitment moderates the relationship between work-family policies, the human resource model, and firm performance. Several propositions for future work-family research are presented.

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 219–252 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22005-3

219

220

MICHELLE M. ARTHUR AND ALISON COOK

INTRODUCTION
One of the most significant changes in the last quarter of century is the entry and continued presence of women in the working population. The percentage of women that comprise the labor force has increased from 33% in 1977 to 46% in 2002 (Bureau of Labor Statistics, 2003). Concurrently, the structure of the family has changed. Dual-career marriages make-up 60% of families; these couples comprise 45% of the workforce (Catalyst, 2002). One-quarter of all U.S. households with children are headed by women alone (NCWO, 2002). Of these women, 63% with children under six are in the labor force (NCWO, 2002). As the demographic composition of the labor force and the structure of the family have changed, organizations have introduced several types of work-family programs to help employees deal with the work and family interface. Organizations often cite the need for productive, stable and satisfied employees as the underlying impetus to the adoption work-family programs (Mosanto, 2003). In response, researchers have examined the effects of work-family programs on psychological outcomes. Broadly, these studies have examined the relationship between work-family programs and employee affective, cognitive, and behavioral outcomes. The results suggest that work-family programs increase potential employee attraction, generate more productive employees, and reduce employee turnover. The implicit suggestion of psychological research is that work-family programs increase employee performance, hence firm profitability. However, there are few empirical papers linking work-family programs to firm-level outcomes. The majority of institutional studies have investigated firm characteristics as predictors of firm adoption of work-family programs. A few studies employing institutional and symbolic action theories have found some support for a relationship between work-family programs and firm value (Arthur, in press; Perry-Smith & Blum, 2000). However, these studies have focused on perceptual measures of firm performance such as company stock and perceived firm performance (e.g. employee speculation of future firm performance). To date, work-family researchers have investigated performance measures that rely on constituents’ expectations of firm performance. Since work-family programs have long been considered progressive, and most recent arguments suggest a “best practice,” a relationship beyond the perceptual level may exist. There is a lack of research investigating a relationship between work-family programs and actual firm performance. The purpose of the current paper is to develop several propositions linking workfamily programs, the human resource model, and firm-level outcomes. We begin with a discussion of the definition and evolution of work-family programs. Next, we provide an overview of the psychological and sociological literature addressing

A Multi-Theoretical Perspective

221

work-family conflict and programs. Following, we synthesize two streams of research, the organizational learning research and the human resource literature, to develop propositions based on the idea that firm commitment to work-family policies moderates the relationship between work-family programs and firm performance. A discussion of future work-family research and a brief conclusion follow. We define work-family programs broadly as any program designed to alleviate individual conflict between work and family. In general, work-family programs fall into three categories; dependent care, family leave, and flexible working hours. The earliest work-family programs focused on childcare concerns. During World War II, when women temporarily entered the workforce, the government provided childcare centers to increase the availability of women for work (Glass & Estes, 1997). After World War II, firms faced with increased demand for workers, followed suit and began to offer childcare centers. One of the first widely cited childcare centers was established at Stride Rite in 1971. Early work-family programs tended to focus on the single strategy of childcare centers. In the 1970s, the operationalization of childcare centers evolved from onsite childcare centers to numerous offshoots such as off-site resources, emergency childcare, inter-organizational consortia, referral services and voucher programs (Friedman, 1990). Similar to the evolution of childcare center programs, workfamily programs evolved to a broader definition. In the 1980s, firms tended to focus on family-leave strategies (Friedman, 1990). In doing so, firm programs included maternity leave, paid and unpaid, extended leave and the like. In the 1990s firms began to directly address the conflict presented by work-family tensions with flexible work arrangements such as telecommuting, job sharing, and alternative workweeks. Work-family programs have evolved substantively in the past thirty years. Concurrently, scholars from many disciplines have examined the work and family interface.

LITERATURE REVIEW
Researchers have explored the relationship between work and family from various perspectives outside of organizational studies. The topics investigated include the effects of a working mother or unemployed father on the family, such as examining the intellectual development of children when the mother works outside the home, or the loss of respect and status that may result when a father is unemployed (Bronfenbrenner & Crouter, 1983; Meneghan & Parcel, 1990); the effects of work-family conflict on children, for example, its influence on their social outcomes (Parcel & Menaghan, 1994; Perry-Jenkins et al., 2000); work-family issues

222

MICHELLE M. ARTHUR AND ALISON COOK

within political and legal domains, such as the arguable necessity of government involvement in order to proliferate work-family initiatives (Edwards, 2001; Ferber & O’Farrell, 1991; Guthrie & Roth, 1999; Haas, 1999; Kelly & Dobbin, 1999; Perry-Jenkins et al., 2000); and cross-cultural comparisons of work-family conflict (Yang et al., 2000) among others. However, the next section of this paper is limited to work-family issues within the domain of the organization; a stream of research that builds on the foundation set forth by Kanter’s (1977) seminal work.

Psychological Framework Examining work-family issues at the individual-level of analysis, scholars typically focus on one of two paths. One avenue of research investigates the conflict individuals face as a result of involvement in multiple roles, and its effects on the individual (Bacharach et al., 1991; Barling & MacEwen, 1991; Cooke & Rousseau, 1984; Greenhaus, Parasuraman & Collins, 2001). The other avenue of research examines the implementation of work-family policies and their effects on the individual (Lambert, 2000; Roehling et al., 2001; Shinn et al., 1989; Tausig & Fenwick, 2001). The outcomes measured often coincide; both lines of research are interested in constructs such as job satisfaction, organizational commitment, withdrawal, and work performance, to name a few. The notable difference in the two streams of research is the variables investigated between the work-family conflict and employee behavior relationship. For example, in explaining intentions to turnover, scholars within the role conflict perspective (Grandey & Cropanzano, 1999) have examined role stress as a predictor, whereas in the policy perspective (Grover & Crooker, 1995), scholars have examined an employee’s accessibility to work-family policies. Currently, the research is limited to integrating these two perspectives; nonetheless, some scholars are making inroads by testing mediating variables of the work-family policy to employee outcome relationship (Lambert, 2000; Thomas & Ganster, 1995). The first part of the psychological literature review focuses on role theory and its derivatives. The latter part of the literature review focuses on work-family programs. Role Theory Perspective Role theory (Katz & Kahn, 1978) provided the theoretical underpinnings for several of the early investigations of work-family conflict (e.g. Cooke & Rousseau, 1984). According to the theory, individuals are involved in multiple roles in the family and work domains. These multiple roles have the potential to cause strain as a function

A Multi-Theoretical Perspective

223

of two mechanisms. Given the likelihood that the roles in each domain do not coincide, role overload and/or interrole conflict may result. Role overload occurs when the multiple role expectations cause one to feel overwhelmed as a result of increased workload. Interrole conflict results from the lack of compatibility among the roles. One’s resources of time and energy are limited; therefore, fulfillment of one role may prevent fulfillment of another role. Cooke and Rousseau’s (1984) results support role theory with regard to work-role expectations. The multiple roles led to stressors (work overload and interrole conflict), and those stressors led to strain. Though not routinely cited in the literature, role theory offers a basis from which other theories and models have developed. Greenhaus and Beutell (1985) expanded on role theory by proposing underlying dimensions of interrole conflict. Their model includes time-based pressures, strain-based pressures, and behavior-based pressures placed on the roles within both the work and the family domains. Additionally, they noted the variation possible in the intensity of the conflict given the salience of the role and the strength of the penalty for noncompliance. Therefore, Greenhaus and Beutell (1985) suggest factors that reduce the negative consequences of interrole conflict will likely lessen role pressures. Burke (1988), using a similar framework of work stressors and non-work stressors as antecedents of work-family conflict, investigated numerous variables that may be affected. Consistent with Greenhaus and Beutell’s (1985) proposition of time-based pressure, Burke (1988) found shiftwork resulted in higher levels of work-family conflict. Shifting the balance slightly from the Greenhaus and Beutell (1985) model, though, Burke (1988) noted work characteristics and work stressors had a greater effect on work-family conflict than non-work stressors. Furthermore, in his sample of over eight hundred police officers, he found work-family conflict to be significantly related to intentions to turnover, job burnout, and lower job satisfaction. Confirming this relationship, Bacharach et al. (1991) found that work-family conflict, indeed, was related to lower job satisfaction and job burnout. Other researchers have also supported similar findings. Kossek and Ozeki (1998), for example, conducted a meta-analysis of numerous studies that investigated work-family conflict and its relationship to job satisfaction. Most studies in their sample analyzed the conflict in the direction of work-to-family rather than family-to-work; nonetheless, the results in the studies are noticeably consistent with regard to the job satisfaction, work-family relationship. Adams et al. (1996) also examined the work-family conflict, job satisfaction relationship; however, they proposed a model integrating work-family conflict and social support.

224

MICHELLE M. ARTHUR AND ALISON COOK

Involvement in Roles The Adams et al. (1996) model is predicated on the belief that the work and family relationship can concurrently experience support and conflict. In addition to a standard work-conflict measure, they also incorporated emotional support and involvement in their model. Therefore, they assessed job and family involvement; instrumental and emotional support at home; family to work interference and work to family interference; and job and life satisfaction. Adams et al.’s (1996) results support the additions to the model. It is already established that the work and family relationship affects job satisfaction, yet their expanded model offers further insight. The level to which an individual is involved in a role affects that association. High levels of job involvement are associated with increased job satisfaction, as well as increased interference of family by work. High levels of family involvement are associated with increased life satisfaction. The initial model incorporating involvement was proposed by Frone et al. (1992) to predict depression. Although Adams et al. (1996) borrowed from this earlier model, each offers distinct contributions to the work-family field. A primary contribution of the Frone et al. (1992) model was the bidirectional nature of the work-family, family-work construct. Findings have substantiated the influence of family stressors on family to work conflict; suggesting the effect family has on work life. Adams et al.’s (1996) model notes the importance of role involvement. Their findings support that family and job involvement affect the relationship between work and family. Enrichment and Depletion Level of involvement also played a role in Barnett and Baruch’s (1985) study investigating women in multiple roles and the effect it had on stress. They based their work on two opposing views, the scarcity hypothesis and the expansion hypothesis. The scarcity hypothesis contends the greater the number of roles a person is involved with, the greater the role strain they may experience. The expansion hypothesis, on the other hand, asserts as a person accumulates more roles the role strain lessens. The difference results from the quality of experience the person has with the given role. Within this study, the quality of experience was determined by the disparity between rewards and concerns reported for each role; wife, mother, and paid worker. Barnett and Baruch (1985) found the role of a paid worker was not related to their indicators of stress. Their results support the expansion, rather than the scarcity hypothesis, contrary to typical beliefs of multiple roles. The implication is that not all roles are equal in how they affect individuals. The authors suggest women who are not employed experience some deleterious effects resulting from a lack of structure and legitimacy. In the same vein, Hochschild (1997) contends many women prefer to spend time in their

A Multi-Theoretical Perspective

225

work role rather than their family role because of the structure and consistency of rewards and appreciation. In essence, these findings assert the additional role of a paid worker is energizing and enriching rather than depleting one’s resources. Lending credence to the expansion hypothesis, Gutek and her colleagues (1991) explored both the rational perspective and the gender-role perspective. The rational view proposes a positive linear relationship between hours worked and workfamily conflict. The gender-role view predicts gender will influence the interaction with the number of hours worked and the resulting conflict. The results of this study, to some extent, offer a variation from the assertion that more roles create added benefits, but they do suggest that additional roles, if in the “sex-appropriate” domain, will not likely increase work-family conflict. A study of performance by Barling and MacEwen (1991) surveyed female employees to determine the effects of work-family conflict on performance, controlling for their level of satisfaction and commitment to being a mother. A negative indirect effect on performance via attention and concentration difficulties was confirmed. Rothbard (2001), pursuing the concepts of scarcity and expansion, developed a model of enrichment or depletion as a result of involvement in multiple roles. In her model, the enrichment or depletion process occurs through an emotional response mechanism. Depletion results in role strain and conflict. Enrichment, on the other hand, argues against fixed resources and claims multiple roles will result in added benefits. Her results demonstrate support for both arguments. In fact, women experienced both depletion and enrichment. In the direction of work to family, they experienced depletion; family to work, enrichment. Interestingly, men did not experience depletion, and they only experienced enrichment in the direction of work to family. Rothbard, noting the gender differences found, suggests work and family domains are likely more permeable for women than for men. Viewing depletion in another perspective, Grandey and Cropanzano (1999) proposed the conservation of resources theory (Hobfoll, 1989) as a potential framework from which work-family researchers could base future work. The model’s premise is that individuals value resources. Threats to the resources one possesses, or to the resources one expects to possess, will result in stress. For an example, the conservation of resources model predicts multiple roles will result in stress developing from the resources lost (extra time spent) attempting to balance the roles. Findings were supportive of the model. Grandey and Cropanzano (1999) examined sources and consequences of resource drain over time. Their results did, indeed, support the drain of resources by both work and family stressors. Role stress at home affected family to work conflict, just as role ambiguity and conflict at work affected work to family conflict. Thus, family and work satisfaction deteriorated. This deterioration exemplifies the drain of resources. Eventually, as a response to replenishing needed resources, intentions to turnover may develop.

226

MICHELLE M. ARTHUR AND ALISON COOK

Social Identity and Role Investment Withdrawal intentions have also been investigated within the social identity framework (Tajfel & Turner, 1985). Greenhaus et al. (2001) examined the level of role investment with work-family conflict and withdrawal. He and his colleagues found, consistent with social identity theory, the salience of the role affects an individual’s reaction to the conflict. For instance, if one is highly invested in one’s career and not as invested in the family, work-to-family conflict will most likely be ignored. On the other hand, if one is not highly invested in one’s career and is more invested in the family, work-to-family conflict will likely lead to turnover. Only work-to-family conflict resulted in turnover intentions; family-to-work conflict did not. This finding is expected given that family-to-work conflict is generated from stressors at home, and leaving the job would not likely reduce the stress. Within both the social identity approach and the utilitarian approach, Lobel (1991) explored the differences between the underlying causes of work-family conflict. The utilitarian approach perceives value of the role in terms of costs and benefits. If one perceives valued rewards for greater role involvement, investment will increase. Conflict will develop as the investment in the work and family domains becomes more equivalent. As the involvement in the roles becomes more uneven, work-family balance will increase. The social identity theory, in contrast, predicts an increase in role investment will correspond to a greater sense of identity within the role. Costs and benefits are not predictors of role investment within the social identity framework. The salience of the role determines investment. Work-family conflict results when the two roles become less aligned in their underlying values. The more aligned, the greater the work-family balance. Spillover, Compensation and Segmentation The boundary between family and work has become permeable; therefore, the assertion that work and family affect each other is largely accepted in the workfamily literature. Spillover and compensation both contend work flows into family and vice versa. However, the difference is that compensation asserts a deficiency in one domain will cause an individual to compensate for it in the other domain. Therefore, work may not be going well, so an individual will attempt to gain satisfaction from the family domain. Spillover, on the other hand, contends feelings in one sphere will permeate into the other sphere. A positive outcome in one will remit a positive to the other. Segmentation theory stresses separating the work and family domains. As noted by Edwards and Rothbard (2000), segmentation is now considered a deliberate strategy an individual may develop in order to keep the two domains from overlapping. Spillover: whether referred to as a theory (Grandey & Cropanzano, 1999), a hypothesis (Sumer & Knight, 2001), or a linking mechanism (Edwards &

A Multi-Theoretical Perspective

227

Rothbard, 2000), is strongly supported in the work-family literature. Williams and Alliger (1994) examined mood spillover between work and family. Results suggest negative moods have greater spillover than positive moods, and women exhibit more spillover in both directions than men. Doby and Caplan (1995) also investigated spillover. Examining organizational stressors, they found work-experienced anxiety spills over to home-experienced anxiety. Furthermore, spillover is more likely to occur for high-threat stressors than low-threat stressors. Applying a slightly different tactic, Sumer and Knight (2001) analyzed spillover through an individual differences approach. Using attachment theory as their basis, they found differences in how people experience the work-family relationship. A personality perspective suggests some people have difficulty discriminating between the two domains, resulting in maximum spillover. Initially, role theory provided theoretical guidance into the relationship between work and family. Increased scholarly interest spurred investigations of other theoretical frameworks. No single theory has dominated the literature in this area (Edwards & Rothbard, 2000). Nonetheless, there seems to be some consensus in the area, that work-family conflict, absent any support, may lead to organizational outcomes such as turnover, lower commitment, job burnout, lower job satisfaction, decreased job performance, and the like. Newer investigations have led scholars to examine asymmetric predictions of role theory for men and women. Furthermore, scholars have not only examined the spillover from work to family but also from family to work. The idea of increased social support at home as an intervening variable between work and family conflict lends itself to the idea that support at work may also intervene in what would otherwise be considered a seemingly consistent relationship between increased work and family conflict and increased withdrawal type behaviors.

Human Resource Policies The human resource perspective is predicated on the idea that human resource programs, such as work-family programs, will increase employee productivity. It is an efficiency-based perspective that suggests organizations’ policies and practices can be wholly explained as a drive toward efficiency or profits. That is, the benefits of work-family programs will outweigh the costs. The basic assumption is that, if an organization helps employees balance work and family, organizations will be better able to attract, retain, and extract increases in productivity from current employees. Scholars have investigated several individual-level outcomes resulting from organizational interventions such as dependent care, family leave policies and flexible hours. While the results of studies have been, in some cases mixed,

228

MICHELLE M. ARTHUR AND ALISON COOK

on the whole the results suggest work-family programs provide employees with the infrastructure to become more efficient. Flexible Scheduling Within the realm of work-family policies, flexible scheduling is the most diffuse. However, unlike other work-family policies, the exact cost of implementing flexible scheduling is difficult to quantify. Benefits to the organization, though, have been substantiated. Christensen and Staines (1990) suggest flextime programs decrease withdrawal behaviors such as tardiness, absenteeism, and turnover, while Dunham et al. (1987) found minor, positive effects for several general reactions as a result. Attendance and turnover were also investigated by Dalton and Mesch (1990). Though no effect was found for turnover, absenteeism dramatically decreased with flexible scheduling. A finding also supported in an Israeli study conducted by Krausz and Freibach (1983). Dalton and Mesch (1990) suggest their findings are consistent with an “ability-to-attend” dimension. Flexible scheduling offers control of time to employees, which, in turn, potentially decreases external conflicts (Tausig & Fenwick, 2001). The practical effects of flexible scheduling are consistent with lower absenteeism, yet the theoretical implication does not support a relationship with turnover. An indirect link to turnover, however, has been suggested through the mechanism of job satisfaction. Several scholars have affirmed a relationship between flexible schedules and job satisfaction (Christensen & Staines, 1990; Thomas & Ganster, 1995). Grover and Crooker (1995), using a national sample, found a positive relationship between flexible scheduling and affective commitment. This is consistent with Mellor et al. (2001) who suggest organizations should offer work-family policies to help their employees feel less “trapped”; therefore, building affective rather than continuance commitment. With affective commitment, employees accept the values and beliefs of the organization and have a desire to maintain membership and put forth effort; whereas with continuance commitment, employees have developed “side bets” and remain with the organization due to their lack of alternatives that are able to match what they have accumulated (Meyer & Allen, 1997). Pierce and Newstrom (1982), investigating workers in insurance firms, found flexible schedules positively influence organizational commitment and job involvement. Extending the Grover and Crooker (1995) study, Roehling et al. (2001) assessed employee loyalty as a function of flexible scheduling. Additionally, they introduced the variables of life stage and gender. Their findings affirm that flexible scheduling affects employee loyalty for both men and women at all life stages. The positive results for all life stages, as well as for men, are somewhat surprising given the belief that those experiencing greater work-family conflict

A Multi-Theoretical Perspective

229

should be more affected by work-family policies. Nonetheless, the positive relationship demonstrated between flexible scheduling and employee loyalty should be noted. Dependent Care Contrary to the consistent findings of flexible scheduling research, studies examining dependent care interventions have found mixed results. In Grover and Crooker’s (1995) study, childcare information services were related to affective commitment, yet childcare assistance did not yield a significant effect. Other scholars affirm a positive relationship between childcare services and organizational commitment (Goldberg et al., 1989; Youngblood & Chambers-Clark, 1984). In an extension of the past work, Roehling et al. (2001) found a positive relationship between childcare policies and employee loyalty, but not as expected. They hypothesized that the individuals who benefit the most from the policy (women with preschoolaged children) would demonstrate the greatest employee loyalty. This was not confirmed, however. Women with school-aged children, rather than women with preschool-aged children reported higher loyalty scores. A plausible explanation is that women with school-aged children have more variable needs with regards to childcare; therefore, the services provided by the organization are particularly helpful to them. Kossek and Nichol (1992) studied the effects of an on-site childcare center. Although the center did not affect performance or absenteeism, it did positively affect attitudes and behaviors such as recruitment and retention. These findings mirror an earlier study by Goff et al. (1990). The authors did not find a link between on-site childcare centers and absenteeism, but suggest on-site centers are important in recruitment and retention. This consistent finding is understandable given that most childcare centers are not generally equipped to take care of sick children; therefore, parents must still be absent in specific situations. Additionally, Kossek and Nichol (1992) introduced an idea termed the “frustration effect.” They suggest employees become frustrated when their needs at the childcare center cannot be accommodated; therefore, diminishing the value of the childcare center for those individuals. Exploring the “frustration effect,” Rothausen et al. (1998) conducted a study within an organizational justice framework. The assertion is that employees wishing to, but not being able to use the center should perceive it as less fair. Additionally, equity norms may be a consideration for those who do not directly benefit from the center. Rothausen et al. (1998), though, did not find a backlash. They found current users, past users, and potential future users to have a more positive attitude toward the childcare center than the general worker, but there was no frustration effect. At most, non-positive attitudes were proximal to the center. In the same regard, the positive attitudes

230

MICHELLE M. ARTHUR AND ALISON COOK

reported by the users were strongest when they pertained to the childcare center. General work attitudes were not significantly related. Dependent care programs have been more consistently related with turnover than with absenteeism. In a study of elder care assistance, Denton et al. (1990) suggest the organizational intervention encourages employee retention. Grover and Crooker’s (1995) results support a significant negative relationship between childcare information services and intentions to turnover. And further, Youngblood and Chambers-Clark (1984) found that corporate sponsored childcare centers were linked to lower turnover intentions. Family Leave Policies The number of organizations offering maternity leave and other family-related leave policies has increased substantially over the past quarter century. The Family and Medical Leave Act of 1993 has propelled this trend. As noted by Haas (1999), though, exclusions in the legislation prevent all employees from being covered. Firms with fewer than 25 employees are allowed to prohibit family leave, and “key” employees of any size firm may also be denied the option. Moreover, it is often not possible for an employee to bear the financial hardship of unpaid leave, essentially rendering this benefit a moot point to them. Though not all organizations allow family leave, especially paid leave; benefits have been demonstrated. Grover and Crooker (1995) found that maternity leave was significantly related to affective commitment, as well as lower intentions to turnover. Offshoots of typical work-family studies have developed. Recent studies have examined supportive climate as a moderator of the work-family program and employee behavior relationship. These studies address the prominent idea emerging in the work-family literature that firms that are not truly committed to work-family programs will hinder employees from using the benefits. In doing so, the firm does not gain the efficiencies or benefits associated with the work-family programs.

Supportive Climate Organizations, as a means of reducing work-family conflict and achieving firm benefits or efficiencies, may exhibit supportive measures beyond adoption of work-family programs. Examining emotional support, Thomas and Ganster (1995) investigated the effects of supervisory support on work-family conflict, absenteeism, job satisfaction, as well as various health issues. They concluded that, indeed, a supportive workplace yields benefits. Results suggest a direct benefit of supervisory support on job satisfaction. Additionally, indirect relationships to both

A Multi-Theoretical Perspective

231

job satisfaction and depression were supported through a work-family conflict link. Kossek and Ozeki (1998), in their meta-analysis, confirmed a consistent negative relationship between work-family conflict and job satisfaction. The importance of supervisory support on work outcomes has been consistently positive (Roehling et al., 2001; Shinn et al., 1989). Expanding beyond the notion of supervisory support, Nielson et al. (2001) explored mentoring as a means of lessening work-family conflict. They proposed mentoring as a unique form of social support within the organization. Findings support their claim that a supportive mentor reduces work-family conflict. Although more significant findings were reported in the direction of family-to-work conflict than work-to-family conflict, each was negatively related to supportive mentorship. An interesting correlation that deserves mention, though, is that supportive mentoring was also negatively related to the number of hours worked. Hours worked strongly influences work-to-family conflict, but it does not affect family-to-work conflict; however, hours worked and job performance may be related and is worth exploration. Other studies explored job performance as a function of work-family interventions finding mixed results (Judiesch & Lyness, 1999; Kossek et al., 2001; Kossek & Nichol, 1992). Kossek et al. (2001) also assert the importance of a supportive work climate. Incorporating variables regarding dependent type and climates in the work and family domains, they suggest a supportive work climate positively affects job performance; a negative work climate hinders job performance. Affective commitment, work-family conflict, and intentions to turnover were also affected by work-family climate (Thompson et al., 1999). Thompson et al. (1999) investigated work-family culture and its effect on organizational attachment, benefit utilization, and work-family conflict. Their findings affirm a positive link to affective commitment and benefit usage, and a negative link to intentions to turnover and work-family conflict. Due to a lack of theory beyond an efficiency-based framework in the workfamily policy perspective, scholars have recently investigated social exchange theory as a theoretical underpinning to the work-family adoption and employee behavior relationship. Social exchange theory provides a solid theoretical framework for future work-family research. Social Exchange Social exchange theory predicts a relationship based on mutual, though not equal reciprocation (Gouldner, 1960). Predicated on the assumption one does not like to feel obligated, social exchange theory suggests when organizations provide work-family benefits to their employees, the employees will reciprocate with a positive action. It has been suggested that signaling theory would contradict this

232

MICHELLE M. ARTHUR AND ALISON COOK

assertion (Lambert, 2000). Signaling theory’s basic premise is that actions on part of the organization have the potential to signal to the employees that they are valued. The point in contention is whether a general work-family policy available to all employees has the potential to signal value. Signaling theory contends for employees to feel valued, the action must specifically target them. Lambert (2000), however, believes employees will perceive greater organizational support as a result of the work-family policies and react positively. Other scholars also support that conclusion; when organizations implement work-family policies irrespective of their target, all employees have the potential to feel valued (Drago et al., 2001; Grover & Crooker, 1995). As a direct measure of the obligation felt by employees, due to social exchange, Lambert (2000) examined the impact of work-family policies on organizational citizenship behaviors. She also investigated perceived organizational support as a mediating variable. Results support the impact of work-family policies on organizational citizenship behaviors, consistent with social exchange theory. Surprisingly, however, an increase in organizational citizenship behaviors was not a function of perceived organizational support. This finding seems contrary to expectations and worth further exploration. Nonetheless, the benefits realized for the organization were a function of the work-family policies and their direct effect on employees’ willingness to exhibit citizenship behaviors. In spite of the often-mixed results, work-family policy studies do not show an increase in negative employee behaviors. For example, studies may not find a significant negative relationship between childcare centers and absenteeism, but they also do not find an increase in absenteeism resulting from childcare centers. Therefore, if firm-level effects are positive, an interaction of individualand firm-level effects should produce a significant, incremental positive effect on firm performance. The following discussion focuses on firm-level studies of work-family programs.

Sociological Framework The primary stream of research within the sociological framework examines the characteristics of organizations that make them more likely to adopt work-family initiatives. The variation among studies occurs in which antecedents scholars believe drive the adoption decision. Examining the mechanisms that influence organizations to respond to work-family pressures and adopt progressive initiatives dominates firm-level work-family research. Most scholars draw on institutional theory (DiMaggio & Powell, 1983), resource dependence theory (Pfeffer & Salancik, 1978), organizational adaptation theory (Daft & Weick, 1984), or a

A Multi-Theoretical Perspective

233

combination to explain firms’ actions. Organizational learning (Argyris & Schon, 1978) and management dominant logic (Prahalad & Bettis, 1986) have also been proposed as useful frameworks for addressing this issue. Institutional Theory According to institutional theory, organizations may adopt work-family initiatives in order to gain legitimacy. The adoption occurs regardless of possible operational efficiencies. Legitimacy occurs as a result of an organization’s alignment with its environment. Three pressures are detailed in institutional theory that may influence this alignment, or isomorphism: normative, coercive, and mimetic (DiMaggio & Powell, 1983). Normative pressures may be exerted by professional organizations or groups that want the organization to be more legitimate. An example is a group of female employees pressuring an organization to implement a childcare center, not necessarily because they need it, but because comparable occupations outside of their organization offer it and implementation of the center would legitimize their job. Coercive pressures may be exerted by those in power, often deriving from the control of resources or backing of the government; an example, organizations granting maternity leave only after legislation has been passed. Mimetic pressures develop when the environment is changing and the organizations seek legitimacy by mimicking successful competitors in their industry. For instance, in order to appear family-friendly, the organization may implement work-family policies comparable to those of their competitors. This symbolizes to the environment that they are legitimate; therefore, appearing acceptable to customers and potential employees. Organizational responsiveness to the adoption of work-family policies is susceptible to all institutional pressures. Coercive pressure has influenced the adoption of paid and unpaid maternity leave policies examined the trend of organizations to offer maternity leave programs (Guthrie & Roth, 1999; Kelly & Dobbin, 1999). They assert these programs were adopted as a response to governmental pressure. Although many programs were implemented prior to the Family and Medical Leave Act of 1993, Kelly and Dobbin (1999) have linked that implementation to prior government actions. Noting the Civil Rights Act of 1964 and the EEOC’s administrative ruling in 1972, they examined the spread of maternity leave programs and government interventions over a 30-year period. Some analysts, however, claim the increase of maternity leave programs resulted from market pressures to remain competitive in a changing labor force. Nevertheless, Kelly and Dobbin (1999) conclude from their analysis that it was the interventions by the government that aided in the proliferation of maternity leave programs. Guthrie and Roth (1999) also examined the implementation of maternity policies; however, they tied maternity leave to paid sick leave. This link was

234

MICHELLE M. ARTHUR AND ALISON COOK

made with the assumption that women, during pregnancy and childbirth, deviate from their healthy body. They proposed that paid sick leave was a mechanism by which organizations could effectively include pregnant women in existing policies. With this link, Guthrie and Roth’s (1999) findings regarding paid sick leave are in concordance with their underlying logic for paid maternity leave. Consistent with coercive pressure, they found organizations governed by states including pregnancy as employment discrimination were more likely to offer paid sick leave. Contrary to Kelly and Dobbin (1999), though, they suggest some organizations may take an active role in offering paid maternity leave or paid sick leave. Guthrie and Roth (1999) found support that factors such as a female CEO or a large proportion of women in the industry were related to paid sick leave being offered in the organization. It should be noted, however, if they had not linked maternity leave with sick leave, their conclusions would not have been supported. Integration of Resource Dependence Theory Consistent with resource dependence theory (Pfeffer & Salancik, 1978), other scholars have also suggested the active role of management as an antecedent to firm adoption (Goodstein, 1994; Ingram & Simons, 1995; Milliken et al., 1998). Resource dependency refers to the acquisition of resources essential to survival. In the realm of work-family, firms adopt initiatives as a survival response to ascertain needed resources in the changing environment. In other words, with more women in the workforce, organizations need to actively react in order to secure qualified human resources. Although many scholars believe resource dependence theory is relevant to the issue of work-family policy adoptions, they acknowledge its influence in concurrence with institutional theory. Suggesting organizations are able to strategically respond to institutional pressures, Goodstein (1994) applied Oliver’s (1991) framework to assess organizational adoption of childcare programs and flexible scheduling. Oliver (1991) proposes five factors in determining institutional response: cause, constituents, content, control, and context. Cause refers to normative pressure. Goodstein (1994) suggests that larger organizations, as a result of their visibility, will receive more normative pressure to address work-family issues. Given the importance of legitimacy for large organizations, they will likely comply. Constituents represent important groups to the organization. Within the realm of work-family, this is a key concern if the organization is dependent on a female labor force. Work-family issues are particularly salient for working women; therefore, Goodstein (1994) suggests that organizations dependent on female employees and parents will be more responsive to family-friendly policies. Content is the composition of institutional pressures. When institutional demands are aligned

A Multi-Theoretical Perspective

235

with organizational goals, compliance is more likely. The contention is public, rather than private sector organizations will be more responsive. This may be a result of less pressure to thrive economically, or it may be a voluntary diffusion on part of the organizations to exhibit legitimacy. Control represents the pressure faced by an organization to conform to the norms set by its competitors. If an organization does not conform (i.e. adopt a similar work-family initiative), it risks its legitimacy. Context represents the interconnectedness of the organizational field. The more interconnected the organizations, the greater the institutional pressure to conform. Goodstein’s (1994) findings confirm that organizational response varies depending on the pressure exerted. For instance, he found organizational size and responsiveness to be highly related; as well as responsiveness to be greater in organizations highly dependent on female employees. Additionally, consistent with control, mimetic pressures were greatest within the same industry and geographic location. Replicating Goodstein’s (1994) study, Ingram and Simons (1995) corroborated most findings; however, some discrepancies were found. Unlike Goodstein’s (1994) study, Ingram and Simons (1995) found support for the hypothesis that public rather than private sector organizations are more responsive. Also, they added a refinement to the test of constituency. In addition to measuring response as a function of the percentage of females in the organization, they also measured the response as a function of the percentage of female managers. Only the latter was found to be significant. Nonetheless, results from both Goodstein’s (1994) study and Ingram and Simons’ (1995) study support the integration of a strategic choice perspective with institutional theory. Managerial Interpretation Not discounting the importance of resource dependence theory and institutional theory as determinants of organizational responsiveness, Milliken et al. (1998) suggest managers’ interpretations of environmental concerns also affect responsiveness towards adoption independent of the two theories. Their findings support the assertion that managerial interpretation plays an important role in organizational actions. Benefits were more likely offered in organizations when the managers interpreted them as relevant and tied to productivity. Milliken et al. (1998) suggest different managers may react differently given the same resource limitations and institutional pressures. The variance results from their interpretation of the issue. Osterman (1995) analyzed adoption of work-family programs within the framework of the organization’s employment strategy. As with the Milliken et al. (1998) study, Osterman (1995) found organizations were more likely to adopt programs if they perceived them as relevant and tied to productivity. In Osterman’s (1995)

236

MICHELLE M. ARTHUR AND ALISON COOK

framework, however, the ties to productivity are indirect. Adoption of work-family policies occurs in response to increasing problems of withdrawal or as an attempt to build organizational commitment. He suggests high-commitment work systems as an employment strategy is a strong predictor of adoption. For this strategy to be effective, employee commitment is a necessity. Management chooses to adopt work-family initiatives with the belief the overall level of commitment will be enhanced. Management dominant logic (Prahalad & Bettis, 1986) has also been integrated with institutional theory to predict responsiveness of organizations to adopt workfamily initiatives (Kossek et al., 1994). Prahalad and Bettis’s (1986) framework suggests managers develop “dominant logic” based on their past experiences. Demographic variables as well as environmental influences may affect a person’s approach. Kossek and her colleagues (1994), examining adoption of childcare initiatives, suggest managerial logic may influence an organization’s conformity or lack thereof to institutional pressure. Results confirmed managerial influence in the adoption of childcare programs; when the human resource manager perceived the program favorably, as well as perceiving executive attitudes to be positive, conformity to institutional pressures to adopt was most likely. Organizational Adaptation Expanding on Daft and Weick’s (1984) organizational adaptation model, Milliken et al. (1990) proposed a framework in which organizations recognize and interpret changes in order to better align themselves with the environment. Their model is comprised of five phases: scanning, noticing, interpreting, choosing, and learning. Scanning is the process of gathering relevant information regarding changes in the environment. Noticing is the process of allocating time and energy to the issues that may surface as a result of the change. The third step, interpretation, is an analysis of the issue. Significance to and implications for the organization are assessed. Some issues may not be interpreted to be significant factors for the organization. Choosing which issues need response and which can be discarded is the fourth step. The model is complete at the learning phase in which the organization selects a fitting response to the issue. Implementing Milliken et al.’s (1990) model, Goodstein (1995) examined organizational responsiveness to eldercare interventions. Consistent with the proposed model of organizational adaptation, he suggests organizational responsiveness will result from the recognition and interpretation of change in the environment. Eldercare is a recent, yet significant concern for many working adults. The U.S. Census Bureau (2000) reports more than 34 million Americans are currently 65 and older, and by 2020, that number is projected to exceed 53 million. Furthermore, 42% of workers in 2002 have been estimated as being responsible

A Multi-Theoretical Perspective

237

for elder care (Working Woman Magazine, 2000). Goodstein (1995) investigated the internal and external awareness issues of eldercare, the interpretation of potential implications of the issue, the scope of existing work-family benefits offered, and organizational demography. Surprisingly, age and gender did not influence organizational responsiveness. The findings did confirm the importance of external and internal awareness, as well as the breadth of work-family programs offered. Additionally, if involvement in eldercare was perceived to impact productivity or organizational commitment, organizations were more likely to respond. This study illustrates the importance of interpretation in organizational adaptation. Morgan and Milliken (1992) conducted an exploratory study to determine key factors of organizational response. Although not set within any specific theoretical framework, they explored dynamics that pertain to the theories discussed above. As indicated by organizational adaptation and managerial interpretation, their findings of internal awareness and perceptions of productivity are consistent with Goodstein’s (1995) study. Consistent with institutional theory, Morgan and Milliken (1992) found geographic location, size of the organization, and industry are all determinants of responsiveness. Extent of unionization was not significant in their study and yielded mixed results in the Kossek (1987) analysis. Although not finding support, their investigation of percentage female is consistent with resource dependence theory. Organizational Learning Lee et al. (2000) explored organizational responsiveness to reduced-load work in the framework of organizational learning (Argyris & Schon, 1978). They suggest three paradigms: accommodation, elaboration, and transformation; viewed on a continuum from exploitation to exploration. Exploitation increases efficiency by reducing variety, whereas exploration increases variety by taking risks. Organizations need to both exploit and explore in order to adapt to a changing environment. From their analysis, Lee et al. (2000) determined noticeable differences among firms in their interpretation and implementation of reduced-load work. They classified these differences as accommodation, elaboration, or transformation. Accommodation represents minimalist organizations. Such firms treat a request for reduced-load work on a case-by-cases basis, not something worthy of a new routine (Lee et al., 2000). Elaboration represents supportive organizations. They typically had formal policies supporting reduced-load work; however, employees who take advantage of this policy may limit their career potential. A fast-moving organization would be representative of the transformation paradigm. Regardless of formal policies, reduced-load work is accepted and viewed as a measure taken to retain quality employees. These firms accept and are amenable to change.

238

MICHELLE M. ARTHUR AND ALISON COOK

Firm-Level Outcomes As demonstrated by scholars, perceptions of productivity may affect firms’ responsiveness to adopt work-family policies (Gonyea & Googins, 1993; Goodstein, 1995; Morgan & Milliken, 1992). Few scholars have examined the link between work-family initiatives and firm-level outcomes; however, addressing the bottom line is a compelling argument. Similar to research in the psychological framework, this recent direction attempts to link adoption of work-family initiatives to organizational benefits. However, the benefits are realized at the firm-level rather than at the individual-level. For instance, scholars have investigated links between work-family initiatives and firm-level outcomes such as perceived firm performance and firm value. Perry-Smith and Blum (2000) investigated the link between work-family policies and perceived firm-level performance, and Arthur (in press) explored the link between work-family policies and firm value. Perry-Smith and Blum’s (2000) study offers a connection between bundles of work-family initiatives and benefits to the organization. In line with strategic human resources, their analysis focuses on work-family bundles instead of single initiatives. Bundles are comprised of complementary policies that are aligned with the organization’s values. Work-family bundles likely include dependent care services, flexible scheduling, and referral services. Perry-Smith and Blum (2000) assert bundles are more suitable for examining firm-level outcomes since they assure a broader effect. Their arguments are framed in resource-based and symbolic action perspectives. The symbolic action perspective (Pfeffer, 1981) asserts actions by the organization, such as offering work-family initiatives, signal to the employees they are valued. In return, employees put forth extra effort and performance is increased. Within the resource-based view, Perry-Smith and Blum (2000) contend workfamily bundles are difficult to imitate and, therefore, have the potential to create a sustainable competitive advantage. Their findings support both perspectives in that there is a positive relationship between perceived firm-level performance and bundles of work-family initiatives. They divided firm-level performance into profit-sales growth, market share, and organizational performance. All three performance measures were positively related to organizations that offered a wide range of work-family policies. A shortcoming exists, though, in that all measures of performance are perceptual. Nonetheless, this study has a strong impact. Not only do Perry-Smith and Blum (2000) link work-family policies to firm-level performance, but they also assert the possibility of creating a sustainable competitive advantage. Arthur (in press), framing her argument in institutional theory, linked workfamily initiatives with firm value. Her underlying assumption is that organizations

A Multi-Theoretical Perspective

239

will be better able to acquire scarce resources as a result of the legitimacy gained by adoption of work-family policies. Realizing the potential benefits to an organization, constituents of the firm, at the time of the work-family initiative announcement, will react to influence the share price; hence, increasing the firm’s value. Additionally, Arthur (in press) examined industry characteristics such as percentage female, unemployment rates, and high-tech classifications as potential moderators of this relationship. Her findings support the assertion that work-family initiative announcements positively affect firm value. Moreover, high-tech industries and industries with a large percentage of female employees were found to moderate the relationship. This study magnifies the importance of work-family policies beyond the contribution of Perry-Smith and Blum (2000) in that its measure is objective. While an argument could be made that share price is a function of constituents’ perceptions of the firm’s value; it is, nonetheless, a bottom-line number to analyze. By only assessing firm value at the time of the announcement, though, Arthur’s (in press) study is a conclusion on reactions to announcements rather than a study on the effectiveness of the policies. If a policy was announced and never implemented, implemented and rarely used, or implemented and highly effective is unknown. Also, the policy’s impact on an objective measure of firm performance is uncertain.

Summary Although other frameworks have been suggested, institutional, resource dependence, and organizational adaptation theories dominate firm-level research on work-family policies. Scholars have found, by and large, consistent results in identifying the antecedents to firm adoption of work-family initiatives. Furthermore, scholars have recently made inroads in the investigation of the relationship between work-family initiatives and firm-level outcomes using macro-level theories of the firm. Beyond organizational theory studies, the efficiency or the human resource-based view of the firm implicitly suggests a relationship between work-family programs and firm performance via increased human resource efficiencies. The next section summarizes what we know about work-family conflict from the psychological literature. Second, a review of the literature on the under-utilization of work-family policies is presented. Following, we develop a model integrating the support literature with the human resource model of work-family policies. In doing so, we present several propositions. Numerous investigations in the psychological literature have revealed a negative relationship between work-family conflict and attachment behaviors.

240

MICHELLE M. ARTHUR AND ALISON COOK

In general, work-family conflict has been related to lower job satisfaction, increased job burnout, lower commitment, lower performance, intentions to turnover and the like. In sum, the effects of work-family conflict have almost uniformly been related to poor work outcomes. These analyses spurred research into the effectiveness of organizational interventions in halting these deleterious outcomes. As such, a human resource model was developed proposing that firm interventions to alleviate work and family conflict would garner positive work outcomes to firms. The human resource or efficiency-based model proposes several advantages to firms of work-family programs. In general, the model suggests that firms with work-family programs will benefit via human resources with increased ability to attract employees, retain employees and produce more efficiently. An implicit assumption of this view is that work-family programs are interpreted, enforced, and used uniformly across firms. However, several qualitative studies have found considerable variation in the application of work-family programs across and within firms (e.g. Lee et al., 2000). Scholars have found considerable variance in the use of work-family programs. Specifically, several scholars have noted the under-utilization of work-family programs (Bailyn, 1997; Hochschild, 1997; Kamerman & Kahn, 1987; Kossek et al., 1999; Kossek & Ozeki, 1999). Kamerman and Kahn (1987), for instance, reported closures of childcare centers at AT&T during the 1970s as a result of lack of use. Additionally, Hochschild (1997) found even when parents were offered progressive programs at work they failed to take advantage of them. Recognizing this phenomenon, several scholars have examined potential barriers facing the utilization of work-family programs (Bailyn, 1997; Hochschild, 1997; Kossek et al., 1999; Perlow, 1995; Powell, 1999; Secret, 2000). Bailyn (1997), incorporating both societal perceptions and organizational culture, offers a broad framework for understanding barriers that may exist. To overcome barriers, she claims society must view work-family issues as a larger concern, one for the organization and its culture, not just the individual. People’s current perceptions regarding time at work, such as their beliefs about the importance of being visible in the organization, persuade some not to use the benefits offered. Bailyn (1997) concludes that to overcome these barriers, perceptions will have to shift. A similar appeal is made by Perlow (1995). Consistent with Bailyn’s (1997) view, she believes current perceptions regarding time and success need to change. Furthermore, she asserts that work-family policies are only “band-aid” solutions that may bring credit to the organization, but may thwart the long-term career potential of those who use the policy. Perlow (1995) believes work-family policies are treating a symptom rather than the problem. To progress, organizations need

A Multi-Theoretical Perspective

241

to stop rewarding the old ways of working. For example, current assumptions hold that one must be visible at work to be successful; one must work extended periods of time to be committed; and one must always place work as the top priority. Perlow (1995) contends these underlying assumptions create a contradiction. Work-family policies are instigated in order to help the employee and the employer; however, by using a work-family program one may violate the assumptions, which, in turn, may hinder overall career advancement. This assertion is supported by both Perlow’s (1995) finding on flexible scheduling and by Judiesch and Lyness’ (1999) finding on leaves of absence. Both studies suggest deleterious consequences for the employee such as being passed over for promotions or salary increases. Therefore, the hesitancy on part of the employee to utilize a work-family program is not surprising. Kossek et al. (1999) proposed a suggestion to promote employee use of work-family policies. In their study, approximately 1,000 managers and their use of alternative work schedules were investigated. Consistent with previous work, many employees did not take advantage of the work-family policy (Hochschild, 1997; Kamerman & Kahn, 1987; Kossek & Ozeki, 1999). To overcome apprehension of use within the workplace, Kossek et al. (1999) suggest organizations need to foster cultural change within the firm. By encouraging managers to take the first step within their work group and use the policy, they believe it will signal to other employees the acceptability of the program. Supporting this assertion, Thompson et al. (1999) and Secret (2000) found employees’ use of work-family initiatives was positively related to perceptions of a supportive culture. Expanding beyond workplace culture, Secret (2000) also investigated organizational characteristics as predictors of employee use of work-family policies. Secret (2000) asserts organizational characteristics such as size and industry are valuable predictors of employee use; moreover, she found that organizational characteristics are better predictors of use than individual differences. For instance she surprisingly found parents were no more likely to use work-family programs than other employees. However, Secret’s (2000) results confirm employees in public and non-profit firms, as well as employees in larger organizations are more likely to use the programs. Clearly, the extent to which the culture or managers support work-family practices is crucial to the efficiency gains in the human resource model and, perhaps less so, for macro-level theories (e.g. institutional theories) of the firm. A model excluding the level of managerial or firm commitment to work-family policies would be remiss. In most cases, the gains in efficiency may not be realized by firms without managerial support for the work-family program. The following discussion focuses on support and what it means in the human resource model. In the discussion, several propositions are introduced.

242

MICHELLE M. ARTHUR AND ALISON COOK

PROPOSITIONS LINKING WORK-FAMILY PROGRAMS AND FIRM PERFORMANCE
The following propositions suggest that the relationship between work-family programs and firm performance relies on a synthesis of the human resource model and the level of managerial support for the program. We begin by discussing the research presented by Lee et al. (2000) that examines the variation in the levels of managerial support for work-family programs across firms. Next, we combine this research with the underlying human resource model constructs of attraction, retention and productivity gains from work-family programs. In doing so, we propose that complete managerial support is necessary to maximize firm performance in response to a work-family program. Support Literature Work-family policies can have various levels of support at the managerial level (e.g. Perlow, 2000). As previously discussed, a typology presented by Lee et al. (2000) classifies firm responsiveness to work-family policies as accommodation, elaboration or transformation. That is, work-family policies could be in an accommodation phase, wherein minimal firm support is found. For example, a firm could announce a policy or have a policy that does not change actual practices in the firm or may accommodate one individual on a piecemeal basis. Second, a firm could be in the elaboration phase, wherein firms have a new routine in response to a workfamily policy. However, the firm does not give up old ways of doing things. Thus, a penalty may exist for violating previous work arrangements. If so, employees may be penalized for use with lack of promotion, public admonition, and/or lack of salary increases (Judiesch & Lyness, 1999; Perlow, 1995); thereby, illustrating an overall lack of support for the work-family policy. Third, a firm could be in a transformation phase, a phase wherein the work-family policy is accepted as a new routine to help employees become more efficient. Depending on the stage of responsiveness to a work-family policy, the relationship between a work-family policy and the constructs of the human resource perspective should vary.

Human Resource Model Attraction The first link in the human resource model or the efficiency-based model of the firm is attraction. That is, prospective employees being more attracted to a firm

A Multi-Theoretical Perspective

243

because of a work-family human resource policy. The underlying assumption is that employees who desire these policies will be more attracted to firms with work-family policies than those without. In addition, prospective employees who do not desire these types of policies will not be repelled from the firm, perhaps, because other factors such as pay, career advancement potential, and the like may be more salient human resource considerations to them. Thus, the overall effect is an increase in the applicant pool. If workers characteristics are normally distributed and selection processes are reliable, firms should have more productive workers from which to select from the applicant pool (Hannon & Milkovich, 1996). Attracting and selecting productive employees should allow a firm to gain a competitive advantage (Hannon & Milkovich, 1996). In spite of this implicit chain of events, there are few empirical studies that examine the relationship between work-family policies and candidate attraction (Carmichael, 1984). Attraction may be a firm-level response to a work-family policy. Therefore, one would expect that, to the extent that a candidate is unfamiliar with the internal application of a work-family policy, the level of support within a firm would not moderate the level of attraction for individuals outside the firm with no knowledge of the level of support. However, attraction also takes place by way of internal recommendation (Kossek & Nichol, 1992). Kossek and Nichol (1992) found that employees were more likely to recommend potential employees to an organization due to an on-site childcare center. We expect current of former employees to strongly recommend a firm for employment that is fully committed to a workfamily policy. Thus, more potential employees will be drawn to firms with full managerial commitment to work-family programs than firms with lower levels of commitment. Hence, Proposition 1a. Firms with work-family programs in a transformational phase will attract more candidates than firms in an elaboration phase. Furthermore, Proposition 1b. Firms with work-family programs in an elaboration phase will attract more candidates than firms in an accommodation phase. Retention The second often cited efficiency-based gain from work-family programs is retaining skilled employees. Several studies have examined the relationship between work-family programs and retention or similar constructs. Studies directly examining retention have found a relationship between work-family programs and retention (Denton et al., 1990; Kossek & Nichol, 1992). Both studies

244

MICHELLE M. ARTHUR AND ALISON COOK

found work-family programs increase employees’ desire to stay with a firm. Similarly, studies have suggested that work-family programs decrease intentions to turnover (Christensen & Staines, 1990; Grover & Crooker, 1995; Youngblood & Chambers-Clark, 1984) and increase employees’ affective commitment and organizational commitment (Goldberg et al., 1989; Grover & Crooker, 1995; Mellor et al., 2001; Pierce & Newstrom, 1982; Youngblood & Chambers-Clark, 1984). Recently, Roehling et al. (2001) found work-family programs increase employee loyalty for all employees, regardless of applicability to their personal needs. Last, studies suggest work-family programs increase job satisfaction (Christensen & Staines, 1990; Thomas & Ganster, 1995). Job satisfaction is, perhaps, the strongest predictor of retention. Tests of retention and its underlying constructs have produced the most consistent results of the relationships between the human resource model and work-family policies. The underlying assumption is that work-family programs provide a benefit because they provide employees with better capabilities to balance work and family, hence higher job satisfaction and retention. That said, the programs must be functional to provide balance between work and family. Furthermore, a relationship between a work-family program and retention relies on managerial commitment to the work-family program. Therefore, Proposition 2a. Firms with work-family programs in a transformational phase will retain more employees than firms in an elaboration phase. Similarly, Proposition 2b. Firms with work-family programs in an elaboration phase will retain more employees than firms in an accommodation phase. Productivity Gains Productivity gains are the final link in the human resource model. Clearly, productivity gains are realized through attraction and retention. However, studies have investigated a few other constructs. The underlying idea is that if employees can balance work and family or minimize work and family conflict, employees may be more productive at work. Employees may become more productive for several reasons. Studies have shown that employees may be mentally healthier (e.g. Frone et al., 1992). Also, studies suggest employees are less likely to be absent (Christensen & Staines, 1990; Dalton & Mesch, 1990; Krausz & Freibach, 1983) and tardy (Christensen & Staines, 1990). Mental health and steady attendance at work should allow employees to be more productive. However, this is only the case if the work-family program applies to them, and the program has thorough managerial commitment. Thus,

A Multi-Theoretical Perspective

245

Proposition 3a. Firms with work-family programs in a transformational phase will extract more productivity gains from employees than firms in an elaboration phase. Likewise, Proposition 3b. Firms with work-family programs in an elaboration phase will extract more productivity gains from employees than firms in an accommodation phase. Based on the human resource model of the firm, work-family programs that are in a transformational phase should increase the level of benefits realized by the firm. It is commonly understood by human resource scholars that if a firm can acquire good employees, retain skilled workers and allow employees to work more efficiently at work, we expect a firm to be able to reduce employee per unit costs (e.g. Hannon & Milkovich, 1996). That is, if a firm can attract, keep, and increase the efficiency of its employees, we expect future recruiting, hiring, and training costs to decrease. Concurrently, if employees are more productive, firms should be able to extract more profits from the employees work. We must also acknowledge that the costs of a program increase when it requires commitment to cultural change and, potentially, a restructuring of work. Clearly, the costs of work-family programs vary with program type. However, studies have suggested that the benefits of work-family programs outweigh the costs (Grover & Crooker, 1985). Changing the structure of work to accommodate work and family is a long-term approach to employees. That said, the benefits at a specific moment in time are difficult to measure. Longitudinal studies are necessary to accurately estimate the costs and benefits of such programs. Several work-family scholars have made a call for more longitudinal research (e.g. Kossek, 1987). Aside from real gains in employee efficiency, work-family programs affect performance at the macro-level as well. Macro-level studies suggest firm value increases via reputational gains or the “intangible wealth” of an organization following adoption of work-family programs (Arthur & Cook, 2003). Two recent studies have shown the adoption of work-family programs has a positive effect on firm performance, measured as stock price and perceived performance (Arthur, in press; Perry-Smith & Blum, 2000). These studies have not investigated the level of managerial commitment to the work-family programs. For symbolic action and institutional theories, the level of managerial commitment is not as important to theoretical predictions as the human resource model. Nonetheless, a positive relationship exists. In sum, macro-level research tends to suggest firm adoption of work-family programs positively affects firm-level outcomes.

246

MICHELLE M. ARTHUR AND ALISON COOK

Based on the discussion above, we expect a positive relationship to exist between adoption as well as transformational work-family programs and firm performance. That is, both micro- and macro-level research suggests a positive relationship between work-family programs and firm performance. Several investigations of the relationship between the constructs of the human resource model and work-family policies have suggested positive work outcomes to firms; others have found no relationships for some constructs. No study, has found that work-family programs have no effect for all of the outcomes predicted. Similarly, no study has found that work-family programs have a negative effect on any of the constructs measured. That said, if research of the human resource model generally supports a positive relationship at the micro-level for transformational firms and the institutional models exhibit a positive effect, the overarching effect of a work-family program should be positive. Proposition 4a. Firm adoption of a transformative work-family program will increase firm performance more than firm adoption of an elaborative workfamily program. Likewise, Proposition 4b. Firm adoption of an elaborative work-family program will increase firm performance more than firm adoption of an accommodative workfamily program.

FUTURE RESEARCH
Firm-level studies must move beyond measuring firm performance in response to the adoption of work-family policies. We propose that firms with managerial commitment to work-family programs will reap more benefits that those with less commitment. Measures of program use and firm performance seem an obvious direction for future work-family research. We acknowledge that this is somewhat of a Catch-22 for researchers. That is, firm adoption often requires that managers believe that there are bottom-line outcomes of work-family programs. However, managerial commitment is an antecedent to showing a relationship between workfamily programs and firm performance. Nonetheless, causality may be investigated with longitudinal studies. Work-family research is dominated by cross-sectional studies. Thus, accurate costs and benefits are difficult to measure. More longitudinal studies would provide insight into the actual benefits and costs of work-family programs. We know that firms that adopt work-family policies, specifically at the transformational

A Multi-Theoretical Perspective

247

level, are taking a long-term view of employees. That said, an accurate estimation of the costs and benefits must also take place over time. Aside from firm-level performance, more research of the under-investigated human resource model constructs must be tested. More specifically attraction and direct productivity have been largely overlooked in the work-family research. We know people are more likely to recommend a firm with work-family programs (Kossek & Nichol, 1992). However, actual applicant pool fluctuation would provide further support for the human resource model. Direct productivity, perhaps, is the most under-investigated construct. Several studies have examined variables that suggest productivity increases. However, a longitudinal study of employee productivity before and after a work-family intervention would be discerning.

CONCLUSION
It is essential to recognize and accept the demographic transition in the labor force. Several scholars assert organizations must take an active role in assisting their employees in the balance of work and family domains (Bailyn, 1993; Glass, 2000; Jacobs & Gerson, 2001; Kossek et al., 1999). Glass (2000) and Bailyn (1993) both recognize the importance of abandoning the traditional assumptions of work and time. Bailyn (1993), acknowledging the importance of productivity, asserts personal needs still need to be addressed. Haas (1999), however, contends organizations’ responsiveness to work-family issues will only stem from an economic concern. As illustrated throughout the review, researchers have consistently demonstrated positive work outcomes as a result of firms addressing work-family issues. It is time for researchers to connect the use of work-family programs to firm performance.

ACKNOWLEDGMENTS
The authors deeply appreciate the comments and suggestions of Joe Martocchio on previous versions of this manuscript.

REFERENCES
Adams, G. A., King, L. A., & King, D. W. (1996). Relationships of job and family involvement, family social support, and work-family conflict with job and life satisfaction. Journal of Applied Psychology, 81, 411–420. Argyris, C., & Schon, D. A. (1978). Organizational Learning: A Theory of Action Perspective. Reading, MA: Addison-Wesley.

248

MICHELLE M. ARTHUR AND ALISON COOK

Arthur, M. M. (in press). Work-family initiatives and share price reaction: An institutional perspective. Academy of Management Journal. Arthur, M. M., & Cook, A. (2003). Taking stock of work-family initiatives. Unpublished manuscript. Bacharach, S. B., Bamberger, P., & Conley, S. (1991). Work-home conflict among nurses and engineers: Mediating the impact of role stress on burnout and satisfaction at work. Journal of Organizational Behavior, 12, 39–53. Bailyn, L. (1993). Breaking the mold: Women, men, and time in the new corporate world. New York: Free Press. Bailyn, L. (1997). The impact of corporate culture on work-family integration. In: S. Parasuraman & J. H. Greenhaus (Eds), Integrating Work and Family: Challenges and Choices for a Changing World (pp. 209–219). Barling, J., & MacEwen, K. E. (1991). Maternal employment experiences, attention problems and behavioral performance: A mediational model. Journal of Organizational Behavior, 12, 495–505. Barnett, R. C., & Baruch, G. R. (1985). Women’s involvement in multiple roles and psychological distress. Journal of Personality and Social Psychology, 49, 135–145. Bronfenbrenner, U., & Crouter, A. (1983). Work and family through time and space. In: S. B. Kamerman & C. D. Hayes (Eds), Families that work: Children in a changing world. Washington, DC: National Academy Press. Burke, R. J. (1988). Some antecedents and consequences of work-family conflict. Journal of Social Behavior and Personality, 3, 287–302. Carmichael, H. L. (1984). Reputations in the labor market. The American Economic Review, 74, 713–724. Christensen, K. E., & Staines, G. L. (1990). Flextime: A viable solution to work/family conflict? Journal of Family Issues, 11, 455–476. Cooke, R. A., & Rousseau, D. M. (1984). Stress and strain from family roles and work-role expectations. Journal of Applied Psychology, 69, 252–260. Daft, R. L., & Weick, K. E. (1984). Toward a model of organizations as interpretation systems. Academy of Management Review, 9, 284–295. Dalton, D. R., & Mesch, D. J. (1990). The impact of flexible scheduling on employee attendance and turnover. Administrative Science Quarterly, 35, 370–387. Denton, K., Love, L. T., & Slate, R. (1990). Elder care in the 90s: Employee responsibility, employer challenge. Journal of Contemporary Human Services, 349–359. DiMaggio, P., & Powell, W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 23, 111–136. Doby, V. J., & Caplan, R. D. (1995). Organizational stress as threat to reputation: Effects on anxiety at work and at home. Academy of Management Journal, 38, 1105–1123. Drago, R., Costanza, D., Caplan, R., Brubaker, T., Cloud, D., Harris, N., Kashian, R., & Riggs, T. L. (2001). The willingness-to-pay for work/family policies: A study of teachers. Industrial and Labor Relations Review, 55, 22–41. Dunham, R. B., Pierce, J. L., & Castaneda, M. B. (1987). Alternative work schedules: Two field quasi-experiments. Personnel Psychology, 40, 215–242. Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work and family: Clarifying the relationship between work and family constructs. Academy of Management Review, 25, 178–199. Edwards, M. E. (2001). Uncertainty and the rise of the work-family dilemma. Journal of Marriage and Family, 63, 183–196.

A Multi-Theoretical Perspective

249

Ferber, M. A., & O’Farrell, B. (1991). Work and family: Policies for a changing work force. Washington, DC: National Academy Press. Frone, M. R., Russell, M., & Cooper, M. L. (1992). Antecedents and outcomes of work-family conflict: Testing a model of the work-family interface. Journal of Applied Psychology, 77, 65–78. Glass, J. (2000). Envisioning the integration of family and work: Toward a kinder, gentler workplace. Contemporary Sociology, 29, 129–143. Glass, J., & Estes, S. B. (1997). The family responsive workplace. Annual Review of Sociology, 23, 289–313. Goff, S. J., Mount, M. K., & Jamison, R. L. (1990). Employer supported child care, work/family conflict, and absenteeism: A field study. Personnel Psychology, 43, 793–809. Goldberg, W. A., Greenberger, E., Koch-Jones, J., O’Neil, R., & Hamill, S. (1989). Attractiveness of child care and related employer-supported benefits and policies to married and single employees. Child and Youth Care Quarterly, 18, 23–37. Gonyea, J. G., & Googins, B. K. (1993). Linking the worlds of work and family: Beyond the productivity trap. Human Resource Management, 31, 209–226. Goodstein, J. D. (1994). Institutional pressures and strategic responsiveness: Employer involvement in work-family issues. Academy of Management Journal, 37, 350–382. Goodstein, J. (1995). Employer involvement in eldercare: An organizational adaptation perspective. Academy of Management Journal, 38, 1657–1671. Gouldner, A. W. (1960). Cosmopolitans and locals: Toward an analysis of latent social roles. Administrative Science Quarterly, 5, 444–479. Grandey, A. A., & Cropanzano, R. (1999). The conservation of resources model applied to work-family conflict and strain. Journal of Vocational Behavior, 54, 350–370. Greenhaus, J. H., & Beutell, N. J. (1985). Sources of conflict between work and family roles. Academy of Management Review, 10, 76–88. Greenhaus, J. H., Parasuraman, S., & Collins, K. M. (2001). Career involvement and family involvement as moderators of relationships between work-family conflict and withdrawal from a profession. Journal of Occupational Health Psychology, 6, 91–100. Grover, S. L., & Crooker, K. J. (1995). Who appreciates family-responsive human resource policies: The impact of family-friendly policies on the organizational attachment of parents and non-parents. Personnel Psychology, 48, 271–288. Gutek, B. A., Searle, S., & Klepa, L. (1991). Rational versus gender role explanations for work-family conflict. Journal of Applied Psychology, 76, 560–568. Guthrie, D., & Roth, L. M. (1999). The state, courts, and maternity policies in U.S. organizations: Specifying institutional mechanisms. American Sociological Review, 64, 41–63. Haas, L. (1999). Families and work. In: M. B. Sussman & S. K. Steinmetz (Eds), Handbook of Marriage and the Family. New York: Plenum. Hannon, J. M., & Milkovich, G. T. (1996). The effect of human resource reputation signals on share prices: An event study. Human Resource Management, 35, 405–424. Hobfoll, S. E. (1989). Conservation of resources: A new attempt at conceptualizing stress. American Psychologist, 44, 513–524. Hochschild, A. (1997). The Time Bind. New York: Metropolitan. Ingram, P., & Simons, T. (1995). Institutional and resource dependence determinants of responsiveness to work-family issues. Academy of Management Journal, 38, 1466–1482. Jacobs, J. A., & Gerson, K. (2001). Overworked individuals or overworked families? Explaining trends in work, leisure, and family time. Work and Occupations, 28, 40–63.

250

MICHELLE M. ARTHUR AND ALISON COOK

Judiesch, M. K., & Lyness, K. S. (1999). Left behind? The impact of leaves of absence on managers’ career success. Academy of Management Journal, 42, 641–651. Kamerman, S. B., & Kahn. A. J. (1987). The responsive workplace: Employers and a changing labor force. New York: Columbia University Press. Kanter, R. M. (1977). Work and Family in the United States: A critical review and agenda for research and policy. New York: Russell Sage. Katz, D., & Kahn, R. L. (1978). The social psychology of organizations (2nd ed.). New York: Wiley. Kelly, E., & Dobbin, F. (1999). Civil rights law at work: Sex discrimination and the rise of maternity leave policies. American Journal of Sociology, 105, 455–492. Kossek, E. E. (1987). Human resources management innovation. Human Resource Management, 26, 71–92. Kossek, E. E., Barber, A. E., & Winters, D. (1999). Using flexible schedules in the managerial world: The power of peers. Human Resource Management, 38, 33–46. Kossek, E. E., Colquitt, J. A., & Noe, R. A. (2001). Caregiving decisions, well-being, and performance: The effects of place and provider as a function of dependent type and work-family climates. Academy of Management Journal, 44, 29–44. Kossek, E. E., Dass, P., & DeMarr, B. (1994). The dominant logic of employer-sponsored work and family initiatives: Human resource managers’ institutional role. Human Relations, 47, 1121– 1149. Kossek, E. E., & Nichol, V. (1992). The effects of on-site child chare on employee attitudes and performance. Personnel Psychology, 45, 485–509. Kossek, E. E., & Ozeki, C. (1998). Work-family conflict, policies, and the job-life satisfaction relationship: A review and directions for organizational behavior-human resources research. Journal of Applied Psychology, 83, 139–149. Kossek, E. E., & Ozeki, C. (1999). Bridging the work-family policy and productivity gap: A literature review. Community, Work and Family, 2, 7–32. Krausz, M., & Freibach, N. (1983). Effects of flexible working time for employed women upon satisfaction, strains, and absenteeism. Journal of Occupational Psychology, 56, 155–159. Lambert, S. J. (2000). Added benefits: The link between work-life benefits and organizational citizenship behavior. Academy of Management Journal, 43, 801–815. Lee, M. D., MacDermid, S. M., & Buck, M. L. (2000). Organizational paradigms of reduced-load work: Accommodation, elaboration, and transformation. Academy of Management Journal, 43, 1211–1226. Lobel, S. A. (1991). Allocation of investment in work and family roles: Alternative theories and implications for research. Academy of Management Review, 16, 507–521. Mellor, S., Mathieu, J. E., Barnes-Farrell, J. L., & Rogelberg, S. G. (2001). Employees’ nonwork obligations and organizational commitments: A new way to look at the relationships. Human Resource Management, 40, 171–184. Mellor, S., Mathieu, J. E., Barnes-Farrell, J. L., & Rogelberg, S. G. (2001). Employees’ nonwork obligations and organizational commitments: A new way to look at the relationships. Human Resource Management, 40, 171–184. Meneghan, E. G., & Parcel, T. L. (1990). Parental employment and family life: Research in the 1980s. Journal of Marriage and the Family, 52, 1079–1098. Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace: Theory, research, and application. Thousand Oaks, CA: Sage. Milliken, F. J., Dutton, J. E., & Beyer, J. M. (1990). Understanding organizational adaptation to change: The case of work-family issues. Human Resource Planning, 13, 91–106.

A Multi-Theoretical Perspective

251

Milliken, F. J., Martins, L. L., & Morgan, H. (1998). Explaining organizational responsiveness to work-family issues: The role of human resource executives as issue interpreters. Academy of Management Journal, 41, 580–592. Morgan, H., & Milliken, F. J. (1992). Keys to action: Understanding differences in organizations’ responsiveness to work-and-family issues. Human Resource Management, 31, 227– 248. Nielson, T. R., Carlson, D. S., & Lankau, M. J. (2001). The supportive mentor as a means of reducing work-family conflict. Journal of Vocational Behavior, 59, 364–381. Oliver, C. (1991). Strategic responses to institutional processes. Academy of Management Review, 16, 145–179. Osterman, P. (1995). Work/family programs and the employment relationship. Administrative Science Quarterly, 40, 681–700. Parcel, T. L., & Menaghan, E. G. (1994). Work, Family, and Young Children’s Lives, 157–178. Perlow, L. A. (1995). Putting the work back into work/family. Group & Organization Management, 20, 227–239. Perry-Jenkins, M., Repetti, R. L., & Crouter, A. (2000). Work and family in the 1990s. Journal of Marriage and the Family, 62, 981–998. Perry-Smith, J. E., & Blum, T. C. (2000). Work-family human resource bundles and perceived organizational performance. Academy of Management Journal, 43, 1107–1117. Pfeffer, J. (1981). Management as symbolic action: The creation and maintenance of organizational paradigms. In: L. L. Cummings & B. M. Staw (Eds), Research in Organizational Behavior (Vol. 3, pp. 1–52). Greenwich, CT: JAI Press. Pfeffer, J., & Salancik, G. R. (1978) The External Control of Organizations. New York: Harper and Row. Pierce, J. L., & Newstrom, J. W. (1982). Employee responses to flexible work schedules: An interorganization, inter-system comparison. Journal of Management, 8, 9–25. Prahalad, C. K., & Bettis, R. A. (1986). The dominant logic: A new linkage between diversity and performance. Strategic Management Journal, 7, 485–501. Roehling, P. V., Roehling, M. V., & Moen, P. (2001). The relationship between work-life policies and practices and employee loyalty: A life course perspective. Journal of Family and Economic Issues, 22, 141–170. Rothausen, T. J., Gonzalez, J. A., Clarke, N. E., & O’Dell, L. L. (1998). Family-friendly backlash – fact or fiction? The case of organizations’ on-site child care centers. Personnel Psychology, 51, 685–706. Rothbard, N. P. (2001). Enriching or depleting? The dynamics of engagement in work and family roles. Administrative Science Quarterly, 46, 655–684. Secret, M. (2000). Identifying the family, job, and workplace characteristics of employees who use work-family benefits. Family Relations, 49, 217–225. Shinn, M., Wong, N. W., Simko, P. A., & Ortiz-Torres, B. (1989). Promoting the well-being of working parents: Coping, social support, and flexible job schedules. American Journal of Community Psychology, 17, 31–55. Sumer, H. C., & Knight, P. A. (2001). How do people with different attachment styles balance work and family? A personality perspective on work-family linkage. Journal of Applied Psychology, 86, 653–663. Tajfel, H., & Turner, J. C. (1985). The social identity theory of intergroup behavior. In: S. Worchel & W. G. Austin (Eds), Psychology of Intergroup Relations (2nd ed., pp. 7–24). Chicago: Nelson-Hall.

252

MICHELLE M. ARTHUR AND ALISON COOK

Tausig, M., & Fenwick, R. (2001). Unbinding time: Alternate work schedules and work-life balance. Journal of Family and Economic Issues, 22, 101–119. Thomas, L. T., & Ganster, D. C. (1995). Impact of family-supportive work variables on work-family conflict and strain: A control perspective. Journal of Applied Psychology, 80, 6–15. Thompson, C. A., Beauvais, L. L., & Lyness, K. S. (1999). When work-family benefits are not enough: The influence of work-family culture on benefit utilization, organizational attachment, and work-family conflict. Journal of Vocational Behavior, 54, 392–415. Williams, K. J., & Alliger, G. M. (1994). Role stressors, mood spillover, and perceptions of work-family conflict in employed parents. Academy of Management Journal, 37, 837–868. Yang, N., Chen, C. C., Choi, J., & Zou, Y. (2000). Sources of work-family conflict: A Sino-U.S. comparison of the effects of work and family demands. Academy of Management Journal, 43, 113–123. Youngblood, S. A., & Chambers-Clark, K. (1984). Child care assistance can improve employee attitudes and behavior. Personnel Administrator, 45–46, 93–95.

TOWARD UNDERSTANDING AND MANAGING STEREOTYPICAL BELIEFS ABOUT OLDER WORKERS’ ABILITY AND DESIRE FOR LEARNING AND DEVELOPMENT
Todd J. Maurer, Kimberly A. Wrenn and Elizabeth M. Weiss
ABSTRACT
A model of stereotypical beliefs that older workers have difficulty learning and developing and are not motivated to learn is presented. Three categories of antecedents of the stereotypical beliefs are addressed: (1) experience with stereotype-consistent behaviors and promulgation of the stereotype by others; (2) perceived learning and development inhibitors internal to the older worker; and (3) perceived learning and development inhibitors external to the older worker. Potential consequences of the stereotypical beliefs for older workers and employing organizations are also explored. Individuating information and knowledge of within-older-group differences are posited to attenuate the influence of group-based stereotypes. Processes and tactics within organizations that should increase this information and knowledge are presented. The proposed model provides a framework to help

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 253–285 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22006-5

253

254

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

guide future research on this topic and also some suggestions for managing a work place where these beliefs may exist.

INTRODUCTION
Learning and development has become increasingly important to career success and longevity. The workplace is changing rapidly and work itself is being redefined, increasingly requiring constant learning (Kiechel, 1993). Consistent with these changes, Hall and Mirvis (1995) have described a shift from the organizational career to a “protean” career. In the former, one spends an entire career with the same organization and type of work. In the latter, the focus is heavily oriented toward learning and adapting to new organizations and careers. According to Hall and Mirvis, “learning how to learn and continuous learning have become core career competencies” (pp. 280–281). Similarly, Greller and Stroh (1995) note that changes in technology and business strategies mean that “pursuing a career requires considerable commitment to learning and reshaping one’s skills. Workers who do not take the steps necessary to stay skilled and knowledgeable may be forced to cut short their careers as their skills become obsolete or irrelevant” (p. 236). In the business community, learning and development are increasingly more important. One survey asked 400 HR executives to judge the importance of 29 employee qualities (AARP, 2000). Scoring in the top 15 were “Flexible in doing different tasks,” “Will participate in training programs,” “Try new approaches,” “Up-todate skills,” and “Learn new technology.” All of these things relate to training and continuous learning of new job skills. In the same study, “Training your current workforce,” “Introducing new information technology into the workplace,” and “Retraining your workforce” were ranked in the top ten general employment issues that the HR executives considered important for today’s workforce. Further, given the nature of changing job content, cognitive (as opposed to physical) skills seem increasingly important. While these changes are taking place, the demographics of the workforce are changing dramatically. One of the most important changes is what has been referred to as the “graying of the work force,” which is the increasing average age of workers. Likewise, Capowski (1994) notes that the median age of the work force is projected to reach 40 by the year 2010. The number of people age 50–65 will increase at more than twice the rate of the overall population by the end of the century. Fullerton (1999) reports that the highest population growth over the 1998–2008 period is expected to occur in the 45–64 age group. From this discussion we can conclude that training and development are becoming increasingly important in today’s changing work environment, and the work

Toward Understanding and Managing Stereotypical Beliefs

255

force affected by these changes is composed of an increasing number of older workers. This clearly indicates the importance of a focus on the development and training of older workers. However, some research and work place literature, to be discussed below, has suggested that stereotypical beliefs exist in which older workers are thought to lack both ability and inclination to learn and develop. In light of the discussion above, these beliefs pose a serious potential problem in the work place.

A Need for Better Understanding of Stereotypical Beliefs in the Work Place Stereotypical beliefs may influence both decision-makers and older workers themselves. To the extent that decision-makers and other key people are influenced by these beliefs, this may result in a relative lack of opportunity or support for older workers to participate in training and development activities. They may be excluded or discouraged from participation, may not be selected for particular assignments, or may not be encouraged or supported in more subtle, but psychologically meaningful ways. This may be especially true to the extent that decision makers are trying to maximize their returns on investments in training or learning resources. To the extent that older workers themselves are exposed to or hold stereotype-consistent beliefs, this may influence their motivational state of mind and behavior (Maurer, 2001; Salthouse & Maurer, 1996). Maurer and Rafuse (2001) outlined the legal and behavioral implications of differences in both opportunities and support for development as a function of age. Also, while most people have been sensitized to offensive sexist or racist stereotypes at work, the idea that “you can’t teach an old dog new tricks” seems not to have acquired the same taboo status, even though workers over the age of 40 are legally protected from discrimination, just as minorities and women are. However, beyond the recognition that this stereotypical belief exists in the work place and that it may influence behavior, there has been relatively little analysis of the nature of that belief, and more importantly, its specific origins and basis. An analysis of the origins and basis for these beliefs not only enhances understanding of them but also provides clues regarding how to deal with them effectively in the workplace. In the present paper, a model is presented of beliefs about older workers’ capacity for and inclination toward learning and development in organizational settings (see Fig. 1). This model is intended to be relevant to settings in which older and younger workers alike are in need of continuous learning and development, as described earlier in this paper. Situations in which a need for development or learning are precluded are not the concern here (such as when workers already

256 TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Fig. 1. Model of Stereotypical Beliefs About Older Workers’ Ability and Desire for Learning and Development.

Toward Understanding and Managing Stereotypical Beliefs

257

possess needed knowledge or skill, or when they have announced an immediately impending retirement or departure from the organization). The model draws on literature in aging, organizational behavior, management, and industrial-organizational psychology to help explain the origins and basis for beliefs about older workers’ capability and inclination to learn and develop. Within the framework, a system of antecedents and potential outcomes of these beliefs is presented. The role that individuating information and knowledge of within-age-group variability can play in reducing effects of stereotypes is highlighted and ideas are offered regarding interventions that may provide individuating information within organizations. An important and unique aspect of this model is that it shows not only the potential effects the stereotypes have on decision-makers in the workplace (supervisors, peers, subordinates), but also the potential effects of stereotypes on the behavior of older workers themselves. The overall model can help guide future research on this topic. Also, the framework is intended to be one which is not only based on research literature and theory, but also one which is practical enough to provide sound and reasonable suggestions for practice in dealing with stereotypical beliefs about older workers in organizations. In the next section of this paper, a description of the model will begin with a discussion of the belief that older workers are less capable of learning and developing and less inclined to participate in such activities. Some potential outcomes of these beliefs will be discussed. Following this, three general categories of factors which are proposed to contribute to this belief will be described, including: (a) experience with stereotype-consistent behaviors and promulgation of the stereotype by others; (b) perceived learning and development inhibitors internal to the older worker; and (c) perceived learning and development inhibitors external to the older worker. In the final section of the paper, data on variability in actual age differences will be described and the way in which knowledge of this information may reduce the effects of group-based stereotypes on decisions and behavior will be illustrated. Based on this, ideas for providing individuating information and knowledge of within group variability are offered, along with suggestions for managing a work place where these beliefs may exist.

Definition of “Older Worker” First, however, some discussion is warranted on the subject of what defines an “older worker.” This issue is relevant in age research because there is no universal operationalization of the term, and it may depend on the specific context at hand.

258

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Legally, an older worker is defined as anyone in the workforce at or above the age of 40 (Age Discrimination in Employment Act). However, in the research literature, the definitions are much more varied, from 35 and over (Downs, 1967; Newsham, 1969), to 36 through 60 (Caplan & Schooler, 1990) to 55 through 67 (Elias, Elias, Robbins & Gage, 1987) to people aged 58–84 (Zandri & Charness, 1989). Adding to the confusion is the fact that many of these studies seem to lack any rationale for choosing a particular age at which to define a worker as “older” (e.g. Downs, 1967; Elias, Elias, Robbins & Gage, 1987; Zandri & Charness, 1989) while others used a median split in order to separate a group of participants into younger and older adults (Caplan & Schooler, 1990). In a review of 105 studies that defined the term “older worker,” Ashbaugh and Fay (1987) found the average chronological age as operationalized in research of an older worker to be 53.4 years. Thus, perhaps this can serve as a reasonable lower bound for the term “older worker,” although what is considered old may depend on the context.

A MODEL OF STEREOTYPICAL BELIEFS ABOUT OLDER WORKERS’ ABILITY AND INCLINATION TO LEARN/DEVELOP AND POSSIBLE CONSEQUENCES OF THESE BELIEFS
We present the rationale for the specific links of the model in the following sections.

Ability to Develop Research has suggested that there is a common belief that older workers have less potential for learning and development. In a study by Rosen and Jerdee (1976a), 65 items were developed and classified into four work-related scales: performance capacity, potential for development, stability, and interpersonal skills. Respondents were asked to respond to each item twice, once as it applies to the average 60-year-old man and once as it applies to the average 30-year-old man. In the category of “potential for development,” the mean rating was significantly higher for the 30-year old man than for the 60-year-old man. In a study by Perry and Varney (1978), targets varied on age and level of competence. The older worker was rated lower than the younger worker on two dimensions: catching on to new ideas and contributions made in future. In research performed by the Institute of Personnel Management (1993), a large-scale survey

Toward Understanding and Managing Stereotypical Beliefs

259

was conducted in five European countries. Older workers were perceived to “Learn less quickly” and they “Are less able to grasp new ideas.” These items were part of factor termed Adaptability. Older workers were rated significantly below the scale midpoint on this factor. A recent article by Forte and Hansvick (1999) partially supported the idea that people believe older workers are less capable of development. They asked 98 employers about their attitudes toward older workers on 12 dimensions related to work. Overall, older workers were rated lower on the attributes of computer skills, ability to learn quickly, energy and stamina, and flexibility. They were rated higher than younger workers on academic skill levels, attendance, ability to get along with co-workers, work ethics, salary expectations, and supervisory skills. The dimensions on which older workers were rated lower are those closely related to learning and development. Interestingly, this study did show evidence of an ingroup bias among younger employers toward older workers. Employers who were 50 or over themselves rated workers of similar age as being more desirable on all attributes except “possesses stamina and energy to perform job.” Thus, older employers viewed older workers as more able in development-related areas than did younger employers. Warr, in his (1994a) review article, notes that several other articles (e.g. Doering, Rhodes & Schuster, 1983; Heron & Chown, 1961; Stagner, 1985) have “documented the common tendency to view older workers as slower, less interested in new training, less flexible, and more likely to become weary than their younger colleagues (p. 488).” Summarizing many of these kinds of studies, Finkelstein, Burke and Raju (1995) conducted a meta-analysis which showed that older workers were rated lower than younger workers on having potential for development. The studies included in this review addressed perceived differences (based on group membership) between older workers and younger workers as opposed to actual differences. It might be noted that older adults were not included as subjects in this research. The results of the Rosen and Jerdee (1976b) and the Perry and Varney (1978) studies were based entirely on the responses of undergraduate students. Rosen and Jerdee’s (1976a) study and the Forte and Hansvick (1999) research did include some adult participants from the work force; however, undergraduates still comprised half of the Rosen and Jerdee (1976a) sample. This raises the question of whether age would moderate the difference in development potential ratings observed for older and younger workers if greater numbers of older respondents participated. But overall it seems that, based on the research literature, older adults are perceived as less capable of learning and development. However, it also seems that people may have the perception that older workers are not only less capable, but are also less motivated or inclined to participate in learning and development activities.

260

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Desire to Develop A common stereotype of older people is that they have no interest in adapting to the times. In fact, Capowski (1994) describes research which showed that 57% of businesses claim that older employees show resistance to training. The AARP (1995) has suggested two obstacles to the hiring and training of older workers: (1) perceptions of a lack of flexibility or perceived unwillingness to do more and new tasks, and unwillingness to adapt to new work environment (e.g. older workers = old ways); and (2) perceptions of resistance to technology – unwilling to adapt to new technology in the work place. In a large study conducted in Europe by the Institute of Personnel Management (1993), results reflected a perception that older workers less readily accept the introduction of new technology, and are less interested in being trained. This research suggests that some people perceive older workers as being less interested in or motivated toward engaging in learning and development activities. Additionally, studies finding stereotypes that characterize older adults as being unable to develop often report that older adults are also seen as not wanting to develop. Warr (1994a) cites previous reviews as characterizing older adults as being less interested in new training opportunities. The Rosen and Jerdee (1976b, 1977) studies also found that older workers are seen as being less interested in training. It should be noted that these studies investigating beliefs about older workers’ ability and inclination to learn and develop are to be distinguished from studies which examine actual differences between younger workers and older workers. The above studies refer to judgments of older workers as a group and not appraisals of individual workers and their ability or willingness to successfully participate in development activities.

Potential Consequences of the Beliefs: Behavior by Administrators, Peers and Observers There are several potential consequences of these types of beliefs to people in organizational settings. One significant outcome, given the importance of learning and development opportunities to careers, is a possible lack of access to training and development opportunities. If decision-makers possess these beliefs, older workers may not be nominated, selected, or informed of opportunities to be involved in training or development activities such as training classes, conferences, seminars, on-the-job training, special temporary job assignments, learning-rich job rotations, committee or task force assignments, coaching from supervisors (or mentors, peers), opportunities for feedback, or career planning.

Toward Understanding and Managing Stereotypical Beliefs

261

Rosen and Jerdee (1976b) provided evidence for the link between stereotypes against older workers and discrimination against older workers in terms of recommended action. In their study of simulated managerial decisions, participants reacted to a scenario involving either an older or a younger target. Six of these scenarios represented different stereotypical beliefs about older workers. Results showed that respondents viewed older targets as being more difficult to persuade to change, significantly less motivated to keep up to date, and less desirable for retraining opportunities (preferring termination and replacement to retraining). In a replication of this study using Harvard Business Review subscribers ranging in age from younger than 30 to over 60, similar results were found (Rosen & Jerdee, 1977). In addition to the research described above, literature on actual workforce behavior suggests that older workers may not participate in and/or may not have access to some types of development activities. We cannot be sure across all of these settings whether they are not participating because they don’t want to, don’t need to, or because their employers do not give them the opportunity. However, some of this literature suggests that older workers might be sometimes excluded or may be influenced to choose not to participate, and either of these explanations could be influenced by stereotypical beliefs held by decision-makers or older workers themselves. For example, Simon (1996) asserts that many older workers are being denied the opportunity to improve their skills, citing that 55–64-year-olds are only a third as likely as 35–44-year-olds to receive training in the workplace, according to a study by the U.S. Department of Labor. Simon suggests that many employers may discourage older workers from obtaining training. Boerlijst (1994) found that participation in training and development courses relevant to one’s functional area decreases with age. Also, 65% of workers over 40 had never or hardly ever participated in training and development courses outside of their functional area. In the 1995, Study of Employer-Provided Training (U.S. Bureau of Labor Statistics) researchers reported that 50.7% of employees who were aged 55 and over received formal training from their employers in the 12 months prior to the survey. In contrast, 78.5% of their counterparts in the 25–34 age group received formal training from their employers. Capowski (1994) reports that a study by the AARP found that only three out of ten companies include older adults in their training programs. A study conducted by SHRM and AARP (1995) found that 47% of respondents’ organizations provide training to upgrade skills of older workers. Greller and Stroh (1995) assert that “the message conveyed to older workers . . . may be that they no longer have any potential” (p. 235). Hall and Mirvis (1995) suggest several barriers to career development in the form of stereotypes about older workers. Among them is the perception that it is

262

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

too costly to continue to invest in developing older employees, the belief that the older worker is too inflexible and difficult to train, and the perception that retraining for the older worker would represent too much effort for a relatively small group of employees. And in simulated hiring decisions, age can have a significant effect on a decision to hire (younger workers being more likely to be hired) (Haefner, 1977). There is little experimental research that documents age discrimination in actual work place decisions or simulated decisions. Simple differences in participation rates do not reflect direct evidence of discrimination. Several types of decisions and judgments should be addressed in future research. First, the most overt types of decisions are those in which a learning-rich job or a training opportunity is being formally sought and an acceptance (or funding) or rejection decision must be made. This is the type of decision made in the Rosen and Jerdee studies described previously. Are older workers’ applications for learning-rich jobs or requests for training opportunities accepted to the same degree as younger workers? Likewise, when a temporary job assignment or transfer from an existing job must be made that will involve a high degree of learning, is an older worker nominated or chosen for this assignment? Job assignments are a significant source of development. A related type of judgment is an evaluation of an individual’s potential for development, and/or his/her learning and development capabilities as part of a performance appraisal process or perhaps ratings of promotability or even as part of a reduction in force. Another measure or action might be the amount of attention and effort devoted to the developmental part of an annual performance evaluation. Does a judge and feedback-provider invest more effort into development and future-oriented learning for a younger as opposed to an older feedback recipient? Another set of variables has to do with more subtle social effects, including support or dissuasion from developmental experiences. For example, does a supervisor or co-worker provide encouragement and prodding toward a challenging, developmental situation or task or is there a subtle “silence” on the part of the would-be supporter when it comes to considering a challenging task or learning experience on the part of the older worker? Likewise, another type of action is informing an older worker of developmental opportunities. If an opportunity came along to attend a workshop or participate in a challenging skill development task, does the person pass the information along to an older worker in case he/she might be interested? Likewise, does a co-worker share recently learned information and skills with an older person (for example, a new computer software technique)? Are older workers kept “in the loop” when it comes to learning and development? Based on the literature reviewed above along with the research suggestions just described, several research propositions are given below.

Toward Understanding and Managing Stereotypical Beliefs

263

Proposition 1. People in organizational settings who believe that older workers are typically neither able nor interested in learning and development will be less likely to: (a) make decisions that are favorable toward older workers (selection into learning-rich jobs, acceptance/funding of training requests, choose for a temporary and learning-rich job assignment or task, etc.); (b) make judgments that are favorable toward older workers (potential for development or learning ability judgments made in performance appraisal, promotability ratings, or human capital judgments made in reductions in force, etc.); (c) provide support for learning and development (encouragement and prodding toward challenging learning, informing about opportunities, sharing of knowledge and skills, etc.). While some research has shown age effects in decisions and judgments and other research has documented beliefs that older workers are not able nor inclined to learn and develop, research has generally not investigated actual empirical linkages between the beliefs and decisions, judgments and support constructs described above. Although relationships have been inferred, causal (or even empirical) relationships have not been investigated directly. Future research should include both constructs (beliefs and outcomes) in investigations to determine if they are related. Further, most research has not addressed social support constructs as outcomes (e.g. encouragement, information sharing, etc.) despite the fact that these variables may have an influence on older workers’ behavior and work life. Additionally, the Rosen and Jerdee studies that are highly cited in this area were conducted over 25 years ago. A replication of Rosen and Jerdee’s (1976b) study conducted by Weiss and Maurer (in press), using primarily engineering and computer science majors, failed to replicate the original findings, suggesting that this research may be in need of updating. It is unclear whether effects in prior literature are cohort-based, or if they will continue to be observed as time goes on. More current research on the effects of stereotypes is needed that addresses decisions, judgments and social support involving older workers in learning and development settings. Perhaps this research can examine different characteristics of decision-makers and of their contexts for relationships with age-related stereotypes, and tendency to make age-based decisions and judgments.

Potential Consequences of Beliefs: Behavior by Older Workers Themselves One unique aspect of this model is that the possibility of stereotypes influencing behavior by older workers themselves is recognized. Based on the literature,

264

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

it seems that stereotypes might indirectly influence behavior of older workers themselves through a self-fulfilling prophecy. They may influence older workers’ perceptions of what is appropriate or possible for individuals in their age group. Greller and Stroh (1995) cite Areis – “Like people at other stages in their lives, older people come to understand what is expected of them by looking for cues and role definitions provided by others (Aries, 1962; Cole, 1992)” (p. 235). Greller and Stroh suggest that stereotypes may “influence workers’ concepts of appropriate aging behavior, which may lead older individuals to conform more closely to others’ expectations. They may resist taking on challenging work assignments, for example, or ‘decide to retire’ even though they would rather continue to work.” (p. 239). Capowski (1994) also cites Denise Loftus (of the AARP) – “In many cases, older people were passed over for training because it was assumed that they couldn’t learn – this gets reinforced, and it just becomes a self-fulfilling prophesy. It’s sad, because the ability to learn continues” (pp. 12–13). Capowski (1994) asserts that older employees may become disillusioned, and as a result, unmotivated. Maurer (2001) and Salthouse and Maurer (1996) describe how exposure to negative age stereotypes may influence older workers’ self-efficacy or self-confidence for learning and development, a key factor in motivation. Maurer and Rafuse (2001) describe how a workplace which is not supportive of the development of older workers could adversely affect the older workers’ “terms, conditions, and privileges of employment,” a legal term used to define illegal discrimination. Perceived discrimination may also have influences on older worker’s selfesteem, personal control, and satisfaction. In a study by Hassell and Perrewe (1993), perceived age discrimination was measured by assessing employees’ personal experiences with age discrimination in their workplaces. Perceived discrimination had a significant interaction with age in terms of global self-esteem. The global self-esteem of older workers who perceived age discrimination was lower than those who did not. Also, both age and perceived age discrimination had an interactive effect in relation to personal control. While perceptions of control generally increased with age, they were found to decrease in cases where age discrimination was perceived. Older workers who perceived age discrimination were also less satisfied with growth opportunities. In a study by Miller, Cox, Gieson, Bean, Adams-Price, Sanderson and Topping (1993), older employees who believed that perceived performance deterioration of older employees exists and who perceived preferential treatment of younger employees were more likely to experience a low degree of job involvement. Also, older employees who believe that perceived performance deterioration of older employees exists were more likely to experience alienation from the job.

Toward Understanding and Managing Stereotypical Beliefs

265

Although stereotypical beliefs about older workers’ learning and development behavior might be possessed by older workers themselves, this is not to say that all older workers or even most of them will have strong stereotypical beliefs. Perhaps older workers would be less likely to possess strong beliefs along these lines. However, there may be individual differences in the degree to which all people possess these beliefs, including older workers. Possessing these beliefs to varying degrees may be related to actions by the older workers. To the extent that older workers believe that, in general, older workers do have more difficulty and also generally are less interested in learning and development, this may affect their specific beliefs about their own ability to develop and the utility of that pursuit. However, research has generally not explored these potential linkages, and more research seems warranted. It seems likely that these stereotypical beliefs could influence several selfrelevant cognitions and motivationally important variables. As mentioned above, stereotypes may influence self-efficacy for development, to the extent that such stereotypes help persuade the person that he/she is less capable of performing learning tasks or serve as abstract “models” depicting older workers as incompetent (Bandura, 1997; Maurer, 2001; Salthouse & Maurer, 1996). Likewise, as older workers look to their context for role definitions and appropriate behavior (Greller & Stroh, 1995), stereotypes may influence self-concepts regarding learning-relevant capabilities. Further, stereotypical beliefs that older workers typically cannot develop and do not want to develop may lower perceptions by older workers that engaging in learning and development will be intrinsically rewarding or that financial and tangible rewards are linked to that activity. These are two perceived benefits that have been tied to motivation and participation in development (Maurer & Tarulli, 1994). To the extent that both self-efficacy and perceived rewards are lower, motivation and interest in development should also be lower. Proposition 2. Older workers who possess beliefs that aging workers typically are not able nor interested in learning and development will have lower self-relevant cognitions related to learning (e.g. self-concept as “developer,” lower self-efficacy for development, fewer perceived personal benefits for development such as intrinsic or extrinsic rewards), and their motivation and participation in development will also be lower. Additionally, although research has addressed predictors of retirement such as finances, work characteristics, and non-work characteristics (cf. Beehr et al., 2000), research has generally not investigated the potential effects of stereotypical beliefs about learning and development capability in relation to this decision. This is critical because learning and development are getting increasingly important

266

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

for job performance in the work place. Based on the literature described above, beliefs about the appropriateness of learning and development for older workers may impact subsequent intentions for job-related behavior, including continuing to work. Proposition 3. Older workers employed in jobs requiring learning who possess beliefs that aging workers typically are not able nor interested in learning and development will be more likely to have intentions to retire sooner than older workers who do not possess these beliefs, holding other predictors of retirement constant. In summary, more research on this topic is clearly needed. It appears that not only may beliefs exist that older workers are neither able nor inclined to successfully pursue learning and development activities at work, but also there may be serious consequences of such beliefs. Such beliefs may contribute to a lack of access to and support for learning and development activities and resources by older workers, self-fulfilling prophecies, reduced motivation, and higher probability of withdrawal (e.g. retirement). Additionally, if older workers are not maintaining and developing knowledge and skills, then this could represent a significant liability for employing organizations. Although literature does suggest the existence of these beliefs and the potential effects they have, there is relatively less understanding of the specific origins and basis for these beliefs. In this model, the possible antecedents of such beliefs are outlined next. An understanding of antecedents is likely to be helpful for providing ideas for future research and clues on how to affect the beliefs in practice.

Antecedents of Beliefs About Older Workers’ Ability and Inclination to Learn/Develop This section will describe three major categories of factors that will have an influence on beliefs about older workers’ ability and inclination to learn and develop: Experience and promulgation, Perceived learning and development inhibitors internal to the older worker, and Perceived learning and development inhibitors external to the older worker. These antecedents are posited to be based on aggregated experiences and information which lead to beliefs about typical behavior (e.g. older workers in general). However, in a later section of this paper these antecedents will also be contrasted with another key component of the model: Individuating information and knowledge of within-group variability. This information can operate to attenuate the influence of beliefs about typical behavior of older workers on behavior by administrators and behavior by older workers themselves.

Toward Understanding and Managing Stereotypical Beliefs

267

Experience and Promulgation The first category of influences on the belief that older workers cannot and do not want to develop is “Experience and Promulgation.” This category consists of three influences. These influences are: first-hand observations of stereotypeconsistent or confirming behavior, insufficient information to disconfirm or negate the stereotype, and promulgation of the stereotype by others. First-hand observations of stereotype-consistent behavior can come from many different sources. Observation of instances of failure and/or lack of participation or interest on the part of an older person at work can have a very strong effect on an individual’s beliefs. The fact that some older workers do not participate in training as often as younger adults also helps to reduce the number of available stereotype-disconfirming instances. Rosen, Williams and Foltman (1965) reported that younger adults are more likely to volunteer for training. Cleveland and Shore (1992) reported that age was negatively related to developmental experiences. Tucker (1985) investigated training needs of technology and management workers in the U.S. Geological Survey. Most of the workers indicated that their last training experience was less than one year ago. The majority of those who answered “more than three years ago” were in the sixty and over age group. Warr (1994b) reported from the 1991 Labor Force Survey that the proportion of older workers participating in job-related training in the last four weeks and the hours of off-the-job training in the last week declined across age. In the 1992, Institute of Management Survey, age predicted training incidence and duration negatively irrespective of tenure. Those under forty reported a median of ten days training in 1991; those over sixty reported half as many days. This difference was due to activities during work time. Days of training activities on their own time were the same. Some older workers may not be able to succeed in training under the same conditions as younger workers. Studies conducted in laboratory settings using real-world training simulations have demonstrated several adjustments that may be necessary for older trainees to succeed. Using exercises that are self-paced rather than those with a time limit (Gomez, Egan & Bowers, 1986), allowing plenty of time (Elias, Elias, Robbins & Gage, 1987; Siemen, 1976; Zandri & Charness, 1989), different social contexts for training (Zandri & Charness, 1989), having plenty of help available (Elias, Elias, Robbins & Gage, 1987), and reducing stress associated with the training environment (Jamieson, 1969) are all strategies that may facilitate older workers’ performance in training. In traditional training situations where such accommodations are not available, some older learners’ performance may suffer, providing a situation in which the stereotype of older learners as less capable will be supported.

268

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Regardless of the underlying reasons, the point remains that older adults are viewed less frequently in training and development situations, and their performance may be, on average, lower than that of their younger colleagues (Kubeck, Delp, Haslett & McDaniel, 1996). This can have a profound impact on one’s beliefs about older workers’ capability and motivation for professional development. People may, in addition to observing stereotype-supporting situations, lack the information they need to invalidate their beliefs. Research on stereotypes has put forth several ways in which individuals may end up with a lack of information to disconfirm their stereotypes. People may be more likely to attend to things that confirm their stereotype, so they may not remember disconfirming occurrences even when they are available. People may attend longer to confirming than to disconfirming information, and prefer to receive stereotype-confirming information when given a choice (Erber & Fiske, 1984; Neuberg & Fiske, 1987; Ruscher & Fiske, 1990). Individuals may also encode less information when they can use existing stereotypes to fill in missing information, so even if disconfirming information is available, it may not be encoded and thus, may not be remembered (Von Hippel et al., 1993). So even if there are disconfirming instances available (older adults engaging in development, etc.), people may preferentially recall the instances in which the stereotype is confirmed. Even recognizing and remembering a few notable exceptions need not change the more global stereotype. If an individual creates a “subgroup,” or a smaller stereotype within the larger main stereotype, then the main stereotype can stay intact. For example, if an individual knows an older employee who has attended multiple training seminars on computer use and is an expert user, he may classify this older adult as a sort of “exception to the rule” and place him in his own category away from other older adults. This subgrouping process is especially likely when the subgrouped individual does not fit the existing stereotype in other ways too (Lambert, 1995), because then he is not seen as being representative of the larger stereotyped group. In order for disconfirming information to have an impact on an individual’s stereotype, it must be widely dispersed throughout several members of the group who are typical of the group in other ways (Crocker, Hannah & Weber, 1983; Hewstone, Johnston & Aird, 1992; Hewstone et al., 1994; Johnston & Hewstone, 1992). In this way, subgrouping may perpetuate the stereotype, even in the face of contradictory behavior by a few individuals (Allport, 1954). Another phenomenon that would make stereotype-confirming incidents more salient than disconfirming ones occurs within the context of ingroup–outgroup perceptions. For example, if younger workers see themselves as part of a “younger worker” ingroup and place older workers into an outgroup, the younger workers will thus perceive older workers as having very little variability as a group (Mullen & Hu, 1989). This would mean that a single instance of an older worker’s

Toward Understanding and Managing Stereotypical Beliefs

269

failure in training or absence from a training program would be very likely to be integrated into the younger worker’s stereotype of older workers as a group. Since the outgroup is perceived as having little variability, the actions of a single member are easily generalized to the rest of the group (Judd, Ryan & Park, 1991). In cases in which sufficient motivation is present, people can sometimes override their automatic stereotypical response (Fiske, 1998). However, if one is not exposed to older workers engaging in various development activities and being successful in them, or does not encode this information when it is present, they will not have enough disconfirming information available to process. One can also be exposed to stereotypical beliefs through statements of opinion by others. This type of stereotype promulgation might occur in the context of explicit managerial justification (i.e. not allocating money for training older workers), or it could occur implicitly as an employee notices that older adults are not selected to attend training. It could also occur more directly when others express the stereotype-consistent opinions directly through statements that older workers are too hard to train (e.g. “old dog, new tricks”) or don’t want to learn (e.g. “stuck in their ways”). According to social information processing theory (Zalesny & Ford, 1990), attitudes and ideas in the work environment can have a powerful effect on an individual’s own development of or adherence to similar attitudes. Although it seems likely that experience with stereotype-consistent behavior and promulgation of the belief by others should have an effect on the belief that older workers are not able nor motivated to learn and develop, little or no research has examined these linkages. Research on this seems needed. Based on this discussion, the following propositions are offered. Proposition 4a. First-hand experiences and observations of older workers performing lower in training and learning experiences and not participating in learning and development experiences will increase the beliefs that older workers are not able nor interested in learning and development. Proposition 4b. Promulgation of the beliefs by others at work will increase the stereotypical beliefs by observers.

Perceived Internal Inhibitors This category of influences includes both the beliefs about general changes or differences with age in mental and physical characteristics as well as changes in motives that may be relevant to many domains of behavior, including learning and development at work. These beliefs are antecedents or contributors but are conceptually distinct from the belief about learning and development activities.

270

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

That is, one’s beliefs about the decline of abilities with age and the controllability of these changes, general physical and mental health stereotypes, and beliefs about interests, needs, values, or motives may have implications for many behaviors, including training and development. Thus, these more general age-related beliefs could contribute to beliefs that older workers are not able and/or interested when it comes to learning and developing at work. This set of factors is a potentially important influence, especially since these factors involve negatively valenced personal characteristics of the older worker that may be considered stable personal and dispositional aspects of the person (Pettigrew, 1979). In fact, there is good reason to believe that individuals may hold many negative beliefs about older people. Nuessel (1982) looked at terms in society that are used to describe older individuals and found that they were overwhelmingly negative. Nussbaum and Robinson (1984) examined fifty articles from popular magazines in order to observe the portrayal of aging and they were found to reinforce negative stereotypes held by the population in general. In fact, people may hold beliefs about older people in many different domains (e.g. physical characteristics, health, personality, social characteristics, emotions, cf. Schmidt & Boland, 1986), and many (though not all) of these beliefs may be negative. To the extent that these domains have implications for learning and development behavior, they may have implications for beliefs about that behavior. For example, research has shown that aging people may be increasingly likely to be thought of as being forgetful, absentminded, or slower (Heckhausen, Dixon & Baltes, 1989; MacNeil, Ramos & Magafas, 1996), less creative (Rosen & Jerdee, 1976b), and possessing lower physical abilities (Netz & Ben-Sira, 1993; Rosen & Jerdee, 1976b; Slotterback & Saarnio, 1996). In fact, there may be a high consensus about the nature of expected changes during adulthood, and research has suggested that people believe that there is an increasingly larger proportion of expected losses and fewer expected gains associated with older ages. Heckhausen and Baltes (1991) show that increases in undesirable attributes were expected to be more likely in older adulthood and that aging-related increases in undesirable attributes may be less controllable (e.g. nothing can be done to mitigate the negative changes). Research has shown a relationship between one’s age and a perception of decline in memory abilities (Ryan & See, 1993; see Dixon, 1989, for a review), and has also shown a decline in various cognitive reasoning capabilities with age (Verhaeghen & Salthouse, 1997). Other research has suggested that the motives and needs of older people may change (cf. Salthouse & Maurer, 1996). Learning, growth and challenge may have less importance to older workers as suggested in a meta-analysis by Engle, Miguel, Steelman and McDaniel (1994) in which small mean age differences in growth needs were observed. Hall and Mansfield (1975) reported that the

Toward Understanding and Managing Stereotypical Beliefs

271

importance of esteem, security, and affiliation increased with age while the importance of self-actualization decreased. Markus and Herzog (1991) assert that older adults may have a greater interest in the prevention of feared outcomes and a lesser interest in occupational or career-oriented goals. If people perceive negative changes to occur with age in general characteristics of the person which are also relevant to learning and development at work, these general beliefs may lead one to hold the more specific beliefs that older workers are not very capable of successful learning and development due to declines in these key characteristics. Further, to the extent that the general motives or needs of older people are perceived to change, beliefs about their interest in pursuing learning and self-development activities may change in kind. Moreover, if people believe these changes are not controllable or changeable, then it is likely they will be perceived as stable, internal to the individual and potentially capable of negatively influencing development. Research has generally not examined these types of variables for relationships with beliefs about the learning and development behavior of older workers. Proposition 5. Those people who have greater beliefs in general age-related decline of mental and physical characteristics will be more likely to possess the more specific beliefs that older workers are less likely to be able and willing to learn and develop at work. Note that research should attempt to identify which specific types of mental or physical characteristics are most closely associated with beliefs about development behavior. At this point, the literature provides very little guidance.

Perceived External Inhibitors This category includes beliefs about organizational and social processes outside of the older person that inhibit his/her ability and motivation to learn and develop at work. This includes a perceived lack of access to training and development resources by older workers as well as discriminatory practices. These perceptions may contribute to the belief that older workers have great difficulty at learning and development. Likewise, perceived social and organizational constraints such as less frequent membership in important social networks, less support or incentives from colleagues, superiors, or non-work social members with respect to learning and development may also contribute significantly to a belief that older workers will not learn and develop effectively at work, nor be interested in that goal. When the context is perceived to impede the person, it is likely that observers’ expectations for success by the person will be eroded.

272

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Maurer and Tarulli (1996) found that subordinates and peers of managers within an organization were more likely to believe that the managers were capable of developing their skills to the extent that they also believed the managers had social support for development, had access to resources that would help the mangers develop, and also had time available to them at work which they could use to improve and develop their skills. While this research can be applied to employees of any age, it has particular relevance to this model in that it suggests that beliefs about important factors in another employee’s work context may influence beliefs about that person’s ability to develop. That is, perceived external inhibitors in an individual’s environment can impact others’ perceptions of him or her. Applied to older workers, it implies that if people believe there are external constraints on an older person’s pursuit of learning and development at work, they logically are likely to believe that the older worker is less likely to be able to learn and develop. Stagner (1985) reviews some research that suggests that people are aware of discriminatory treatment and obstacles that older workers may face when it comes to the pursuit of training. Mercer (1981) reported that 61% of a sample of executives agreed that older workers are discriminated against, although these executives denied that such discrimination took place in their own firms. Likewise, Tieger (1994) notes that “the well known and rather widespread consequences (of prejudice against older workers in terms of their learning capabilities) are that workers are then excluded . . . especially from access to the continuous training given in companies” (p. 79). Older workers may also be confronted with diminishing social support from others regarding their learning and development, and social support has been shown to be significantly related to employee development activities (Maurer & Tarulli, 1996; Noe & Wilk, 1993; Tharenou, Latimer & Conroy, 1994). Likewise, older workers are less likely to receive career counseling from supervisors (Cleveland & Shore, 1992). Kram and Isabella (1985) suggest that older workers may not form peer relationships that would lead to psychosocial support. Adding to the potential problem of restricted access is the fact that many managers think that the amount it would cost to train older employees would not be recovered because their potential years of service to the company are fewer (Stagner, 1985; Warr, 1994a). That is, training older workers is not seen as being worth the investment. Fossum, Arvey and Paradise (1986) discusses this dilemma with regard to human capital theory. Fossum et al. suggest that from a managerial standpoint, organizations must decide which employees will receive specific human capital (SHC) investments, and these judgments may consider the rate of return expected on the investment. From this point of view, older workers may be considered more prone to obsolescence.

Toward Understanding and Managing Stereotypical Beliefs

273

Perceptions of decreasing access to resources, impediments, diminishing incentives and a shrinking support for workers’ learning and development with age may contribute to a belief that older workers will find it difficult to learn and develop, and they may not have much interest in this pursuit. Thus, perceived inhibitors external to the older worker may also contribute to beliefs that he/she is not able or does not want to develop and learn. Proposition 6. Those people who hold greater beliefs that organizational and social processes generally inhibit older workers’ learning and development behavior will be more likely to possess the beliefs that older workers are typically less able and willing to learn and develop at work. Although it seems that the antecedents described above should have effects on the stereotypical beliefs, and as discussed previously, the beliefs could have effects on important outcomes (behavior by others, behavior by older workers), it also seems likely that people can possess individuating information and knowledge of within-group variability, and that this information/knowledge could attenuate the effects that stereotypical beliefs will have.

Individuating Information and Knowledge of Within-(Older) Group Variability It is important to note that many stereotypes have some truth as a basis. There are older workers who have difficulty learning and/or who have lost interest in learning and developing at work. However, there are younger workers with these characteristics too. Most importantly, the variation in ability and desire among older workers is notable. When considering the experience and promulgation effects discussed above, it is important to examine certain data reported in the meta-analysis by Kubeck et al. (1996) that addressed age differences in training performance. Those authors reported that in 35 of the 43 samples providing variance data, the variance for older adults’ performance was larger than for younger adults. They suggest that this indicates greater individual differences among older adults than among younger adults. In light of these results, the characteristics of individuals need to be considered independently from aggregate findings about a group of which he or she is a member. Further, the study by Kubeck et al. (1996) found that although there was a negative relationship between age and training mastery performance (−0.26), based on data from five samples and three studies there was a negative relationship between age and pretraining competence of similar size (−0.21). This suggests that much if not all of the performance differences associated with age after

274

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

training could be accounted for by existing competence differences before training. There was little data available to directly reflect actual amount learned in training (gain scores), but available data from four studies showed only a comparatively much smaller negative relationship (−0.09) between age and amount learned in training. This suggests that existing cohort differences between age groups (e.g. education, experience/skill with technology, etc.) might be causing the pretraining differences. This means that if equally competent older and younger workers are selected for training, the negative effects on success might be much smaller. In a later study, Warr, Allan and Birdi (1999) found that knowledge gain scores before and after training were associated with age −0.25 (p < 0.01), although the association could be accounted for in multivariate analyses by other variables. The authors linked that finding to age differences in constructs such as technical qualification, use of a help-seeking learning strategy, and learning confidence. Those authors found no age differences in later behavior on the job in terms of frequency of use of the learned technology, a goal of the training. These facts from the Kubeck et al. (1996) meta-analysis and the Warr et al. (1999) study have important implications. First, there is great variability within groups of older workers in training performance – more so than younger workers. This means that group averages do not represent older individuals very well, and thus, stereotypes based on averages may not be very accurate in describing individual behavior. Second, if individual differences are considered in selecting people for training (e.g. pre-training competence and other training-relevant characteristics), the negative age effects suggested in this prior research may be significantly smaller. Knowledge of the variability and individuating effects should help to attenuate effects of group-based stereotypes on assessments of individuals. To the extent that there is individuating information provided about an individual, the impact of stereotypes may be reduced. In the meta-analysis on age discrimination effects by Finkelstein, Burke and Raju (1995), one study looked at the effects of providing positive information about both older and younger targets. In this study, older workers about whom there was positive information got higher ratings on potential for development, as opposed to findings in other research which found lower ratings when positive information was not provided. Further, directly relevant to effects of perceived internal inhibitors discussed above, in an analysis of results from several studies published in Psychology and Aging and the Journal of Gerontology (Morse, 1993) reported that older adults as a group are more variable than younger groups in basic and general abilities like memory, fluid intelligence and reaction time. This suggests that there are more differences among older people in general abilities relevant to learning and development than among younger people. Further, differences in motivational

Toward Understanding and Managing Stereotypical Beliefs

275

constructs such as growth needs that are associated with age are quite small (Engle et al., 1994), bringing to question the importance of the any mean differences in defining older adults as a group. Finally, although literature discussed above illustrated that external inhibitors such as social exclusion and discriminatory practices might affect older workers’ ability and inclination to learn and develop, it is possible that individuating information regarding potential inhibitors within one’s own organization may lead people to perceive it as being free from ageism. For example, as discussed in Stagner (1985), 61% of a sample of executives agreed that older workers are discriminated against, although the executives denied that such discrimination took place in their own firms. There are steps that organizations can take to rid their cultures of ageism and perceived roadblocks for older workers (cf. Maurer, 2001; Maurer & Rafuse, 2001). This discussion is not meant in any way to suggest that there are not important age differences and changes that could negatively affect learning and development behavior by older workers. In fact, there is research to suggest that negative age-related effects do exist that are both meaningful and potentially practically significant (cf. Kubeck et al., 1996). However, it is interesting to contrast the literature on aging, which reflects important differences in performance and personal characteristics between older and younger groups, with literature on race and sex. In literature on race and sex differences, there are also important differences in certain job-relevant characteristics such as general mental ability or physical strength. However, just as important in that literature is the idea that people should be treated as individuals because of wide variability within race and sex category. Stereotypical ideas that broadly portray members of those categories in an unfavorable light are offensive. However, as reported by Maurer and Rafuse (2001), although most employees are very sensitive to sexism and racism, somehow the idea that “you can’t teach an old dog new tricks” does not seem to carry the same taboo status in society and the work place. It is just as important in this area to be sensitive to the fact that one size does not fit all, and there can be older individuals who are both interested in and able when it comes to learning. Given this discussion, it appears that some research has found significant variability in training performance by older workers and also important age differences in preexisting competence and other variables which may account for performance differences at the conclusion of training. Other research has examined the influence of individuating information on age discrimination in ratings of potential for development. Still other research has found larger variability in basic and general abilities like memory, reaction time and fluid intelligence among older people. Also, in the literature it has been illustrated that steps can be taken to attempt to rid an organization of ageism and to potentially reduce perceptions

276

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

that there are constraints on older workers’ learning and development. However, no research has examined the interrelationships of: (a) beliefs about older workers ability and inclination to develop; (b) knowledge of the performance variability within older groups in training and in basic, general abilities and motives; (c) individuating information about specific older workers; (d) actions taken by organizations to eliminate perceptions of constraints on older workers; and (e) behavior by observers or older workers themselves. Based on the discussion above, several predictions seem possible. Proposition 7. Higher amounts of individuating information regarding specific older workers’ ability and desire for learning and also knowledge about within(older)group variability in relevant characteristics will: (a) be associated with weaker beliefs that older workers are typically unable and unwilling to learn and develop; (b) attenuate the effect of the stereotypical beliefs on behavior by others and by older workers. Higher amounts of individuating knowledge about the organization’s attempts to eliminate constraints on older workers in learning and development will have similar effects.

Processes and Tactics that Should Provide Individuating Information and Knowledge of Within-(Older) Group Variability There are numerous reasons to avoid stereotype-based decisions and actions in the area of training and development. First, there are potential legal implications in this area (Maurer & Rafuse, 2001), and avoiding litigation is always a motivator. Second, differential treatment may have implications for the effectiveness of older workers themselves, given the importance of continuous learning at work. Third, utilization of the human resources offered by older workers is a key opportunity in a very tight workforce increasingly populated by larger numbers of older workers. A good business case can be made for effectively managing this aspect of workforce diversity. While the review above highlights places where research is necessary, the review and framework provide some basis for tentatively offering ideas for dealing with age stereotypes in the workforce. Specifically, in the previous section of this paper, individuating information and knowledge of within older-group variability was described as key to dealing with older workers as individuals and to attenuating the effects that general stereotypical beliefs may have. In the next section of this paper, several processes and tactics will be outlined that may provide this knowledge. These will be covered within the three categories

Toward Understanding and Managing Stereotypical Beliefs

277

of factors proposed to contribute to the stereotypical beliefs: Experience and Promulgation, Perceived Internal Inhibitors, and Perceived External Inhibitors. Experience and Promulgation In the “Experience and Promulgation” category, the key is to provide the information that people need to move beyond their general stereotypical beliefs. Research in the stereotyping literature provides support for the idea that increasing the individuality of members of the “other” group can reduce inter-group discrimination. For example, in a simulated group situation, people were less likely to discriminate against an outgroup when they were told that the group did not unanimously agree on something; knowledge of even one dissenter was enough to reduce the amount of discrimination (Wilder, 1978). Lee and Clemons (1985) reported that, in a simulated employment decision context, interviewers made more favorable decisions about older applicants when they had specific behavioral performance information about that individual. Taylor, Crino and Rubenfeld (1989) found that actually having worked with older employees was related to more positive impressions of job performance of workers over age sixty-five, suggesting that actual exposure to and contact with older workers has enhanced impressions of members of that group. From this kind of research, we can conclude that stereotypical information is most diagnostic when little or no other information is available to the person forming the impression. When additional information becomes available about an individual, he or she becomes less a member of a uniform group and more an individual. Taylor et al. (1989) suggested that perhaps employers should assign older workers to jobs involving frequent interactions with younger workers. Along these lines, perhaps managers should mix older workers and younger workers in work teams to increase exposure and interaction to foster the perception of older workers as individuals. In addition, organizations should strive to ensure that employees are considered for work-related opportunities based on their individual qualifications and characteristics, rather than on group-based stereotypes or impressions. For example, explicit individual development plans and discussions with all employees may help to develop perceptions of each worker as an individual and avoid assumptions based on age (cf. Maurer & Rafuse, 2001). Another important strategy for reducing promulgation of stereotypes is to reduce the amount of ageism in an organization. This can be done by “signaling the inappropriateness of ageist remarks, jokes and attitudes just as you would for racist or sexist ones” (Maurer, 2001, p. 135). Ageism should be incorporated into the company diversity awareness curriculum. Maurer and Rafuse (2001) suggest providing training for managers concerning the Age Discrimination and

278

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Employment Act (ADEA) and the possible effects of stereotypes. This may be another strategy for decreasing promulgation of stereotypes in an organization. Perceived Internal Inhibitors With regard to internal inhibitors, it seems critical to provide an education to employees in some form regarding the idea that the variability within groups is much greater than the variability between groups, just as efforts have been made in dealing with race and sex. As with any group, older people posses a wide range of abilities related to training and development. While age-related differences between older and younger groups are present, in any one group, people of all ages would likely be performing at many different levels, and quite possibly some older workers could be outperforming younger ones. Steps could be taken to include information regarding the variability among older workers in company diversity awareness programs. Perhaps through exposure to effective older workers in work groups or teams, the knowledge could also be obtained. Perceived External Inhibitors Because external inhibitors are largely environmental characteristics, one way these can be addressed is through formal policies and procedures. Ensuring adequate training opportunities at all ages and in all levels of the company would imply an open training system. The organization needs to create an environment of equal access for all employees to receive the training they need. Maurer and Rafuse (2001) summarize ways in which organizations can scrutinize management practices and policies to ensure equal treatment of older workers in training and development, avoiding the pitfalls of stereotype-based actions. For example, managers can be rewarded for meeting employee training and development goals of all employees. Upward feedback systems can help monitor managers’ success in developing their employees. Also, employees should receive feedback regarding their development behavior, and goals for future behavior should be set. These actions serve to encourage the learning and development of all employees. They also ensure that the organization is perceived as one that provides learning and development to all employees, including older workers. This serves to decrease the number of perceived external inhibitors to the development of older workers. Each of these three categories of recommendations has a place in a company or organizations’ larger effort to reduce unfair treatment of its employees. Many of the ideas presented here might be folded into a training program designed to reduce stereotyping. Special training could be given to personnel decision-makers regarding appropriate use of individuating information in selection, promotion, and training situations.

Toward Understanding and Managing Stereotypical Beliefs

279

Proposition 8. Tactics and processes that provide: (1) exposure to successfullyperforming older workers actively learning/developing; (2) information about the variability of older workers’ learning-relevant attributes and motives; and (3) information that reflects a lack of ageism and obstacles to development in an organization will increase the amount of individuating information and knowledge of within-(older) group variability that people possess.

Additional Considerations for Future Research While the above propositions provide a multitude of opportunities for exploration into the area of stereotypical beliefs about older workers’ ability and desire to develop, there are also a number of methodological issues which warrant consideration. As mentioned earlier, there has been considerable disagreement among researchers as to what constitutes an “older worker.” According to Ashbaugh and Fay (1987) a variety of mostly arbitrary methods have been used to define the threshold age. Perhaps careful attention in future research can be given to operationalizing the term “older worker.” Ashbaugh and Fay’s (1987) finding that the mean age used to define older worker was 53.4 years provides a useful starting point; however, it may be difficult to identify an age that applies to every setting. The relative ages of others in a given setting or industry (Cleveland, Shore & Murphy, 1997; Lawrence, 1984) may make a widely generalizable definition of “older” elusive. Research which explores constructs such as relative age in relation to individual, situational, motivational and employee development constructs may help to further delineate age-related constructs as they specifically relate to the learning and development of older workers (cf. Maurer, Weiss & Barbeite, 2003). Another methodological issue mentioned earlier is the range of research participants age. The majority of research examining perceptions of older workers seems to have been conducted using undergraduate students (Finkelstein, Burke & Raju, 1995). While undergraduates serve as a very valuable research population, perceptions of older workers’ ability and desire to develop should be examined using participants of varying ages, and more work using adults beyond traditional college age is needed. In order to obtain maximum generalizability a variety of age groups including older workers themselves should be included in such research. This is particularly important given that there is some evidence to suggest that younger people and older people may differ in their perceptions of older workers (Finkelstein, Burke & Raju, 1995). It may also be important to examine any relevant cohort or period effects. That is, are perceptions of older workers such as those studied by Rosen and Jerdee almost thirty years ago similar to those found among

280

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

people today? Do Baby Boomers and Gen X’ers perceive older workers differently? Legislation such as the Age Discrimination in Employment Act may also have had an impact on people’s beliefs about older workers, resulting in cohort effects. A replication of Rosen and Jerdee’s (1976b) study conducted by Weiss and Maurer (in press), failed to replicate the original findings, suggesting that this research may be in need of updating. The use of a wide range of participants and the investigation of cohort effects could help to further understand perceptions of older workers’ ability and desire to develop and to establish important boundary conditions for findings. Finally, it would be useful to attempt to bridge the gap between research on perceptions of older workers’ capabilities (and perceptions of declines) and research on actual capabilities and declines. Knowing where stereotypes depart significantly from actual, mean trends in performance data and where they do seem to correspond to true group differences might be useful not only in enhancing understanding, but also as a point of departure in working to manage both older workers and those who work with them. General Summary Given the importance of continuous learning and development for both workers and organizations, combined with an increasing number of older workers in the workforce, the issue of training and development of older workers is increasingly critical. However, stereotypical beliefs exist which portray older workers as typically unable and uninterested in learning and development. These beliefs may have detrimental effects not only on evaluators and decision-makers in organizations, but also possibly older workers themselves. Therefore, an understanding of the origins and basis for these beliefs is important, as is an understanding of how the effects of these beliefs on individuals can be attenuated. A model was presented in the current paper which is intended to enhance understanding of these beliefs and which may help guide future research on this topic. The framework also provides some suggestions for dealing with these beliefs in practice.

ACKNOWLEDGMENTS
An earlier version of this paper was presented at the Annual Conference of the Academy of Management, Washington, DC, August 2001. Preparation of this article was partially supported by National Institute on Aging grant R01 AG1 4655.

Toward Understanding and Managing Stereotypical Beliefs

281

REFERENCES
AARP (1995). American business and older workers: A road map to the 21st century. Washington, DC: DYG. AARP (2000). American business and older employees. Washington, DC: Author. Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Aries, P. (1962). Centuries of childhood: A social history of family life. New York: Vintage Press. Ashbaugh, D. L., & Fay, C. H. (1987). The threshold for aging in the workplace. Research on Aging, 9, 417–427. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman. Beehr, T. A., Glazer, S., Nielson, N. L., & Farmer, S. J. (2000). Work and nonwork predictors of employees’ retirement ages. Journal of Vocational Behavior, 57, 225–260. Boerlijst, J. (1994). The neglect of growth and development of employees over 40 in organizations: A managerial and training problem. In: J. Snel & R. Cremer (Eds), Work and Aging: A European Perspective (pp. 251–271). London: Taylor & Francis. Bureau of Labor Statistics (1995). Survey of employer-provided training. Washington, DC: Author. Caplan, L. J., & Schooler, C. (1990). The effects of analogical training models and age on problemsolving in a new domain. Experimental Aging Research, 16, 151–154. Capowski, G. (1994). Ageism: The new diversity issue. Management Review, 83, 10–15. Cleveland, J. N., & Shore, L. M. (1992). Self and supervisory perspectives on age and work attitudes and performance. Journal of Applied Psychology, 77, 469–484. Cleveland, J., Shore, L., & Murphy, K. (1997). Person and context oriented perceptual age measures: Additional evidence of distinctiveness and usefulness. Journal of Organizational Behavior, 18, 239–251. Cole, M. (1992). Culture in development. In: M. H. Bornstein, M. E. Lamb et al. (Eds), Developmental Psychology: An Advanced Textbook (3rd ed., pp. 731–789). Hillsdale, NJ: Lawrence Erlbaum Associates. Crocker, J., Hannah, D., & Weber, R. (1983). Person memory and causal attributions. Journal of Personality & Social Psychology, 44, 55–66. Dixon, R. A. (1989). Questionnaire research on metamemory and aging: Issues of structure and function. In: W. L. Poon, D. C. Rubin et al. (Eds), Everyday Cognition in Adulthood and Late Life (pp. 394–415). New York: Cambridge University Press. Doering, M., Rhodes, S. R., & Schuster, M. (1983). The aging worker. Beverly Hills, CA: Sage. Downs, S. (1967). Labour turnover in two public service organizations. Occupational Psychology, 41, 137–142. Elias, P. K., Elias, M. F., Robbins, M. A., & Gage, P. (1987). Acquisition of word-processing skills by younger, middle-age, and older adults. Psychology and Aging, 2, 340–348. Engle, E. M., Miguel, R. F., Steelman, L. A., & McDaniel, M. M. (1994, April). The relationship between age and work needs: A comprehensive research integration. In: H. Sterns (Chair), Age related differences in work attitudes and behaviors. Symposium conducted at the meeting of the Society for Industrial and Organizational Psychology, Nashville, TN. Erber, R., & Fiske, S. (1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709–726. Finkelstein, L. M., Burke, M. J., & Raju, N. S. (1995). Age discrimination in simulated employment contexts: An integrative analysis. Journal of Applied Psychology, 80, 652–663.

282

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Fiske, S. (1998). Stereotyping, prejudice, and discrimination. In: D. T. Gilbert, S. T. Fiske & G. Lindzey (Eds), The Handbook of Social Psychology (pp. 357–411). New York: McGraw-Hill. Forte, C. S., & Hansvick, C. L. (1999). Applicant age as a subjective employability factor: A study of workers over and under age fifty. Journal of Employment Counseling, 36, 24–34. Fossum, J. A., Arvey, R. D., Paradise, C. A., & Robbins, N. E. (1986). Modeling the skills obsolescence process: A psychological/economic integration. Academy of Management Review, 11, 362–374. Fullerton, H. N. (1999, November). Labor force projections to 2008: Steady growth and changing composition. Monthly Labor Review, 122, 19–32. Gomez, L. M., Egan, D. E., & Bowers, C. (1986). Learning to use a text-editor: Some learner characteristics that predict success. Human-Computer Interaction, 2, 1–23. Greller, M. M., & Stroh, L. K. (1995). Careers in midlife and beyond: A fallow field in need of sustenance. Journal of Vocational Behavior, 47, 232–247. Haefner, J. E. (1977). Race, age, sex, and competence as factors in employer selection of the disadvantaged. Journal of Applied Psychology, 62, 199–202. Hall, D. T., & Mansfield, R. (1975). Relationships of age and seniority with career variables of engineers and scientists. Journal of Applied Psychology, 60, 201–210. Hall, D. T., & Mirvis, P. H. (1995). The new career contract: Developing the whole person at midlife and beyond. Journal of Vocational Behavior, 47, 269–289. Hassell, B. L., & Perrewe, P. L. (1993). An examination of the relationship between older workers’ perceptions of age discrimination and employee psychological states. Journal of Management Issues, 5, 109–120. Heckhausen, J., & Baltes, P. B. (1991). Perceived controllabilty of expected psychological change across adulthood and old age. Journal of Gerontology, 46, 165–173. Heckhausen, J., Dixon, R. A., & Baltes, P. B. (1989). Gains and losses in development throughout adulthood as perceived by different age groups. Developmental Psychology, 25, 109–121. Heron, A., & Chown, S. M. (1961). Aging and the semi-skilled: A survey in manufacturing industry on Merseyside. London: Her Majesty’s Stationery Office. Hewstone, M., Johnston, L., & Aird, P. (1992). Cognitive models of stereotype change: II. Perceptions of homogeneous and heterogeneous groups. European Journal of Social Psychology, 22, 235–249. Hewstone, M., Macrae, C., Griffiths, R., & Milne, A. (1994). Cognitive models of stereotype change: Measurement, development, and consequences of subtyping. Journal of Experimental Social Psychology, 30, 505–526. Institute of Personnel Management (1993). Age and employment: Policies, attitudes and practice. London: Institute of Personnel Management. Jamieson, G. H. (1969). Age, speed, and accuracy: A study in industrial retraining. Industrial Gerontology, 8, 50–51. Johnston, L., & Hewstone, M. (1992). Cognitive models of stereotype change: III. Subtyping and the perceived typicality of disconfirming group members. Journal of Experimental Social Psychology, 28, 360–386. Judd, C., Ryan, C., & Park, B. (1991). Accuracy in the judgment of in-group and out-group variability. Journal of Personality & Social Psychology, 61, 366–379. Kiechel, W. (1993). How we will work in the year 2000. Fortune, 130, 38–52. Kram, K. E., & Isabella, L. A. (1985). Mentoring alternatives: The role of peer relationships in career development. Academy of Management Journal, 28, 110–132. Kubeck, J. E., Delp, N. D., Haslett, T. K., & McDaniel, M. A. (1996). Does job-related training performance decline with age? Psychology and Aging, 11, 92–107.

Toward Understanding and Managing Stereotypical Beliefs

283

Lambert, A. (1995). Stereotypes and social judgment: The consequences of group variability. Journal of Personality & Social Psychology, 68, 388–403. Lawrence, B. S. (1984). Age grading: The implicit organizational timetable. Journal of Occupational Behavior, 5, 23–35. Lee, J. A., & Clemons, T. (1985). Factors affecting employment decisions about older workers. Journal of Applied Psychology, 70, 785–788. MacNeil, R. D., Ramos, C. I., & Magafas, A. M. (1996). Age stereotyping among college students: A replication and expansion. Educational Gerontology, 22, 229–243. Markus, H., & Herzog, R. (1991). The role of the self-concept in aging. Annual Review of Gerontology and Geriatrics, 11, 110–143. Maurer, T. (2001). Career-relevant learning and development, worker age, and beliefs about self-efficacy for development. Journal of Management, 27, 123–140. Maurer, T., & Rafuse, N. (2001). Learning not litigating: Managing employee development and avoiding claims of age discrimination. Academy of Management Executive, 15(4), 110–121. Maurer, T., & Tarulli, B. (1994). Perceived environment, perceived outcome, and person variables in relationship to voluntary development activity by employees. Journal of Applied Psychology, 79, 3–14. Maurer, T., & Tarulli, B. (1996). Acceptance of peer/upward performance appraisal systems: Role of work context factors and beliefs about managers’ development capability. Human Resource Management Journal, 35, 217–241. Maurer, T., Weiss, M., & Barbeite, F. (2003). A model of involvement in work-related learning and development activity: The effects of individual, situational, motivational and age variables. Journal of Applied Psychology, 8(4). Mercer, W. M. (1981). Employer attitudes: Implications of an aging work force. New York: William M. Mercer. Miller, D. M., Cox, S., Gieson, M., Bean, C., Adams-Price, C., Sanderson, P., & Topping, J. S. (1993). Further development and validation of an age-based equal opportunity measure for organizations: An operational definition of ecological dissonance. Psychology, A Journal of Human Behavior, 30, 32–37. Morse, C. K. (1993). Does variability increase with age? An archival study of cognitive measures. Psychology and Aging, 8, 156–164. Mullen, B., & Hu, L. (1989). Perceptions of ingroup and outgroup variability: A meta-analytic integration. Basic & Applied Social Psychology, 10, 33–252. Netz, Y., & Ben-Sira, D. (1993). Attitudes of young people, adults, and older adults from threegeneration families toward the concepts “ideal person,” “youth,” and “old person.” Educational Gerontology, 19, 607–621. Neuberg, S., & Fiske, S. (1987). Motivational influences on impression formation: Outcome dependency, accuracy-driven attention, and individuating processes. Journal of Personality and Social Psychology, 53, 431–444. Newsham, D. B. (1969). The challenge of change for the adult trainee. Industrial Gerontology, 3, 32–33. Noe, R., & Wilk, S. (1993). Investigation of factors that influence employees’ participation in development activities. Journal of Applied Psychology, 78, 291–302. Nuessel, F. H. (1982). The language of ageism. Gerontologist, 22, 273–276. Nussbaum, J. F., & Robinson, J. D. (1984). Attitudes toward aging. Communication Research Reports, 1, 21–27. Perry, J. S., & Varney, T. L. (1978). College student’s attitudes toward workers’ competence and age. Psychological Reports, 42, 1319–1322.

284

TODD J. MAURER, KIMBERLY A. WRENN AND ELIZABETH M. WEISS

Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis of prejudice. Personality & Social Psychology Bulletin, 5, 461–476. Rosen, B., & Jerdee, T. H. (1976a). The influence of age stereotypes on managerial decisions. Journal of Applied Psychology, 61, 428–432. Rosen, B., & Jerdee, T. H. (1976b). The nature of job-related stereotypes. Journal of Applied Psychology, 61, 180–183. Rosen, B., & Jerdee, T. H. (1977). Too old or not too old? Harvard Business Review, 7, 97–108. Rosen, N., Williams, L., & Foltman, F. (1965). Motivational constraints in an industrial retraining program. Personnel Psychology, 18, 65–79. Ruscher, J., & Fiske, S. (1990). Interpersonal competition can cause individuating processes. Journal of Personality and Social Psychology, 58, 832–843. Ryan, E., & See, S. (1993). Age-based beliefs about memory changes for self and others across adulthood. Journal of Gerontology, 48, 199–201. Salthouse, T. M., & Maurer, T. J. (1996). Aging, job performance, and career development. In: J. E. Birren & K. W. Schaie (Eds), The Handbook of Aging (pp. 353–364). San Diego, CA: Academic Press. Schmidt, D. F., & Boland, S. M. (1986). Structure of perceptions of older adults: Evidence for multiple stereotypes. Psychology and Aging, 1, 255–260. Siemen, J. R. (1976). Programmed material as a training tool for older persons. Industrial Gerontology, 3, 183–190. Simon, R. (1996). Too damn old. Money, 25, 118–126. Slotterback, C. S., & Saarnio, D. A. (1996). Attitudes toward older adults reported by young adults: Variation based on attitudinal task and attribute categories. Psychology and Aging, 11, 563– 571. Society for Human Resource Management/AARP (1998). Older workers survey. Stagner, R. (1985). Aging in industry. In: J. Birren & W. Schaie (Eds), Handbook of Psychology of Aging (2nd ed., pp. 789–817). New York: Van Nostrand Reinhold Co. Taylor, G. S., Crino, M. D., & Rubenfeld, S. (1989). Coworker attributes as potential correlates to the perceptions of older workers’ job performance: An exploratory study. Journal of Business and Psychology, 3, 449–458. Tharenou, P., Latimer, S., & Conroy, D. (1994). How do you make it to the top? An examination of influences on women and men managerial advancement. Academy of Management Journal, 37, 899–931. Tieger, C. (1994). We are all aging workers: For an interdisciplinary approach to aging at work. In: J. Snel & R. Cremer (Eds), Work and Aging: A European Perspective (pp. 65–84). London: Taylor & Francis. Tucker, F. D. (1985). A study of the training needs of older workers: Implications for human resources development planning. Public Personnel Management Journal, 14, 85–95. Verhaeghen, P., & Salthouse, T. A. (1997). Meta-analyses of age-cognition relations in adulthood: Estimates of linear and nonlinear age effects and structural models. Psychological Bulletin, 122, 231–249. Von Hippel, W., Jonides, J., Hilton, J., & Narayan, S. (1993). Inhibitory effect of schematic processing on perceptual encoding. Journal of Personality and Social Psychology, 64, 921–935. Warr, P. B. (1994a). Age and employment. In: H. C. Triandis & M. D. Dunnette (Eds), Handbook of Industrial and Organizational Psychology (pp. 485–550). Palo Alto, CA: Consulting Psychologists Press. Warr, P. (1994b). Training for older managers. Human Resource Management Journal, 4(2), 22–38.

Toward Understanding and Managing Stereotypical Beliefs

285

Warr, P., Allan, C., & Birdi, K. (1999). Predicting three levels of training outcome. Journal of Occupational and Organizational Psychology, 72, 351–375. Weiss, E., & Maurer, T. (in press). Age discrimination in personnel decisions: A reexamination and extension. Journal of Applied Social Psychology. Wilder, D. A. (1978). Reduction of intergroup discrimination through individuation of the out-group. Journal of Personality and Social Psychology, 36, 1361–1374. Zalesny, M., & Ford, J. (1990). Extending the social information processing perspective: New links to attitudes, behaviors, and perceptions. Organizational Behavior and Human Decision Processes, 47, 205–246. Zandri, E., & Charness, N. (1989). Training older and younger adults to use software. Educational Gerontology, 15, 615–631.

CURRENT DIRECTIONS AND ISSUES IN PERSONNEL SELECTION AND CLASSIFICATION
Walter C. Borman, Jerry W. Hedge, Kerri L. Ferstl, Jennifer D. Kaufman, William L. Farmer and Ronald M. Bearden
ABSTRACT
This chapter provides a contemporary view of state-of-the science research and thinking done in the areas of selection and classification. It takes as a starting point the observation that the world of work is undergoing important changes that are likely to result in different occupational and organizational structures. In this context, we review recent research on criteria, especially models of job performance, followed by sections on predictors, including ability, personality, vocational interests, biodata, and situational judgment tests. The paper also discusses person-organization fit models, as alternatives or complements to the traditional person-job fit paradigm.

INTRODUCTION
The world of work is in the midst of profound changes that are likely to result in different occupational and organizational structures in the future. Technological

Research in Personnel and Human Resources Management Research in Personnel and Human Resources Management, Volume 22, 287–355 Copyright © 2003 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0742-7301/doi:10.1016/S0742-7301(03)22007-7

287

288

WALTER C. BORMAN ET AL.

modernization will only accelerate these changes. Changing technologies will alter the nature of work, with workers required to adjust to new equipment and procedures, adapt to ever-changing environments, and continue to enhance their job-related skills. Some have speculated that there will be no stable jobs in the future; instead, work will be organized around specific projects and initiatives that workers will move to and from (Ilgen & Pulakos, 1999; Pearlman, 1995). This means much more fluidity in the way jobs are performed. Flexibility and adaptability will be necessary to deal effectively with these less well-defined roles. As organizations increasingly operate in a wide and varied set of situations, cultures, and environments, workers will likely need to be more versatile and able to handle a wider variety of diverse and complex tasks. In an age of rapid technological changes and globalization, individuals will be required to be in a continual learning mode. Recognizing that fundamentally new ways of thinking and acting will be necessary to meet these dramatic changes, one critically important intervention, and the focus of this paper, is personnel selection and classification. Historically, the preponderance of predictor validation work has centered on forecasting performance in training, early in a career, or the tenure of employees has been variable (as in concurrent validation studies). However, because finding and training workers in the future will be much more complex and costly than it is today, success on the job during and beyond the first few years will be increasingly important. The prediction of such long-term success indicators as retention and long-term performance will require the use of more complex sets of predictor variables that include such measures as personality, motivation, and vocational interests. To effectively develop measures to predict long-term performance, it will be crucial to better understand the context for the workplace of the future, including the environmental, social, and group structural characteristics. Ultimately, combining the personal and organizational characteristics should lead to improved personnel selection models that go beyond the usual person-job relations, encouraging a closer look at theories of person-organization (P-O) fit (see Borman, Hanson & Hedge, 1997). As we learn more about the organizational needs of the future, we must develop measures of aptitude, knowledge, skill, and personality to address these changes. We can map these requirements onto current selection instruments, and see what is being measured, what is not being measured, and what is being measured poorly. This knowledge would then allow us to focus research efforts on developing new measures to select and possibly classify applicants into future jobs. Expansion of both the predictor and criterion domains will allow organizations to select applicants with a greater chance of long-term success. What follows is a detailed explication of the state-of-the-science relevant to the changing nature of jobs, and

Current Directions and Issues in Personnel Selection and Classification

289

what this means for the future of workplace selection and classification. The next section describes recent research on criteria, with subsequent sections focusing on predictor measurement, in particular, ability, personality, vocational interests, biodata, and situational judgment tests. The paper concludes with a section on an alternative selection model to the traditional person-job fit strategy, personorganization fit.

CRITERION RESEARCH
The central dependent variable of concern in industrial and organizational psychology is job performance. Performance criteria are often what we attempt to predict from our major interventions, including personnel selection, training, and job design. Traditionally, in our field, by far the most attention has been focused on models related to predictors. For example, models of cognitive ability (e.g. Carroll, 1993; Guilford, 1967; Spearman, 1927), personality (e.g. Hogan, 1982; Wiggins, 1985), and vocational interests (e.g. Holland, 1973) have been developed and discussed extensively in the literature.

Models of Job Performance More recently, job performance models and research associated with them are beginning to foster more scientific understanding of criteria. These models have at least two forms. One type examines relations between elements of performance (e.g. job knowledge and proficiency) directed at learning more about criteria. A second type attempts to explicate the central latent variables that can best characterize all performance requirements in work. The first type of performance model emerged in Hunter (1983). His path analysis results suggested that cognitive ability primarily has a direct effect on individuals’ acquisition of job knowledge. Job knowledge in turn influences technical proficiency. Supervisory performance ratings were a function of both job knowledge and technical proficiency, with the job knowledge-ratings path coefficient three times as large as the technical proficiency-ratings coefficient. This line of research continued, with additional variables being added to the models. Schmidt, Hunter and Outerbridge (1986) added job experience to the mix; they found that job experience had a direct effect on the acquisition of job knowledge and an indirect effect on task proficiency through job knowledge. A series of studies by Borman and colleagues (Borman, Hanson, Oppler, Pulakos & White, 1993; Borman, White & Dorsey, 1995; Borman, White,

290

WALTER C. BORMAN ET AL.

Pulakos & Oppler, 1991) added additional variables to the models, and changed somewhat the conceptual focus of these models. Borman et al. (1991) included two personality variables, achievement and dependability, and behavioral indicators of achievement (number of awards and commendations) and dependability (number of disciplinary infractions) in a sample of first tour U.S. Army personnel. The path model results showed that the personality variables had indirect effects on the supervisory performance ratings through their respective behavioral indicators. The best fitting model also had paths from ability to acquisition of job knowledge, job knowledge to technical proficiency, and technical proficiency to the supervisory job performance ratings, arguably the most comprehensive measure of overall performance. Perhaps the most important result of this study was that the variance accounted for in the performance rating exogenous (dependent) variable increased substantially with the addition of personality and the behavioral indicators of personality beyond that found with previous models including only ability along with job knowledge and technical proficiency. Also important, results were interpreted as demonstrating that overall job performance ratings were influenced primarily by ratees’ technical proficiency and behaviors relevant to personality. The Borman et al. (1993) model essentially extended the Schmidt et al. (1986) model to a supervisor sample. They found that contrary to the Schmidt et al. findings, the ability-job experience path was significant. The authors interpreted this path substantively as showing that high ability soldiers are more likely to be given opportunities to supervise others, thus leading to this significant path coefficient. The 1995 study (Borman et al., 1995) had the same basic design as the 1991 study, but included several additional ratee personal characteristic and rater-ratee relationship variables beyond ability, job knowledge, technical proficiency, and supervisory performance ratings. Results from this first tour U.S. Army sample showed that ratee dependability and technical proficiency related highest with the supervisory performance ratings. In a peer rating model, the same two variables plus ratee obnoxiousness (in a negative direction) had significant and approximately equal path coefficients to the overall performance ratings. Again, the interpretation of both sets of results was that ratee dependability and ratee technical proficiency are primary cues supervisors use to make overall judgments about subordinate job performance; peers use the same two cues or factors for these overall performance judgments, but also weight ratee obnoxiousness almost as highly. Regarding the second type of model, Campbell, McCloy, Oppler and Sager (1993) posited eight latent criterion factors that summarize the performance requirements for all jobs. These constructs are: (1) job specific task proficiency; (2) non-job specific task proficiency; (3) written and oral communication;

Current Directions and Issues in Personnel Selection and Classification

291

(4) demonstrating effort; (5) maintaining personal discipline; (6) facilitating peer and team performance; (7) supervision/leadership; and (8) management/administration. Several of these latent constructs emerged consistently in the Project A research, a large-scale selection and classification study conducted in the U.S. Army (Campbell, 1990), across the several jobs studied in that program. Accordingly, there is some impressive empirical support for at least part of this performance taxonomy. The importance of this research direction is considerable. The criterion domain should be carefully mapped, just as various predictor areas have already been. This specification of criterion content should importantly shape the way selection research gets done (Campbell et al., 1993). Consistent with one theme of this review, a criterion taxonomy will help to organize accumulating research findings by addressing questions about links between predictors and individual criteria rather than predictors and overall performance. Research by McCloy, Campbell and Cudek (1994) supports this view of predictor-criterion relations. These authors divided the criterion space into declarative knowledge, procedural knowledge and skill, and motivational components. Using data from eight jobs in Project A, they demonstrated that declarative knowledge is predicted primarily by cognitive ability, whereas the motivational element of performance (indexed by ratings) is linked to personality. Also, Pulakos, Borman and Hough (1988) found very different patterns of personality predictor-criterion relations across three different performance constructs in a sample of Navy recruiters.

Task and Citizenship Performance Another useful way to divide the job performance domain has been according to task and citizenship performance (Borman & Motowidlo, 1993). Borman and Motowidlo argue that organization members may engage in activities that are not directly related to their main task functions but nonetheless are important for organizational effectiveness because they support the “organizational, social, and psychological context that serves as the critical catalyst for task activities and processes” (p. 71). These authors initially presented a five dimension taxonomy that they believed captured the contextual or citizenship performance domain. More recently, Borman, Motowidlo and colleagues have settled on a three-dimension system: (1) personal support; (2) organizational support; and (3) conscientious initiative (Borman, Buck, Hanson, Motowidlo, Stark & Drasgow, 2001; Coleman & Borman, 2000). The notion is to characterize the citizenship performance construct according to the recipient or target of the behavior – other persons, the organization, and oneself, respectively.

292

WALTER C. BORMAN ET AL.

It should be noted that citizenship performance is related to several previously derived concepts. Borman and Motowidlo (1993) point out that important forerunners to this construct are organizational citizenship behavior (Organ, 1988; Smith, Organ & Near, 1983), prosocial organizational behavior (Brief & Motowidlo, 1986), and a model of soldier effectiveness (Borman, Motowidlo & Hanser, 1983). Additional research directions have emerged based on the distinctions among criterion constructs. First, building on the task-citizenship performance distinction, research has examined the weights especially supervisor raters place on ratee task and citizenship performance when making overall performance or similar kinds of global worth-to-the-organization judgments. This research, using various types of methodologies, has consistently found that task and citizenship performance are weighted roughly the same when raters make these overall effectiveness judgments. Podsakoff, MacKenzie, Paine and Bachrach (2000) recently summarized these studies. Across eight studies (e.g. Allen & Rush, 1998; Borman et al., 1995; Van Scotter & Motowidlo, 1996), they found an average of 9.3% of the variance in overall performance ratings was accounted for by task performance; for citizenship performance, the average was 12.0%. Further, Conway (1999) conducted a meta-analysis targeted toward the same issue, and found that collapsing across supervisor and peer raters the respective percentage of variance accounted for in overall performance ratings by task and citizenship performance was 11.5 and 12.0%. The task-citizenship performance distinction has also led to the discovery of different patterns of individual differences predictors being linked to the two different performance domains. For example, the Project A validation results (Campbell & Knapp, 2001) show that ability best predicts performance on the technical proficiency criteria, and personality, especially dependability, best predicts performance on the citizenship dimension, personal discipline. Motowidlo and Van Scotter (1994) and Van Scotter and Motowidlo (1996) found that personality predictors generally correlated higher with citizenship performance than with task performance. More recently, Borman, Penner, Allen and Motowidlo (2001) reviewed studies conducted since 1995 on personality predictors of citizenship performance. They concluded that several personality constructs have at least modest correlations with citizenship performance and that personality-citizenship performance linkages are stronger than personality-task performance links. Hurtz and Donovan (2000) arrived at a similar conclusion based on their meta-analysis, although differences between the personality-citizenship and personality-task performance relations were not large.

Current Directions and Issues in Personnel Selection and Classification

293

Adaptability A potentially important criterion domain that may not fit neatly into the taskcitizenship taxonomy is adaptability. Adaptive job performance is characterized by the ability and willingness to cope with uncertain, new, and rapidly changing conditions on the job. Adaptive performance has become increasingly important in today’s workplace. As mentioned previously, today’s technology is changing at an unprecedented rate. To adapt, personnel must adjust to new equipment and procedures, function in changing environments, and continuously learn new skills. Technological change often requires the organization of work around projects rather than well-defined and stable jobs (Ilgen & Pulakos, 1999; London & Mone, 1999), which in turn requires workers who are sufficiently flexible to be effective in poorly defined roles. Employees must also be able to adapt to increasing diversity: women and nonwhites are moving in growing numbers into roles traditionally held only by white men (Edwards & Morrison, 1994). Furthermore, workers may be assigned to a variety of locations throughout the world during their careers. They must be able to adapt to unfamiliar situations, cultural values, and perspectives. Thus, selection and classification practices may have to be updated in accordance with the changing nature of job performance. If job performance has expanded to include adaptability requirements, it is important to consider how these fit into existing job performance models. Some suggest that adaptive performance should be included as a broad category of job performance, on the same order as task performance and citizenship performance (Hesketh & Neal, 1999; Pulakos, Arad, Donovan & Plamondon, 2000). For example, Pulakos and her colleagues (Pulakos et al., 2000), based on analysis of the literature and a large number of critical performance incidents collected from a group of 21 diverse jobs, identified eight dimensions of adaptive job performance: (1) handling emergencies or crisis situations; (2) handling work stress; (3) solving problems creatively; (4) dealing with uncertain and unpredictable work situations; (5) learning work tasks, technologies, and procedures; (6) demonstrating interpersonal adaptability; (7) demonstrating cultural adaptability; and (8) demonstrating physically oriented adaptability. Others assert that dimensions of adaptive performance can be classified as being most strongly tied to either task performance (e.g. learning work tasks, technologies, and procedures) or citizenship performance (e.g. demonstrating cultural adaptability) (Mumford, Meiman, Russell & Crafts, 1998). This approach can be difficult because the distinction between task and citizenship performance can depend on the nature of the job in question (Borman & Motowidlo, 1993).

294

WALTER C. BORMAN ET AL.

For example, depending upon the job and setting, handling work stress might deal primarily with technical job elements, citizenship elements, or a mix of the two. Others, recognizing that the two approaches are not mutually exclusive, take an integrated perspective. Johnson (in press) subsumes under task and citizenship performance those adaptability dimensions that are clearly related to one or the other of those domains. In sum, adaptive performance appears to be an important additional construct in the criterion domain. It may be possible to categorize elements of adaptability into task or citizenship performance, as appropriate, or the entire adaptability construct may be considered separate from task or citizenship performance. Whichever approach seems more appropriate, adaptability appears to be an important and salient criterion construct.

Work Commitment Another highly important criterion construct is work commitment. Work commitment has been a topic of interest for many years, with theory and research continuing to evolve at a steady pace over the last decade. Some of the earliest, influential work examined employees’ commitment to their employers, thus the early focus addressed “organizational” commitment as a unidimensional construct. Today, however, it is widely recognized that commitment can take different forms, and is characterized as a multidimensional work attitude. In this section, we will briefly highlight principal research findings, with emphasis on recent meta-analytic results; discuss the emerging viewpoint of commitment as a multi-faceted construct; and note relevant future directions. Organizational Commitment Porter, Steers, Mowday and Boulian (1974) defined organizational commitment as “the strength of an individual’s identification with and involvement in a particular organization” (p. 604). They suggested that the construct was characterized by three factors: (1) belief in and acceptance of the organization’s values and goals; (2) willingness to exert effort on behalf of the organization; and (3) desire to maintain organizational membership. This affective or emotional attachment perspective toward organizational commitment dominated the literature during the 1970s and 1980s. In fact, when Mathieu and Zajac (1990) conducted an exhaustive metaanalytic review of the organizational commitment literature, more than 75% of the studies that were included operationalized organizational commitment in this way. Mathieu and Zajac (1990) examined organizational commitment’s relation to 26 antecedents, 14 correlates, and eight consequences. Strongest correlations

Current Directions and Issues in Personnel Selection and Classification

295

within the antecedent category included perceived competence, work ethic, job scope, leader communication, participatory leadership, leader consideration, leader initiating structure, and role conflict. The most noteworthy relationships within the correlates category included overall motivation, internal motivation, job involvement, stress, occupational commitment, union commitment, overall job satisfaction, satisfaction with supervision, and satisfaction with the work itself. Notable correlations within the consequences category included intention to search for job alternatives, and intention to leave one’s job. In the mid-1980s and early 1990s, Meyer and Allen (e.g. Allen & Meyer, 1990; Meyer & Allen, 1984) sought to broaden the nature of organizational commitment. They identified three common themes that they labeled affective, continuance, and normative commitment, and argued that all three forms of commitment reflect an employee’s relationship to an organization. They suggested that affective organizational commitment reflects an emotional attachment to the organization; continuance commitment involves perceived cost associated with discontinuing employment; and normative commitment is characterized by feelings of moral obligation to remain with the organization. Generally, the perspective was that employees with strong affective occupational commitment remain in their occupation because they want to, those with strong normative occupational commitment remain because they feel they ought to, and those with strong continuance commitment remain out of necessity. Meyer and Allen developed item sets to test the three-component model of commitment and, as noted by Meyer (1997) each of the three scales have been subjected to fairly extensive psychometric evaluation, and have received considerable support. Nonetheless, further refinements may be necessary, as several studies have questioned whether affective and normative commitment are truly distinguishable forms of commitment (e.g. Hackett, Bycio & Hausdorf, 1994), and whether continuance commitment is a unidimensional construct, or is in fact composed of two subscales (e.g. McGee & Ford, 1987). In a recent study, Meyer, Allen and Topolnytsky (1998) summarized research findings pertaining to the three facets of commitment. They suggested that affective commitment appears to be strengthened by work experiences that contribute to employees’ comfort in the organization (e.g. good interpersonal relations; role clarity), as well as their sense of competence and self-worth (e.g. participation, feedback, challenge). Increases in continuance commitment result from actions or decisions (in or outside of the workplace), that link the retention of valued assets (e.g. company benefits; status in the community) to continued employment in the organization. Normative commitment appears to be influenced by family, culture, or organization socialization experiences that underscore the appropriateness of continued service, or by the receipt of benefits from the organization

296

WALTER C. BORMAN ET AL.

(e.g. investments in education or training) that create a sense of obligation or reciprocity in the employee. Stanley, Meyer, Topolnytsky and Herscovitch (1999) conducted a series of meta-analyses to examine the relationship between Meyer and Allen’s organizational commitment scales and turnover, turnover intention, absenteeism, job performance, and organizational citizenship behavior. They found that in all cases, affective commitment correlated most strongly with these outcome variables, followed by normative, and then continuance commitment. The magnitude of these correlations with affective commitment was most notable for turnover intention (−0.43), organizational citizenship behavior (0.29), turnover (−0.23), and job performance (0.16). Occupational Commitment Carson and Bedeian (1994) have suggested that coping with the uncertainty associated with changes such as mergers, acquisitions, and layoffs has caused many employees to intensify their focus on, and commitment to, the aspect of their working life that they can control – their occupation. Irving, Coleman and Cooper (1997) reinforced this notion by suggesting that increased emphases on outsourcing and downsizing, as well as changes in organizational structures such as virtual organizations, may have led to a shift in organizations’ relations with their employees – and as a result, individuals may focus their attachment more on their occupations than the organizations for which they work. Meyer, Allen and Smith (1993) tested the generalizability of the Meyer and Allen three-component model of organizational commitment to the domain of occupational commitment. Using a sample of student nurses and registered nurses, they found that: (1) reliable measures of affective, continuance, and normative occupational commitment could be developed; (2) the three components of occupational commitment are differentially related to variables considered to be antecedents or consequences of commitment; and (3) organizational and occupational commitment contribute independently to the prediction of important organizationally-relevant outcome variables such as turnover intention, performance, and citizenship. In a recent meta-analysis of the occupational commitment literature, Lee, Carswell and Allen (2000) reported correlations between occupational commitment and work-related attitudes, work experiences, and outcome variables. With work-related attitudes they found occupational commitment correlated most strongly with job involvement, overall job satisfaction, satisfaction with the work itself, endorsement of work ethic, and career satisfaction. The magnitude of the correlations between occupational commitment and work experiences was not as great as was found with attitudes, but was most noteworthy with stress, role ambiguity, role conflict, supervisor support, coworker

Current Directions and Issues in Personnel Selection and Classification

297

support, and autonomy. Also, occupational and organizational commitment were relatively strongly inter-correlated (0.45), a finding quite similar to the correlation reported by Mathieu and Zajac (1990). The authors also found that the patterns of the correlations were similar for the professional and non-professional categories, although they pointed out that this does not necessarily mean that the mechanisms through which organizational commitment and occupational commitment become correlated are the same for the two groups. The authors speculated, for example, that for professionals, the occupational socialization process likely begins prior to any organizational socialization, whereas they may develop concurrently for those in the non-professional category. In terms of occupational commitment and outcome variables, occupational commitment was most strongly related to occupational turnover intention, organizational turnover intent, organizational turnover, and supervisory ratings of performance. The moderator analyses found some differences between professional and non-professional occupations, however. For professional jobs, the occupational commitment-occupational turnover intention correlation was −0.62, but only −0.50 for non-professionals; and −0.29 vs. −0.15 for organizational turnover intent. Thus, it appears that for professional employees, identification with and attachment to their profession is a particularly important factor in making the decision to leave the profession or the organization. Multiple Perspectives of Work Commitment Although the bulk of research and theorizing has focused on organizational commitment, and more recently occupational commitment, over the years comments have been made about the importance of viewing work commitment even more broadly as a multi-faceted concept. For example, Mowday, Porter and Steers (1982) suggested that researchers attend to the potential influence of multiple commitments, as well as the potential for conflict that multiple commitments may generate. Reichers (1985, 1986) argued that in reality, organizations comprise various coalitions and constituencies (e.g. owners, managers, rank-and-file, customers) whose goals and values may or may not be compatible with those of the organization. Becker (1992) developed commitment indices to measure commitment at a number of levels within the organization, and demonstrated that employees’ commitments to top management, immediate supervisor, and work group accounted for variance in job satisfaction, intention to quit, and prosocial organizational commitment beyond what was accounted for by a unidimensional measure of organizational commitment. The notion here is that employees develop commitments not only to the organization, but also to various constituencies within or outside the organization. In fact, there may be certain circumstances that dictate encouraging commitment at sub-organization levels. For example, as noted by Meyer et al. (1998), under

298

WALTER C. BORMAN ET AL.

uncertain economic conditions, an organization may not be able to guarantee employment even to its most loyal and effective workers, and thus it might be difficult to get employees to commit to work toward organizational goals for the sake of the organization. However, it might be possible to get employees to work toward those same goals if they can be shown to be relevant to a different target. For example, an employee who is highly committed to his/her own personal career is more likely to work hard on projects seen as instrumental for the development of marketable skills. Some of the changes employees face will cause them to look beyond the organization and more carefully consider the nature and limits of organizational attachment. Although the multiple-constituency framework has not been tested extensively, preliminary evidence indicates that there may be some value in measuring commitments to more specific foci within and outside the organization. Several research studies have explored this line of thinking. For example, Gordon, Philpot, Burt, Thompson and Spiller (1980) developed and tested a model of employee commitment to the union. More recently, Lok and Crawford (2001) investigated the relative importance of the organizational culture and subculture of the employee’s local work environment as predictors of employee commitment, and found that subculture had a greater influence on commitment than did the broader organizational culture. The authors suggested that greater attention and resources might need to be given to influencing subcultures in organizations. Allen and Grisaffe (2001) focused on the relations between organizational commitment and customer reactions, reviewing relevant theory and research, discussing methodological issues associated with examining this issue, and making recommendations for both researchers and HR practitioners. McElroy, Morrow and Laczniak (2001) focused on what they labeled external organizational commitment (EOC), that reflects employees involvement and identification with another organization (supplier, customer, partner organization), external to the one in which they are employed. Gallagher and McLean Parks (2001) focused attention on the need for more research to address an increasingly large group of workers – contingent workers. This group is interesting because there is no explicit or implicit expectation of long-term or “ongoing” employment. Gallagher et al. suggested that not only must research focus on determining the level and interrelationships among the facets of commitment within the context of different work arrangements, but also researchers must examine the generalizability of existing commitment measures to the contingent workforce. Changing Perspectives on Work Commitment As mentioned, organizations worldwide are undergoing unprecedented change, with shifts in workforce demographics, technological innovations, and global

Current Directions and Issues in Personnel Selection and Classification

299

competition. Responding to a rapidly changing environment can sometimes place conflicting demands on organizations. On the one hand, organizations need to be flexible in order to be able to adapt to changes as they occur. Thus, they may be required to shrink and expand their workforce as needed – or, they may need a workforce that is flexible with the skills needed to do whatever is required to get the job done. The requirements for flexibility in employment and in employees may not be compatible. As organizations consider the future for selection and retaining their workforce, work commitment may play a vital role in ensuring both near-term and long-term success. To provide some perspective for our review of the work commitment research to date, at least three points are relevant. First, it is becoming increasingly evident that researchers should attend closely to multiple levels and facets of commitment, understanding that they may complement or conflict with the broader goals of organizational commitment. Second, changes in the workplace may affect different components of commitment differently. For example, Meyer et al. (1998) hypothesized that affective commitment is likely to be influenced by changes in comfort – and competence – related work experiences. In addition, they suggested that continuance commitment might be affected by organizational change that draws attention to the costs and benefits of continued employment in the organization. They also suggested that normative commitment might change as a function of altered perceptions of the organization’s willingness to invest in its employees. Consequently, it is possible that the commitment profiles of organizational personnel following organizational change will be much different than they were before the change. Finally, Beck and Wilson (2001) presented the case for viewing work commitment as a developmental process. The authors noted that most of the empirical work reported in the commitment literature has been designed to examine static levels of commitment rather than assessing changes in commitment over time. While noting that little research has been done using a developmental focus, they suggested that researchers should begin to study critical times in individuals’ tenure with an organization when levels of commitment may be volatile, and begin to search for the organizational and individual variables that are causing changes.

Conclusions In summary, criteria have always been important in personnel selection research (e.g. Smith, 1976; Toops, 1944; Wallace, 1965; Weitz, 1961), but a recent trend has been to study job performance in its own right in an attempt to develop substantive models of performance. The vision has been to learn more about the nature of

300

WALTER C. BORMAN ET AL.

job performance, including its components and dimensions, so that performance itself, as well as predictor-performance links, will be better understood (Borman, Hanson & Hedge, 1997; Campbell et al., 1993, 1996; Campbell & Knapp, 2001). Again, if we can make more progress in this direction, the cumulative evidence for individual predictor construct-performance construct relationships will continue to progress and the science of personnel selection will be substantially enhanced. It will also be important to attend to related constructs such as adaptability and work commitment because they are becoming increasingly critical for the modern workforce. Ideally, selection and classification predictors of the future will forecast not only employees’ early career job performance, but also the additional criterion constructs discussed in this section.

ABILITY
Definitional Issues Abilities are relatively stable individual differences that are related to performance on some set of tasks, problems, or other goal oriented activities (Murphy, 1996). Another definition, offered by Carroll (1993), conceptualizes abilities as relatively enduring attributes of an individual’s capability for performing a particular range of tasks. Abilities are similar to traits in that they exhibit some stability over time, yet they are different in the sense that they may develop slowly with exposure to multiple situations (Snow & Lohman, 1984). Despite various definitions of ability, ultimately every ability is defined in terms of some kind of performance, or potential for performance (Carroll, 1993). Although the term ability is widely used in both the academic and applied literature, several other terms have been related loosely to abilities. For example, the term competency has been used to describe individual attributes associated with the quality of work performance. Specifically, competencies have been defined as underlying characteristics of individuals causally related to effective performance in a job (Boyatzis, 1982). This definition of competency is clearly redundant with the definition of ability offered above. However, in practice, lists of competencies often include a mixture of knowledge, skills, abilities, motivation, beliefs, values, and interests (Fleishman, Constanza & Marshall-Mies, 1999). Another term that is often confused in the literature with abilities is skills. Whereas abilities are general traits inferred from relationships among performances of individuals observed across a range of tasks, skills are more dependent on learning and represent the product of training on particular tasks (Fleishman et al., 1999). In general, skills are more situational, but the development of a skill is,

Current Directions and Issues in Personnel Selection and Classification

301

to a large extent, predicted by an individual’s possession of relevant underlying abilities, usually mediated by the acquisition of the requisite knowledge (e.g. Borman, White, Pulakos & Oppler, 1991; Schmidt, Hunter & Outerbridge, 1986). That is, these underlying abilities are related to the rate of acquisition and final levels of performance that a person can achieve in particular skill areas (Fleishman et al., 1999). Ability tests usually measure mental or cognitive ability, but may also measure other constructs such as physical abilities. Ability tests have been almost always paper and pencil tests administered to applicants in a standardized manner, although recent advances in computerized testing have led to more ability tests being administered by computer (e.g. Drasgow & Olson-Buchanon, 1999). The following sections discuss recent developments and the most current topics in the areas of cognitive ability, tacit knowledge or practical intelligence, and physical ability testing.

Cognitive Ability The emergence of industrial and organizational psychology is inextricably linked with cognitive ability testing (Landy, Shankster & Kohler, 1994). Notably, in the early 1900s, Cattell, Scott, Bingham, Viteles and other applied psychologists used cognitive ability tests to lay the foundation for the current practice of personnel selection (Landy, 1993). Recent attention has focused on the usefulness of general cognitive ability “g” versus more specific cognitive abilities for predicting training and job performance, and the contributions of information processing models of cognitive abilities for learning more about ability constructs. First, there has been a debate concerning the “ubiquitousness” of the role of general cognitive ability or g in the prediction of training and job performance. Several recent studies have demonstrated that psychometric g, generally operationalized as the common variance in a battery of cognitive ability tests (e.g. the first principal component), accounts for the majority of the predictive power in the test battery, and that the remaining variance (often referred to in this research as “specific abilities”) accounts for little or no additional variance in the criterion (e.g. Larson & Wolfe, 1995; Ree, Earles & Teachout, 1994). This line of research has also suggested that the specific ability components account for somewhat more variance in job performance than they do in training performance (e.g. Olea & Ree, 1994; Ree et al., 1994). Although there is a growing consensus concerning the predictive efficiency of g, there is less agreement about what this means, especially as it relates to personnel selection practices and research. Some would conclude that “refinements in the measurement of abilities and aptitudes are unlikely to contribute nontrivial increments to validity beyond

302

WALTER C. BORMAN ET AL.

that which is produced by good measures of general mental ability” (Schmidt, 1994, p. 348). Others suggest a variety of scientific and pragmatic reasons why the measurement of specific abilities may still be important and potentially useful. On the scientific side, possibly the most important cautionary note is that our understanding of the basic cognitive processes that underlie intelligent behavior and the reasons why some people are more capable than others is still quite limited (e.g. Murphy, 1996). In addition, an understanding of the processes by which abilities affect performance and the latent structure of the ability-performance relationships is needed to advance the science of personnel selection (e.g. Bobko, 1994; Campbell, 1990). Specific abilities are likely to be more useful (in addition to or in place of g) when our goal is understanding, rather than predictive efficiency (e.g. Alderton & Larson, 1994; Murphy, 1996). However, models put forth to explain ability-performance links rarely examine the roles of specific abilities. Nonetheless, several recent studies have shown meaningful differences in relationships between specific abilities and certain criteria. Carretta and Ree (1995) found that different subtests of the Air Force Officer Qualifying Test (AFOQT) were most valid during different phases of pilot training, and for different training criterion measures. Silva and White (1993) found that a language aptitude battery had significant incremental validity beyond general cognitive ability for predicting success in a language training course. Dror, Kosslyn and Waag (1993) examined differences in spatial ability between pilots and non-pilots, and found that pilots were superior in very specific aspects of spatial ability (e.g. mental rotation) rather than in general spatial ability. In addition to the potential for promoting understanding, improved measures of specific abilities and expanded test batteries have also been found to improve the measurement of g (e.g. Larson & Wolfe, 1995) which, in turn, may improve validities. Cognitive psychologists have been studying the nature of intelligence for years, and although these models have not had many applications in personnel-related contexts, this sort of theory-based approach clearly has potential for improving our understanding of cognitive ability and of ability-performance relationships (although some are less optimistic, e.g. Lohman, 1993). The Air Force conducted a programmatic research effort to determine whether an ability test battery based on a “consensus information processing model” can predict training and job performance better than the ASVAB (Kyllonen, 1994). Preliminary results were promising. The internal structure of the test battery is consistent with the information processing model, it has been shown to predict laboratory learning tasks better than the ASVAB, and at least one of these elementary cognitive tasks (working memory) is strongly related to general cognitive ability. Cognitive psychologists have also identified what appear to be “new” abilities (e.g. time sharing, dynamic spatial ability) which could eventually find selection

Current Directions and Issues in Personnel Selection and Classification

303

applications for jobs that require these special abilities (e.g. Jackson, Vernon & Jackson, 1993). One potential concern is that practice and coaching may be more of a problem for the relatively novel tasks included on these tests when they are used for selection purposes. Another concern is related to the statistical model often used to define g. A general factor or g represents the correlations between specific ability tests, so specific abilities will, by definition, be correlated with the general factor. Thus, it could be argued that it is just as valid to enter specific abilities first and then say that g doesn’t contribute beyond the prediction found with specific abilities alone (e.g. Murphy, 1996). In fact, Muchinsky (1993) found this to be the case for a sample of manufacturing jobs, where mechanical ability was the single best predictor of performance, and an intelligence test had no incremental validity beyond the mechanical test alone. We know very little about specific abilities when they are defined as the variance remaining once a general factor is extracted statistically. Interestingly, what little information is available suggests that these “specific ability” components tend to be most strongly related to cognitive ability tests that have a large knowledge component (e.g. aviation information; Olea & Ree, 1994). This is consistent with previous research showing that job knowledge tests tend to be slightly more valid than ability tests (Hunter & Hunter, 1984), and also with research demonstrating that job knowledge appears to mediate the relationship between abilities and job performance (e.g. Borman, Hanson, Oppler, Pulakos & White, 1993). Meta-analysis has demonstrated the generality of job knowledge tests as predictors of job performance (Dye, Reck & Murphy, 1993). In addition, these authors found that the validity of job knowledge tests was moderated by job complexity and by job-test similarity, with validities significantly higher for studies involving high complexity jobs and those with high job-test similarity. The average corrected validity for job knowledge tests with high job-test similarity was 0.62 in predicting job performance and 0.76 for training performance, and this is somewhat higher than the average corrected validity typically found for cognitive ability tests (e.g. 0.53; Hunter & Hunter, 1984). A more recent meta-analysis showed that different methods and combination of methods have very different validities for predicting future job performance (Schmidt & Hunter, 1998). Furthermore, these recent meta-analytic findings suggest that general cognitive ability measures and work sample measures have high validity (Schmidt & Hunter, 1998). Research findings cumulated over eighty-five years suggest that a general cognitive ability test plus a structured interview may yield validities as high as 0.63 (Schmidt & Hunter, 1998). Finally, conclusions concerning the relative efficiency of g versus specific abilities have been very different when the purpose is classification rather than

304

WALTER C. BORMAN ET AL.

simply selection decisions. Zeidner and Johnson (1994) have shown that when there is a single applicant pool and multiple jobs or job families, specific abilities substantially improve classification efficiency, beyond the use of general cognitive ability alone. Using somewhat different procedures, other researchers have confirmed and extended these findings (Abrahams, Alf, Keickhaefer, Pass, Cole & Walton-Paxton, 1994; Rosse, Peterson & Whetzel, 1995). The amount of gain possible from classification versus selection is negatively related to the intercorrelations among the specific ability tests or composites of tests and positively related to the number of jobs or job families into which applicants are to be classified (see Alf & Abrahams, 1996). However, substantial gains have been found even with predictors that are quite highly correlated. Notice that these results are compatible with the notion that g provides maximum efficiency when prediction is the goal. Scholarios, Johnson and Zeidner (1994) provide some preliminary information about differences in the context of tests (e.g. measures of various abilities and other individual differences) that would be selected for a test battery when classification versus selection is the goal. However, in order to take full advantage of these results, a better understanding of the content of classification batteries and the reasons for these results will be critical (Bobko, 1994). It is interesting to note that these findings are also relevant for vocational/occupational counselors and others concerned with placement (e.g. in the new O∗ Net: Peterson, Mumford, Borman, Jeanneret & Fleishman, 1999), because their decisions are similar to classification decisions in this context.

Tacit or Practical Intelligence Sternberg and colleagues have attempted to broaden the discussion of general intelligence (cf. Sternberg & Wagner, 1992). Based on a triarchic theory of intelligence, Sternberg (1985) suggests that practical intelligence and tacit knowledge play a role in job success. Practical intelligence is often described as the ability to effectively respond to practical problems or demands in situations that people commonly encounter in their jobs, (Sternberg, Wagner & Okagaki, 1993; Wagner & Sternberg, 1985). Conceptually, practical intelligence is distinct from cognitive ability. In fact, some research (cf. Chan, 2001) has found measures of practical intelligence to be uncorrelated with traditional measures of cognitive ability. Sternberg and his colleagues have repeatedly found significant correlations and some incremental validity (over general intelligence) for measures of tacit knowledge in predicting job performance or success (Sternberg, Wagner, Williams & Horvath, 1995). Tacit knowledge has been shown to be trainable, and to differ

Current Directions and Issues in Personnel Selection and Classification

305

in level according to relevant expertise. Certainly tacit knowledge measures deal with content that is quite different from that found in traditional job knowledge tests (e.g. knowledge related to managing oneself and others). When viewed in this way, tacit knowledge measures are very similar to situational judgment tests (SJTs), described in more detail in a subsequent section.

Physical Abilities Another important area for selection into many jobs that require manual labor or other physical demands is the use of physical ability tests. Most physical ability tests are performance tests (i.e. not paper and pencil) that involve demonstration of attributes such as strength, cardiovascular fitness, or coordination. Although physical ability tests are reported to be used widely for selection (Hogan & Quigley, 1994), not much new information has been published in this area during the past few years. In one study, Blakley, Quinones, Crawford and Jago (1994) provided evidence that isometric strength tests are valid predictors across a variety of different physically demanding jobs, and also that the prediction of work simulation performance was better than the prediction of supervisory ratings of physical ability. Blakley et al. (1994) also found, in a large applicant sample, that females scored substantially lower than males on these isometric strength tests. In light of these findings, there is a recent and growing interest in reducing adverse impact through pretest preparation. Hogan and Quigley (1994) demonstrated that participation in a physical training program can improve females’ upper body strength and muscular endurance, and that participation in a pretest physical training program was significantly related to the likelihood of passing a firefighter physical ability test.

Conclusions A cumulation of 85 years of research demonstrates that if we want to hire people without previous experience in a job, the most valid predictor of future performance is general cognitive ability (Schmidt & Hunter, 1998). General cognitive ability measures have many advantages in personnel selection: (1) they show the highest validity for predicting training and job performance; (2) they may be used for all jobs from entry level to advanced; and (3) they are relatively inexpensive to administer. In addition, there is some evidence that measures of “practical intelligence” or tacit knowledge may under certain conditions provide incremental validity beyond general cognitive ability for predicting job

306

WALTER C. BORMAN ET AL.

performance (Sternberg, Wagner, Williams & Horvath, 1995). Finally, physical ability tests may be useful in predicting performance for jobs that are physically demanding.

PERSONALITY
I/O psychology’s long history of interest in personality stems from the desire to predict the motivational aspects of work behavior. Nevertheless, until recently, the prevalent view was that personality variables were a dead end for predicting job performance. Some of the factors fueling this belief were: (1) the view that a person’s behavior is not consistent across situations and thus, traits do not exist; (2) literature reviews concluding that personality variables lack predictive validity in selection contexts; and (3) concern about dishonest responding on personality inventories. However, by the late 1980s, favorable opinions about personality regarding personnel selection began to grow (Hogan, 1991). Evidence accumulated to refute the notion that traits are not real (Kenrick & Funder, 1981) or stable (Conley, 1984). Research showed at least modest validity for some personality traits in predicting job performance (e.g. Barrick & Mount, 1991; McHenry, Hough, Toquam, Hanson & Ashworth, 1990; Ones, Viswesvaran & Schmidt, 1995). Further, evidence mounts that personality measures produce small if any differences between majority and protected classes of people (Ones & Viswesvaran, 1998b) and that response distortion does not necessarily destroy criterion-related validity (e.g. Ones & Viswesvaran, 1998a; Ones, Viswesvaran & Reiss, 1996). The remainder of this section examines these developments in more detail.

Personality Taxonomies Early research examining personality-job performance relationships was purely empirical, with no theory to direct it. Without the guidance of an organizing framework of traits, different researchers often used different labels for the same construct or the same labels for very different constructs. Thus, literature reviews during this era compared studies that were not comparable, and compiled predictor-criterion correlations that might have been reasonably hypothesized along with those that would not be expected (Guion & Gottier, 1965). In light of these circumstances, the pessimistic conclusions of such reviews are not surprising. As many personality researchers have recently pointed out, one cannot expect to find reliable predictor-criterion relationships without satisfying two

Current Directions and Issues in Personnel Selection and Classification

307

conditions: First, the predictors must be organized into reasonable taxonomies, and second, predictors must be matched to relevant criteria. The history of attempts to specify a taxonomy of the personality domain dates back to the late 1800s. Several authors provide a detailed account of this history (e.g. Block, 1995; Cattell, 1943; Goldberg, 1993, 1994; Schneider & Hough, 1995). Today’s well-known, hierarchical, five-factor model (FFM) of personality (alternately, “the Big Five”) was first documented in 1961 by Tupes and Christal (see Tupes & Christal, 1992). The five factors were labeled Surgency, Agreeableness, Dependability, Emotional Stability, and Culture. Following Tupes and Christal, McCrae and Costa (1987) replicated a similar model of the FFM. In their version, Extraversion is comprised of traits such as talkative, assertive, and active; Conscientiousness includes traits such as organized, thorough, and reliable; Agreeableness includes the traits kind, trusting, and warm; Neuroticism includes traits such as nervous, moody, and temperamental; and Openness incorporates such traits as imaginative, curious, and creative (Goldberg, 1992). A large amount of evidence supports the generalizability and robustness of the Big Five. The Big Five can be reliably recovered from various personality measures and rating sources (Costa & McCrae, 1988; Digman & Shmelyov, 1996; Goldberg, 1990; McCrae & Costa, 1997), and using a variety of factor extraction and rotation methods (Goldberg, 1990; Katigbak, Church & Akamine, 1996). The FFM is relevant across cultures, countries, and languages (Benet-Mart´nez ı & John, 1998; Digman & Shmelyov, 1996; Katigbak et al., 1996; McCrae & Costa, 1997). Common descriptions of the status of the Big Five include such phrases as “well-accepted” and “supported by an emerging consensus.” A few authors, however, contest this characterization (e.g. Block, 1995; Eysenck, 1991; Loevinger, 1994; Schneider & Hough, 1995; Tellegen, 1993). Some argue that the theoretical value and the practical usefulness of the Big Five factors are severely limited by their breadth, and that important variables are missing from the model. For example, Hough (1992) proposed a nine-factor personality taxonomy, including Affiliation, Potency, Achievement, Dependability, Agreeableness, Adjustment, Intellectance, Rugged Individualism, and Locus of Control. Neither Rugged Individualism (masculinity vs. femininity) nor Locus of Control are represented in the Big Five. Agreeableness, Adjustment, and Intellectance are similar to Big Five factors. Extraversion is divided into Affiliation and Potency, and Conscientiousness is divided into Achievement and Dependability. Whichever taxonomic structure is favored, from a research perspective, the point we want to emphasize is that the appearance of personality taxonomies has been a crucial development in the history of personality research (e.g. Costa & McCrae, 1995; Schneider & Hough, 1995). Scientific knowledge of the correlates

308

WALTER C. BORMAN ET AL.

of personality traits could not be accumulating as it is without reliance upon a common framework. From an applied perspective, important questions pertain to the criterion-related validity of personality variables and the extent to which it matters whether we focus on broad factors or narrower facets. These issues are addressed next.

Criterion-Related Validity of Personality Measures Predictive and Incremental Validity of the Big Five Although the Big Five model of personality is not universally accepted, considerable research has been conducted using this framework. For example, Barrick, Mount and Judge (2001) conducted a second-order meta-analysis including eleven meta-analyses of the relationship between Big Five personality dimensions and job performance. They included six performance criteria and five occupational groups. Conscientiousness was a valid predictor across all criteria and occupations ( = 0.19–0.26),1 with the highest overall validity of the Big Five dimensions. Emotional Stability was predictive for four criteria and two occupational groups. Extraversion, Agreeableness, and Openness did not predict overall job performance, but each was predictive for some criteria and some occupations. Barrick et al. results echoed conclusions of previous research on this topic (e.g. Barrick & Mount, 1991; Hurtz & Donovan, 2000). Some evidence suggests that Emotional Stability is more important in military settings than it is in most civilian occupations. Barrick et al. (2001) reported = 0.05–0.14 for Emotional Stability predicting performance in five categories of civilian jobs. A meta-analysis based exclusively on European samples reported a validity of 0.12 for Emotional Stability in civilian jobs, compared to a much higher 0.30 in military samples (Salgado, 1998). It is important to note that Salgado’s validities are corrected for predictor unreliability in addition to other artifacts, and that his military data were limited to aviator samples and training proficiency criteria. Vickers (1995) found that Emotional Stability is a useful predictor of military leadership criteria, and that its most predictive facets are depression, pessimism, and stress vulnerability (low scores on each being related to better performance). In Project A, Emotional Stability predicted combat effectiveness (Hough, 1992). The predictive value of Emotional Stability in military settings certainly merits further study. Despite the low to moderate magnitude of the Big Five’s predictive validity, optimism regarding personality’s usefulness in selection contexts remains high. One reason is that even a modestly predictive variable, if uncorrelated with other predictors, offers incremental validity. Personality variables tend to be

Current Directions and Issues in Personnel Selection and Classification

309

unrelated to tests of general cognitive ability (g) and thus they have incremental validity over g alone (Day & Silverman, 1989; Hogan & Hogan, 1989; McHenry et al., 1990; Schmitt, Rogers, Chan, Sheppard & Jennings, 1997). For example, Conscientiousness produces gains in validity of 11–18%, compared to using g alone (Salgado, 1998; Schmidt & Hunter, 1998). Salgado (1998) reported that Emotional Stability measures produced a 10% increment in validity over g, for European civilian and military samples combined. For military samples alone, the incremental validity was 38%. In order to fully realize personality’s potential to predict job performance, we might also consider more than just the Big Five and general performance criteria. This approach is a construct-oriented one, involving a personality taxonomy, a job performance taxonomy, and a nomological net linking the two (Hough & Paullin, 1994). Criteria and promising predictors of them are selected based on theory and/or job analysis. Traditional job analysis techniques focus on cognitive and psychomotor job requirements, but some systems exist that identify job-relevant personality requirements, as well (Borman, Kubisiak & Schneider, 1999; Raymark, Schmit & Guion, 1997; S¨ mer, u S¨ mer, Demirutku & Cifci, 2001). u ¸

Predictive Validity of Personality, Beyond Overall Performance By far, overall job performance ratings are the most commonly-used criteria in selection research. To thoroughly understand the predictive power of personality variables, however, the taxonomic structure of the criterion space is as important as that of the predictor space (Borman, Hanson & Hedge, 1997). For example, overall performance includes both task and citizenship performance. As described previously in the criterion section, research has shown at least a modest improvement in validity of personality predictors when contextual or citizenship performance is the criterion rather than task or overall performance (Borman, Penner, Allen & Motowidlo, 2001; Hurtz & Donovan, 2000; McHenry et al., 1990; Motowidlo & Van Scotter, 1994).

Predictive Validity Beyond the Big Five Meta-analyses based on the Big Five can mask the predictive validity of narrower traits, such as the Affiliation and Potency facets of Extraversion (Hough & Ones, 2001; Schneider, Hough & Dunnette, 1996). Potency, for example, is a much better predictor of salesperson performance than are the broader trait of Extraversion or the other facet of Extraversion, Affiliation (Barrick et al., 2001;

310

WALTER C. BORMAN ET AL.

Vinchur, Schippman, Switzer & Roth, 1998). Across a broad range of occupations, Potency and Affiliation are differently related to a number of criteria; furthermore, Affiliation is often the superior predictor (Hough, 1992). Similar findings emerge when two sub-components of Conscientiousness are considered separately rather than together. Achievement is better than both Dependability and Conscientiousness at predicting sales performance criteria (Vinchur et al., 1998). Across occupations, Achievement is usually a better predictor than is Dependability (Hough, 1992; Vickers, 1995). For predicting personal discipline and law-abiding behavior, however, Dependability is better than Achievement (McHenry et al., 1990). Another important mission for researchers is to devise and agree upon a comprehensive taxonomy of personality at the facet level (i.e. more specific than the Big Five). An organizing framework for narrow personality traits will be crucial to further progress in the understanding of links between personality and various criteria.

Compound Traits In contrast to considering links between narrow personality traits and individual job performance criteria, a very different approach for personality applications in personnel selection is the development of compound traits (e.g. integrity, customer service orientation). In a hierarchical model, covarying facets like anxiety and vulnerability make up higher-order multifaceted traits (like Neuroticism). Compound traits, like multi-faceted traits, are also collections of narrower facets (Schneider & Hough, 1995). The important distinction is that multifaceted traits are said to cause their constituent facets, whereas compound traits emerge from their facets. Consequently, facets of a multifaceted trait covary, but facets of a compound trait may or may not. Multifaceted traits, such as the Big Five factors, are important for summarizing and describing personality on a theoretical level, as well as for their potential relationships with criteria. Compound traits, however, have the potential to show even stronger relationships with criteria, because they are often constructed by identifying the criterion first, and then selecting a heterogeneous group of variables expected to predict it (Hogan & Ones, 1997). Integrity, adaptability, and core self-evaluation are three compound traits that may be especially useful in a selection context. Integrity Integrity consists of facets from all Big Five factors, mainly Conscientiousness, Agreeableness, and Emotional Stability (Ones, Schmidt & Viswesvaran, 1994).

Current Directions and Issues in Personnel Selection and Classification

311

Sackett, Burris and Callahan (1989) distinguished between overt and personalitybased integrity tests. Overt tests include straightforward questions regarding the test-taker’s own dishonest behavior and attitudes toward dishonest behaviors. Personality-based tests use more subtle items and are often derived from traditional personality inventories (e.g. Hogan & Hogan, 1989). It is difficult to compare overt to personality-based integrity tests, in part because the former are most often validated against counterproductive behaviors and the latter are most often validated against overall job performance ratings. Integrity tests have several advantages as part of selection systems, including considerable appeal to employers. Given the enormous costs of employee theft and other counterproductive behaviors (Durhart, 2001; U.S. Department of Health and Human Services, 1997), it is understandable that employers want to avoid hiring dishonest applicants. Also, paper-and-pencil integrity measures have grown in popularity since the 1988 Federal Polygraph Protection Act banned most preemployment uses of the polygraph test. Empirical evidence shows that integrity tests are better than any of the Big Five at predicting job performance, = 0.41 for predicting job performance ratings (Ones et al., 1995). Under some conditions, integrity tests also predict various counterproductive behaviors at work. Integrity tests are uncorrelated with tests of g (Hogan & Hogan, 1989; Ones et al., 1995), and produce a 27% increase in predictive validity over g alone (Schmidt & Hunter, 1998). Understandably, some have been concerned about false negatives in integrity testing: the incorrect classification of honest persons as dishonest. One aspect of this problem is the distasteful nature of the label “dishonest.” This concern is easily addressed: Employers rarely provide job applicants with test scores, score interpretations, or even the exact reason why they were not hired (Sackett & Wanek, 1996). Furthermore, a predictor with relatively high validity will produce fewer false negatives and false positives than will predictors with weaker predictive validity (Ones, Schmidt, Viswesvaran & Lykken, 1996). A more difficult issue related to false negatives is the possibility that these errors are systematic rather than random. Integrity tests might discriminate against highly religious or moral individuals – those compelled to be honest about their faults and past misdeeds (Lilienfeld, 1993; Lilienfeld, Alliger & Mitchell, 1995). Very little data exist, however, to address the accuracy of this possibility. Adaptability As discussed earlier in the criterion section, adaptive job performance has become increasingly important in today’s workplace. In personnel research, adaptability is most often operationalized as a criterion, but we propose that it could also be viewed as a compound personality trait. Recognizing

312

WALTER C. BORMAN ET AL.

that adaptive performance is a function of both general cognitive ability and personality, (Allworth & Hesketh, 1999; Le Pine, Colquitt & Erez, 2000; Mumford, Baughman, Threlfall, Uhlman & Constanza, 1993; Pulakos & Dorsey, 2000), we will focus on personality predictors of adaptive performance here. Existing personality scales carrying the adaptability label tend to be narrowly focused on a particular aspect of adaptability (e.g. International Personality Item Pool, 2001) and would probably not be sufficient to predict the multiple dimensions of adaptive job performance. One way to construct a broad predictor measure of adaptability would be to use a criterion – keying approach, as has been done to construct personality – based integrity tests. Although somewhat speculative, we might surmise that someone who readily adapts on the job is likely to be patient, even-tempered, and confident (facets of Emotional Stability); open to new ideas, values, and experiences (facets of Openness); and determined to do what it takes to achieve goals (a facet of Conscientiousness). It is possible that high levels of other aspects of Conscientiousness, such as dutifulness and orderliness, are detrimental to adaptive performance because they involve over-commitment to established ways of functioning. Again, although speculative, these relationships are in line with the results of several different studies that linked personality variables to adaptive performance (e.g. Le Pine et al., 2000; Mumford et al., 1993; Pulakos & Dorsey, 2000). Core Self-Evaluation As defined by Judge and his colleagues, core self-evaluation (CSE) is a fundamental, global appraisal of oneself (Judge, Locke, Durham & Kluger, 1998). We should note that whereas some have called CSE a compound trait (Schneider & Hough, 1995), Judge and his colleagues might not agree with this characterization. Erez and Judge (2001) described CSE as a single, higher-order factor explaining the association between four more specific traits, namely self-esteem, generalized self-efficacy, internal locus of control, and emotional stability. Indeed, the high intercorrelations between these four traits (Judge & Bono, 2001; Judge, Erez & Bono, 1998) indicate that CSE is perhaps a multifaceted trait rather than a compound trait. A meta-analysis showed that CSE’s constituent traits are impressively predictive of overall job performance (Judge & Bono, 2001). Estimated values, corrected for sampling error and criterion unreliability, ranged from 0.19 (emotional stability) to 0.26 (self-esteem). Considering the strong relationships between the four CSE components, and the fact that self-esteem is the best predictor of CSE, it is not clear whether there is any practical advantage to CSE over self-esteem alone. Further research is needed to settle this question.

Current Directions and Issues in Personnel Selection and Classification

313

As Judge et al. predicted, CSE is even more strongly related to job satisfaction than it is to overall job performance. Predictive validity estimates ranged from 0.24 (emotional stability) to 0.45 (self-efficacy; Judge & Bono, 2001). As with job performance, it is possible that individual CSE components predict job satisfaction as well as or better than the complete construct. People with high CSE seem to seek out challenging jobs and apply a positive mindset to the perception of their jobs (Judge, Bono & Locke, 2000; Judge, Locke et al., 1998). The result is desirable job characteristics, both real and perceived, which contribute to high job satisfaction. The link between CSE and job satisfaction has important implications, because satisfied employees are more likely to stay on the job than are dissatisfied employees (Harter, Schmitt & Hayes, 2002; Tett & Meyer, 1993). Low turnover rates are especially important in organizations that invest heavily in the training of new employees.

Personality and Adverse Impact Based on a summary of the literature, Hough, Oswald and Ployhart (2001) demonstrated that most subgroup differences on personality measures are negligible or small. There are, however, a few exceptions that could make practical differences in selection decisions. For example, Blacks score slightly lower than Whites on Openness and Affiliation. Also, there are some small gender differences in personality variables, favoring men for Emotional Stability and Potency, and favoring women for Agreeableness and Dependability (Hough et al., 2001). Women also score slightly higher on overt integrity tests (Ones & Viswesvaran, 1998b). Nonetheless, when added to cognitive predictors, personality variables can concurrently enhance predictive validity and reduce adverse impact.

Stability of Personality Over Time Longitudinal research shows that individuals’ standing on the Big Five is reasonably consistent over time. For example, testing at five time intervals between the ages of 12 and 62 yielded average test-retest correlations, corrected for internal consistency reliability, from 0.43 (Agreeableness) to 0.59 (Conscientiousness; Judge, Higgins, Thoresen & Barrick, 1999). Roberts and Del Vecchio’s (2000) extensive review also showed that personality is stable, and that it becomes gradually more so later in life. Test-retest reliabilities for the Big Five ranged from 0.50 to 0.54. In short, personality variables can be used in selection without concern that individuals’ standing on them will change dramatically over time.

314

WALTER C. BORMAN ET AL.

Response Distortion One of the most frequent objections to the use of personality in selection settings involves the risk of response distortion. The concern is that applicants will respond to inventory items in ways that they believe will make them look more desirable. There is no doubt that people can fake-good on personality inventories, when instructed to do so (Viswesvaran & Ones, 1999). Knowing this, several questions remain: (1) Do applicants fake on personality inventories in selection settings? If so, to what extent? (2) If faking occurs, what is its impact on criterion-related validity and hiring decisions? (3) If faking does have a negative impact, what can be done about it? The question regarding the extent to which applicants slant their responses on personality inventories is most often addressed by comparing applicant scores to incumbent scores, presuming that incumbents (when told that their scores are for research purposes only) are not motivated to distort their responses, whereas applicants may be. Such comparisons reveal that on average, applicants do fake-good in selection settings, but that applicant versus incumbent differences are markedly smaller than differences found in laboratory “fake-good” versus “honest” conditions (Viswesvaran & Ones, 1999). In Project A, average personality test scores for incumbents directed to be honest were highly similar to those for applicants told that the tests would be used to make decisions about their Army careers (Hough et al., 1990). Some studies, however, have found significantly more favorable personality profiles for applicants than for incumbents (e.g. Rosse, Stecher, Miller & Levin, 1998). Despite the fact that job applicants can and do slant their responses on personality inventories, some believe such faking detracts only slightly, if at all, from criterion-related validity in applied situations (Hogan, Hogan & Roberts, 1996; Hough et al., 1990; Hough & Ones, 2001; Ones et al., 1995; Ones & Viswesvaran, 1998a). Not everyone is convinced that faking can be so easily dismissed. For example, Rosse et al. (1998) considered faking’s effects on selection system fairness. Using a response validity scale to detect and quantify response distortion, they found considerable variation in the magnitude of response distortion across individuals. Rosse et al. demonstrated that using Conscientiousness scores in a top-down selection strategy would result in large proportions of the most flagrant fakers being hired. However, the most dramatic effects were present only in highly selective circumstances (i.e. with a selection ratio of 0.05). Even those who report that faking is not a serious problem often recommend taking action to combat its potential effects (e.g. Hough et al., 1990; Hough & Ones, 2001). Hough and colleagues reviewed evidence in support of two such

Current Directions and Issues in Personnel Selection and Classification

315

strategies, concluding that: (1) warning test takers that distortion can be detected and will have consequences does deter faking; and (2) the response validity scales included in many personality inventories are effective tools for detecting faking.

Conclusions Over the past two decades, personality has enjoyed a well-deserved resurgence in research attention, aided in no small part by the wide acceptance of the Big Five model of personality. Only two Big Five factors, Conscientiousness and Emotional Stability, are consistently related to job performance across occupations and criteria, and their validities are of modest magnitude (Barrick et al., 2001). Nonetheless, even these broad traits offer incremental predictive validity over cognitive ability alone (Salgado, 1998; Schmidt & Hunter, 1998). Using personality predictors narrower than the Big Five shows promise for revealing stronger criterion-related validities. For example, Conscientiousness is related to performance, but the Achievement facet of the Big Five tends to be the stronger predictor (Hough, 1992; McHenry et al., 1990; Vickers, 1995). It will be important for researchers to agree on a more detailed personality taxonomy than the Big Five, even just as a starting point, if systematic research is to accumulate linking more narrow traits to performance. A taxonomy of performance is equally important, especially considering the fact that personality variables are at least somewhat more strongly related to citizenship performance than to technical performance. A complete personality taxonomy, including lower-level facets, will also be useful for identifying compound traits. Such heterogeneous predictor composites are likely to produce stronger correlations with performance in some cases than do homogenous, multifaceted traits. Compound traits may be especially important in organizational environments that are more dynamic and team-oriented (Edwards & Morrison, 1994). Complex, heterogeneous predictors are needed to predict the complex, heterogeneous performance criteria that are likely in such an environment. Recent research has shown not only the criterion-related validity of personality, it has also addressed certain persistent objections to personality testing. Adverse impact appears not to be a problem and response distortion, though certainly problematic in selection settings, may be alleviated with targeted interventions. Also, considerable evidence exists to suggest that personality is impressively stable over the entire life span. Thus, we can be confident that personality variables used in selection will predict performance of selectees throughout their tenure with the organization.

316

WALTER C. BORMAN ET AL.

VOCATIONAL INTERESTS
In this section we examine vocational interests in relation to personnel selection and the prediction of work performance. In particular, we first discuss vocational interests as a domain and where this domain fits with other domains such as values, needs, motivation, and personality. Next, because vocational interest and personality share relatively close conceptual ties and personality has become a popular predictor in selection research, empirical relations between constructs from the two domains are examined. Then, we review the literature linking vocational interests to job satisfaction and job performance criteria.

Definitional Issues Needs, drives, values, and interests are closely related motivational concepts that refer to the intentions or goals of a person’s actions. Interests are generally thought of as the most specific and least abstract of these concepts (Hogan & Blake, 1996). For example, Super (1973) observed that needs are the most general construct, with values and interests serving as lower order constructs. In this conceptualization, interests consist of the specific activities individuals engage in to satisfy their needs. Somewhat in contrast, Dawis (1980) suggests that attitudes are the most general concept, needs and values refer to the importance or unimportance of an object to a person, and interests refer to the liking (or disliking) of the object. In both views, however, interests are seen as the most specific motivational concept. Hogan and Blake (1996) point out that vocational interests have not often been studied in relation to other motivational constructs. Holland and Hough (1976) believe this is unfortunate from a theoretical perspective, but that a likely reason for this lack of attention to theoretical links with other constructs is the early empirical successes of vocational interest inventories, predicting outcomes such as vocational choice. In a sense, there was little reason to relate to the rest of psychology because of these successes. Accordingly, many psychologists have regarded the area of vocational interest measurement as theoretically and conceptually barren (Strong, 1943).

Relations Between Personality and Vocational Interests One important exception to this atheoretical tendency on the part of vocational psychologists is reflected in research to link interests and personality. Many vocational psychologists have considered the two constructs as closely related or even equivalent (Darley & Hagenah, 1955; Hansen, 1984; Holland, 1973). For example,

Current Directions and Issues in Personnel Selection and Classification

317

Holland (1973) has written about vocational interests as “expressions of personality,” that is, expressions of personality in work, school subjects, hobbies, etc. Although the concepts may be closely linked conceptually, it is evident that inventories measuring the two constructs are quite different. Personality inventories present items that presumably reflect the respondent’s tendency to act in a certain way in a particular situation. Vocational interest items elicit like-dislike responses to objects or activities. Hogan and Blake (1996) suggest that the implication is that interest inventories are better at getting closer to a person’s self-concept. They point out that responses to interest items more closely reflect normal social interactions with a new acquaintance. That is, we are more likely to open such a conversation talking about what we like rather than what we tend to do in a particular situation. Thus, several behavioral scientists have suggested that personality and vocational interests are closely related conceptually. And yet, it is clear that inventories measuring the two sets of constructs have quite different types of items. The question arises, how are the two domains linked empirically? There has been a fair amount of empirical research correlating personality and vocational interest responses. In a very useful table, the findings of several studies linking personality and vocational interests (e.g. Blake & Sackett, 1993; Hogan & Hogan, 1995; Holland, Johnson, Asama & Polys, 1993). At the level of the Big Five personality factors and the six Holland types, Extraversion constructs relate primarily with Social and Enterprising types, Conscientiousness correlates with the Conventional type, Agreeableness relates to Social and Realistic (-) types and Openness to Experience relates to the Artistic type and to some extent with the Investigative type. Neuroticism is not linked to any of the Holland types. Although many of these correlations are significant, the magnitude of the relationships is not very large. The highest correlations are in the 40s (one, the NEO Openness and Artistic was 0.52), but most of the significant relations were between 0.20 and 0.30. Thus, it appears that personality constructs and vocational interest constructs have theoretically and conceptually reasonable and coherent relationships but that these linkages are relatively small. What does this mean for selection research? Although the two predictor domains have substantial conceptual overlap, they do not correlate highly, and thus examining vocational interests separately as a predictor of job performance (and job satisfaction and attrition) seems warranted.

Relationships Between Vocational Interests and Job Performance/Job Satisfaction The most obvious link is between vocational interest responses and occupational tenure, and by extension, job satisfaction. For the most part, the tenure relation

318

WALTER C. BORMAN ET AL.

has been confirmed. For example, regarding occupational tenure, Strong (1943) demonstrated that occupational membership could be predicted by vocational interest scores on the Strong Vocational Interest Blank administered between five and 18 years previously. This finding at least implies that persons suited for an occupation on the basis of their interests tend to gravitate to and stay in that occupation. For job satisfaction, the relationships are more mixed. The best estimate of the average interests-satisfaction relation was provided by Barge and Hough (1988). They computed a median correlation across 18 studies and found it to be 0.31. There are several potential reasons why this correlation is not higher. First, restrictionin-range is a possibility, with persons in the U.S. workforce, at least, generally expressing satisfaction with their jobs (e.g. Campbell, 1971). Second, there are many other factors possibly influencing overall job satisfaction, such as satisfaction with pay, the supervisor, and coworkers (e.g. Dawis & Lofquist, 1984). Finally, there is evidence (e.g. Staw & Ross, 1985) that job satisfaction may be, at least in part, an individual differences variable, with many workers tending to be either satisfied or not satisfied no matter what job they are holding. Thus, there are likely reasons why vocational interest-job satisfaction correlations are moderate, at best. Nonetheless, it is disappointing that interest measures do not better predict this highly important dependent variable. Vocational interests are not usually thought of in a personnel selection context, and, in fact, there are not many studies linking vocational interests and job performance. As of 1988, Barge and Hough (1988) identified eleven studies, with a median validity for the interest predictors against job performance of 0.20. Although this level of validity is not very high, and the number of studies represented is not large, this observed correlation of 0.20 compares rather favorably to the validities of personality constructs (e.g. Barrick & Mount, 1991). One of the studies reviewed by Barge and Hough was conducted in the U.S. Navy. Borman, Rosse and Abrahams (1980) administered vocational interest items to Navy recruiters (N = 267) as part of a test validation study. The performance of these same recruiters was evaluated using supervisor and peer raters. Then Borman et al. took all interest items that demonstrated validity above a specified cut-off level and factor analyzed only the valid items to identify potentially valid constructs. The factors emerging from the factor analysis were quite interpretable. Some examples were: (1) interests in extraverted, dominant, leadership activities and occupations; (2) interests in law and politics; (3) interests in sports and competitive activities; and (4) interests in teaching and counseling. The next step was to write additional interest items to reflect these constructs, and the “old” valid items along with the new items were administered to a separate

Current Directions and Issues in Personnel Selection and Classification

319

validation sample of 194 Navy recruiters. This cross-validation study showed reasonably good correlations between the construct scores and job performance ratings, 0.21 or 0.22 for the construct composites against three of the four rating criteria (Selling, Human Relations, and Overall Performance). Vocational interests were also used as a predictor in Project A (the large-scale selection research project conducted in the U.S. Army: Campbell & Knapp, 2001). The Army Vocational Interest Career Examination (AVOICE) was developed to cover interests especially relevant to Army enlisted jobs. Thus, with reference to Holland’s themes, Enterprising items were not represented, Artistic items were under-represented, and Realistic item content was heavily weighted. The AVOICE was administered to several samples during the Project A research program and revised after each administration based on validation results. Also, a number of scoring strategies were employed, including both rational and empirical approaches. In predicting job performance criteria, empirical keying tended to produce slightly higher validities than did rational keys (Paullin, Bruskiewicz, Hanson, Logan & Fellows, 1995). For example, in the concurrent validation sample, the median validity across four jobs was 0.15 for the rational keys; the median cross-validity across the same four jobs for the empirical keys was 0.21. In the longitudinal sample, the respective validities were 0.12 and 0.23. When corrected for range-restriction, the latter two validities reached 0.29 and 0.33. Analogous to personality measures, vocational interest inventories used for selection have serious potential problems with slanting of responses or faking. It has long been evident that people can fake interest inventories (e.g. Campbell, 1971). However, some research has shown that in an actual selection setting, applicants may not slant their responses very much (e.g. Abrahams, Neumann & Githens, 1971), although it is unclear why faking would be any different with vocational interest and personality inventories.

Conclusions Vocational interest measures, similar to personality measures, tap motivationrelated constructs. Interests have substantial conceptual similarity to personality, but empirical links between the two sets of constructs are rather modest. Vocational interest measures are most often used in counseling settings, and have been linked primarily to job satisfaction criteria. However, although not often used in a selection context, limited data suggest reasonable levels of validity. Accordingly, vocational interests may show some promise for predicting job performance.

320

WALTER C. BORMAN ET AL.

BIODATA
Biographical information (i.e. “biodata”) has been widely used in personnel selection since the early part of the century. In general, biodata shows encouraging validities across a number of work-related outcomes (e.g. turnover, tenure, job performance, and training performance). In addition, research indicates that biodata shows less adverse impact in comparison to cognitive ability tests (Wigdor & Garner, 1982). The following section provides an overview of the history and recent developments related to biodata, as well as a discussion of links between biodata and other predictors. Methodological criticisms are also presented. Finally, ethical and legal concerns are considered in the framework of maintaining high predictive power. The primary principle behind the use of biodata is that the best predictor of future behavior is past behavior. In fact, biodata offers a number of advantages when used in personnel selection. Among the most significant is its power as a predictor across a number of work-related criteria. For example, in a recent meta-analytic review of over 85 years of personnel psychology research, Schmidt and Hunter (1998) reported mean biodata validity coefficients of 0.35 and 0.30 against job and training success, respectively. These findings support previous research reporting validities ranging from 0.30 and 0.40 between biodata and a range of criteria such as turnover, absenteeism, job proficiency, and performance appraisal ratings (e.g. Asher, 1972; Hunter & Hunter, 1984; Reilly & Chao, 1982). Based on these meta-analytic results, researchers have concluded that biographical inventories have almost as high validities as cognitive ability tests (Reilly & Chao, 1982). Importantly, the high predictability associated with biodata, the ease of administration of biodata instruments, the low cost, and the lack of adverse impact have led to the widespread use of biodata in both the public and private sectors (Farmer, 2001).

Types of Biodata Items Recently, researchers have begun to develop a taxonomic framework for categorizing biodata research (e.g. Mael, 1991; Stokes, Mumford & Owens, 1994). For example, Mael (1991) classifies biodata attributes into ten dimensions, in turn subsumed under three major themes: (1) historical; (2) methodological; and (3) legal/moral. The historical category includes information regarding events that have previously, or are currently, taking place. The methodological category deals primarily with ensuring the accuracy of the information reported, and includes

Current Directions and Issues in Personnel Selection and Classification

321

externality (whether the actions being recalled could be observed by others), objectivity (whether the recollections were objective as opposed to subjective), first-handedness (whether evaluation of the recalled behavior was made by the respondent or by others), discreteness (whether the recalled behavior was exhibited just one time or on multiple occasions across time), and verifiability. Finally, the legal/moral category includes the notions of controllability (whether or not the respondent chose to engage in the action), equal accessibility (whether or not the activity could have been engaged in by anyone), job relatedness, and invasiveness. Mael (1991) notes that some authors are quite prescriptive regarding the use of certain item attributes (i.e. controllable rather than noncontrollable events), whereas others simply point out the advantages and disadvantages associated with each. In addition, practical constraints (e.g. availability of information, legal issues) will likely influence the biodata attributes attended to in selecting items within the applied sector.

Developing and Validating Biodata Items There are two major strategies for developing and validating biodata inventories. The most common strategy is the empirical keying approach (i.e. the weighted application blank). The process of empirical keying examines relationships between responses on each response option and scores on the criterion of interest. Item selection is based on the magnitude of these relationships, and the weighting of the items is contingent upon the strength and direction of the correlation between the item and the criterion. A composite score is created based on these weights, which is then correlated with the criterion to determine overall validity. Finally, these results are usually cross-validated on a separate sample to control for sample-specificity (Mumford & Owens, 1987; Nickels, 1994; Wesley, Mumford & McLoughlin, 1987). A variety of empirical keying methods exist (see Laurence & Waters, 1993, for a review). However, studies comparing the validities of different techniques have yet to find a conclusive advantage to any of these methods (e.g. Aamodt & Pierce, 1987; Devlin, Abrahams & Edwards, 1992). Although empirical keying remains a common practice, researchers have identified certain drawbacks concerning the method. The first of these has to do with the specific nature of the criterion. That is, the extent to which results can be generalized across jobs from any empirical key is contingent upon “the generality of the criterion measure” (Nickels, 1994, p. 6). In addition, the nature of the

322

WALTER C. BORMAN ET AL.

criterion used can dictate the predictive power of the empirical key (Mumford & Owens, 1987; Thayer, 1977). Two other major criticisms are that empirically derived keys are subject to capitalization on chance (Hunter & Hunter, 1984), and relatedly, that this method is sensitive to sample characteristics, resulting in the shrinkage of the predictor-criterion correlations during cross-validation (Mael, 1994). The most frequent complaint, however, regarding empirical keying is the lack of theory behind its development, leading some researchers to label it as “dustbowl empiricism.” As stated by Laurence and Waters, “A common characterization of biodata inventory construction is that a multitude of items are plucked from the sky, without any rhyme or reason, and retained if they adequately contribute to criterion prediction” (1993, p. 54). Taken together, these criticisms have led researchers to explore alternative methods that are based on a foundation that is more construct-oriented as opposed to empirical. Termed the “rational” approach, this method assigns values to response alternatives based upon theorized relationships with a criterion construct(s) (Mael, 1994). This approach generally begins with a job analysis to define the nature of the criterion. This information defines what area of the criterion’s domain items should tap, and items are generated to cover these criteria. Thus, instead of a purely empirical relationship, items are theoretically and “psychologically” related to the criterion of interest, with an increased focus on both construct and content validity as opposed to pure prediction (Laurence & Waters, 1993). Research has shown that the predictive validities associated with the rational approach are somewhat lower than those obtained using empirical keying (Hornick, James & Jones, 1977; Matterson, 1978). In response, a number of hybrid approaches combining both rational and empirical methods have begun to surface. For example, some studies have used factor analysis to investigate how individual items cluster together, with the factors interpreted as biographical constructs (Steinhaus, 1988). Relationships are then assessed between these factors and the target criterion. Another hybrid technique is subgrouping, which groups individuals according to the similarity of their responses to biodata items and then empirically examines differences between subgroup means on the target criterion (Owens, 1976). For the most part, research has found moderate support for the hybrid approaches. For example, some researchers found subgroups to be reasonably stable over time (Eberhardt & Muchinsky, 1982; Neiner & Owens, 1982), although more recent studies reported that some subgroups tend to be more stable than others (Hecht, Finch, Landau & Stokes, 2002). In general, Mitchell and Klimoski (1982) and Mumford and Owens (1987) have concluded that results based on factorially-based, rationally-keyed, and temperament type measures are more stable than results based on empirically-keyed biodata items, but that, as

Current Directions and Issues in Personnel Selection and Classification

323

mentioned, findings using the former measures tend to show somewhat lower validities (as cited in Laurence & Waters, 1993). In light of this evidence and also in appreciation of increases in the theoretical and conceptual contributions of rational-based methods of item development, measures with a rational foundation that are also empirically-keyed continue to increase in popularity because they combine construct validity, stability, and at least moderate predictive power (Laurence & Waters, 1993).

Relationship of Biodata to other Predictors Biodata and Personality Mumford and Stokes (1992) noted biodata items often appear to be variants of the type of questions found in self-report personality inventories. This observation is reinforced by the empirical result that biodata are often strong predictors of scores on personality scales (Owens, 1976). In fact, Mumford and Owens (1987) found that biodata factors resembling the “Big Five” factors of personality emerged in their research. The aforementioned has led some to assume biodata items are simply another format for measuring personality (Mumford, Snell & Reiter-Palmon, 1994), or temperament (Buss & Plomin, 1975). This position would certainly be consistent with those (Allport, 1937) who include an individual’s experience in their definition of personality. More recently, others (Ashworth, 1989) focused on the distinction between the two being somewhat arbitrary and artificial. If, however, the distinction is made between “hard” (i.e. verifiable and factual) and “soft” (i.e. private and unverifiable) biodata, a clear delineation exists. In fact, in a recent study, Shultz (1996) tested a number of confirmatory factor analytic models of multitrait-multimethod matrices, and found personality and soft biodata items represented one factor, and hard biodata items represented a second. With this in mind, several researchers (Mael, 1991; Mumford & Owens, 1987; Owens, 1976) have defined biodata the way Asher (1972) defined “hard” biodata, though this is in no way a universal characterization (Mael, 1991; Mael & Schwartz, 1991). When considering the domains of interest from a measurement perspective, the differences between biodata and personality become evident. Self-report personality items generally elicit information regarding an individual’s predisposition or general behavioral tendency toward a particular situation. Biodata items on the other hand, focus on prior behavior and experiences occurring in specific situations (Mumford & Stokes, 1992). Also, whereas personality item responses are arguably influenced only by dispositional factors, biodata items

324

WALTER C. BORMAN ET AL.

capture aspects of the environment that affect and are affected by the individual. In addition to personal characteristics, they are tied to social factors as well (Mael, 1991). Mumford, Snell and Reiter-Palmon (1994) noted there are two additional significant points of departure for personality and biodata. The first concerns the element of choice. Biodata measures often capture behavioral patterns that are explicitly tied to the decisions individuals make when presented with a particular situational stimulus. Personality measures, on the other hand, are not tied to a particular decision or choice, but more to a preference. Second, biodata items often tap into content areas that are influenced more by individual knowledge or skills than by personality. In fact, biodata-type items are often used as a preferred vehicle for accessing job-relevant information necessary to assess knowledge, skills, or abilities (Mumford, Snell & Reiter-Palmon, 1994). This is not the case for personality measures. Overall, as a consequence of the shift from empirical to rational and hybrid approaches to development and scoring of biodata, the distinction between biodata and temperament or personality scales has become less clear (Mael, 1991). Nonetheless, a core difference between the two types of predictors remains. In particular, the hallmark of biodata is its emphasis on past behaviors and their influence on future behavior. Biodata, Interests, and Cognitive Abilities Mumford and Stokes (1992) pointed out that biodata items demonstrate a certain amount of overlap with vocational interest inventories (Eberhardt & Muchinsky, 1984). By tapping into past occurrences of behavior, especially those that are directly a function of or are related to particular occupations, biodata measures capture important determinants of interests (Mumford & Stokes, 1992). Mumford and Stokes (1992) noted likely relationships with attitudes and values also exist for biodata. As Mumford and Stokes (1992) stated, relationships between biodata and measures of cognitive abilities have received less attention than that for other areas. As they and others (Mitchell, 1994) have pointed out, there is a fundamental difference between cognitive abilities as they are typically measured and the way they are captured with biodata. Generally, aptitude or ability measures are constructed in such a way as to elicit maximal performance in a somewhat artificial problem-solving situation. However, biodata may be more useful in the prediction of typical or “everyday” behavior (Mitchell, 1994). Though biodata do not usually provide information on the upper bounds for performance, Mumford and Stokes (1992) speculate that they may be tapping into the same variance as measures of practical intelligence (Sternberg, 1985; Wagner & Sternberg, 1985).

Current Directions and Issues in Personnel Selection and Classification

325

In fact, properly constructed biodata measures may be the best way to assess the types of intelligence that are actually better predictors of some real world outcomes (Gordon, 1997) such as job and life success.

Ethical and Legal Issues with Biodata Although most criticism surrounding biodata centers around the use of empirical keys, Mael (1991) reviewed certain ethical and legal concerns that have also been raised. The first of these deals with the controllability of events. That is, there are actions that the respondent chooses to engage in (controllable events), whereas other events were either imposed upon them or happened to them (noncontrollable events). Despite the belief held by numerous biodata researchers that all events, whether controllable or not, have the potential to influence later behavior, some researchers (e.g. Striker, 1987, 1988) argue that it is unethical to evaluate individuals based on events that are out of their control (e.g. parental behavior, socioeconomic status). As a result, some have either deleted all noncontrollable items from their biodata scales (e.g. the Armed Services Application Profile – ASAP), or have even created new measures with the exclusion of these items (e.g. the Air Force’s Leadership Effectiveness Assessment Profile – LEAP). A frequent consequence, however, is that using only controllable items reduces the validity of the biodata instrument (Mael, 1991). Two other ethical and legal concerns that have been raised include equal accessibility and invasion of privacy. That is, some researchers (e.g. Striker, 1987, 1988) argue that items dealing with events not equally accessible to all individuals (e.g. playing varsity football) are inherently unfair, and should not be included. Similarly, the current legal climate does not encourage the use of items perceived as personally invasive. Overall, adhering to certain conditions that minimize such issues as invasiveness might be encouraged. What should be avoided, however, is a reliance on subjective and less verifiable items, that compromises the primary goal of retrieving relatively objective, historical data from applicants.

Conclusions Biodata predictors are a powerful, non-cognitive alternative to cognitive ability tests that have shown significant promise as a predictor in selection. The principle relative to biodata is that past behaviors matter, and should be taken into account when criteria such as performance, absenteeism, and other work-related outcomes

326

WALTER C. BORMAN ET AL.

are being predicted. In addition, efforts are currently underway to develop a more theoretical understanding of the constructs involved with biodata. Finally, ethical and legal concerns are also being addressed in hopes of creating an acceptable compromise between high predictability and overall fairness. Thus, although there is still much to be done in understanding how past behaviors can be used in personnel selection, evidence suggests that enhancing biodata techniques seems like a step in the right direction.

SITUATIONAL JUDGMENT TESTS
A situational judgment test (SJT) presents descriptions of challenging but realistic job-related situations, and respondents indicate which of several alternative courses of action they would take. SJTs have been used for nearly a century (see McDaniel, Morgeson, Finnegan, Campion & Braverman, 2001 for a history of early SJTs). SJTs’ popularity in research and practice has boomed in recent years (Borman, Hanson & Hedge, 1997). Advantages of SJTs include favorable user reactions (Olson-Buchanan et al., 1998) and low adverse impact relative to cognitive tests (Chan & Schmitt, 1997). SJTs are also more economical to develop and administer than are higher-fidelity job simulations such as in-basket tests and role-plays. Importantly, SJTs often produce criterion-related validity comparable to that of some of the best available predictors of job performance in personnel selection (McDaniel et al., 2001; Schmidt & Hunter, 1998), and are related to performance in military (Hanson & Borman, 1993; Hedge et al., 2000; Pulakos & Dorsey, 2000) as well as civilian occupations. In short, the SJT is an attractive candidate for a personnel selection battery. The present section details development issues, research findings, and research needs regarding SJTs.

Development Procedures and Test Characteristics SJTs are typically developed by first having subject matter experts (SMEs) generate challenging, realistic situations. Test developers may then edit the situations to a common format, and these situations are administered to job incumbents, with the instructions to write about what they would do in each of the situations. Typically, themes will emerge from these responses (i.e. several different ways to handle each situation), and the test developer can write response options targeted to each of the themes (e.g. Hanson & Borman, 1993; Motowidlo, Dunnette & Carter, 1990). There are several variations on SJT content and format, instructions to test-takers, judgments requested of test-takers, and scoring procedures. Regarding

Current Directions and Issues in Personnel Selection and Classification

327

content and format, SJTs can be targeted toward personal characteristic constructs or toward the job in general, with the content intended to sample widely the important situations a job incumbent is likely to face. As an example of the former, Motowidlo et al. (1990) targeted with their SJT toward problem solving, communication, and interpersonal skills. With many SJTs, however, the latter content sampling strategy is not employed, and the resulting test is likely to be quite multidimensional. Besides the paper-and-pencil format, video-based and computerized SJTs have been developed (e.g. Chan & Schmitt, 1997; Olson-Buchanan et al., 1998; Weekley & Jones, 1997). When the situations are presented via video, response options may be written, narrated, or both at once. Items and response options also vary in length and complexity. Another issue in SJT development involves the instructions for test-takers. One possibility is for respondents to rate the effectiveness of every available response option. Another is to ask respondents to select, for each situation, the most (or most and least) effective response (i.e. what should be done in each situation). A third option is for respondents to indicate the response they would most (or most and least) likely perform (i.e. what would you do in each situation). Each option carries potential benefits and drawbacks. Rating the effectiveness of each response generates the largest amount of information. The extent to which this information is useful, however, is unknown. Furthermore, more test administration time is probably needed under this instruction set than under the others. Asking test-takers for two judgments per item (e.g. most and least effective response) yields twice as much information as asking for one (e.g. most effective response). Conceptually, it seems appropriate to evaluate the test-taker’s knowledge of what actions to avoid in addition to the most effective responses (Hanson & Borman, 1993). Finally, Hanson and Borman (1993) found that a scoring system that subtracted the effectiveness values of the “least” responses from the effectiveness values of the “most” responses resulted in the highest levels of internal consistency reliability compared to several other scoring strategies. Differences between asking for likely responses (would do) versus asking for effective responses (should do) has generated the most discussion on the topic of SJT instructions (e.g. Weekley & Jones, 1999). Under most/least effective instructions, all respondents are answering based on their knowledge. Under most/least likely instructions, it is likely that only a small proportion of the respondents will answer according to what they would do when they believe that a different response is more effective. This will of course introduce error in the data. There are two general approaches to SJT scoring: rational and empirical. Rational answer keys are usually based on SME ratings of response option effectiveness. Empirical keys assign values to response options based on their

328

WALTER C. BORMAN ET AL.

relationships to a criterion. In two studies that directly compared scoring methods, correlations between SJT scores derived rationally and empirically were about 0.50 (Weekley & Jones, 1997) and 0.80 (MacLane, Barton, Holloway-Lundy & Nickels, 2001). In both studies, rational and empirical scores yielded equivalent relationships with performance criteria. At this time, there is no apparent advantage to either scoring method, but the relevant evidence is sparse and additional direct comparisons of the methods are still needed.

Research Findings Reliability The most commonly reported reliability estimate for SJTs is coefficient alpha, which varies widely across SJTs (McDaniel et al., 2001). SJTs are usually multidimensional by design, however, and should not be expected to have high internal consistency. Test-retest reliability is a more appropriate index (Motowidlo et al., 1990), but it is seldom studied. Data from two studies show test-retest reliabilities of 0.84 (Ployhart & Ryan, 2000) and 0.63 (Borman, Logan, Hedge, Hanson, Bruskiewicz, Schneider & Houston, 1996). Relationship With Other Predictors: General Cognitive Ability (or g) It seems reasonable to expect some association between cognitive ability and SJT scores. A recent meta-analytic estimate of the true correlation between written SJTs and g is 0.46 (McDaniel et al., 2001). McDaniel et al. also found wide variability in the SJT-g relationship across different SJTs. SJTs based on job analysis were more strongly related to cognitive ability ( = 0.50) than those that were not ( = 0.38). Also, SJTs containing shorter questions were more cognitively loaded ( = 0.56) than were those with longer questions ( = 0.47), indicating that reading skill may not account for the SJT-g relationship. (Note that video-based tests were not included in this meta-analysis.) Comparisons between written and video-based SJTs can offer more insight into the reasons why cognitive ability and written SJT performance are related. If reading requirements are largely responsible for the cognitive loading of written SJTs, then video-based SJTs should show lower correlations with g. Two studies that directly compared written and video formats of otherwise-identical tests found that g was moderately related to written SJT scores, but unrelated to video-based SJT scores (Chan & Schmitt, 1997; Swander, 2001). Other studies have reported correlations between video-based SJTs and g ranging from zero to 0.37 (Olson-Buchanan et al., 1998; Smiderle, Perry & Cronshaw, 1994; Weekley & Jones, 1997).

Current Directions and Issues in Personnel Selection and Classification

329

Taken together, the research evidence indicates that video-based SJTs tend to be less cognitively-loaded than their written counterparts, but that reading requirements do not completely account for the cognitive ability influence in SJTs. One study showed that an SJT mediated the g-performance relationship (Hanson & Borman, 1993). That is, g might facilitate learning how to handle particular work situations, which in turn leads to better job performance. More research is needed to understand the factors that drive the cognitive loading for SJTs. Relationship With Other Predictors: Personality McDaniel and Nguyen (2001) conducted a meta-analysis based on a limited number of studies relating SJTs to the Big Five personality traits. SJT scores were related to emotional stability (r = 0.32), conscientiousness (r = 0.26), and agreeableness (r = 0.25; rs corrected for sampling error only). Subsequent to this meta-analysis, four additional SJTs were found to be unrelated or weakly related to the Big Five (Chan, 2001; Clevenger, Pereira, Wiechmann, Schmitt & Harvey, 2001). These results, as well as the large variance displayed in McDaniel and Nguyen’s meta-analytic estimates, indicate that the relationship between SJT scores and personality constructs is not consistent. Two recent SJTs were explicitly designed to measure personality variables. Ployhart and Ryan (2000) targeted emotional stability, agreeableness, and conscientiousness in persons holding customer service jobs. SJT scores correlated about 0.20 (corrected for unreliability in all variables) with each target trait as measured by a traditional personality inventory. These relationships are smaller than those reported in McDaniel and Nguyen’s (2001) meta-analysis, which presumably included mostly SJTs that were not intentionally designed to measure personality constructs. As mentioned in a previous section, Pulakos, Arad, Plamondon and Kiechel (1996) investigated adaptive job performance in U.S. Army personnel. Based on careful consideration of the criterion domain and an extensive literature review, they identified fifteen personality constructs that were promising predictors of adaptive performance. The research team then developed personality measures and an SJT designed to assess ability to adapt to new and changing situations (Pulakos & Dorsey, 2000). The SJT targeted eight dimensions of adaptive performance, with consideration for the personality traits thought to be linked to those dimensions. For example, the trait sociability was linked to the criterion displaying cultural adaptability; thus, some situations presented cultural adaptation dilemmas requiring sociability to resolve. The traits most strongly related to the SJT were willingness to learn, achievement orientation, tolerance of ambiguity, and openness to experience, each with uncorrected correlations in the 0.20s.

330

WALTER C. BORMAN ET AL.

In sum, SJTs have small to moderate relationships with some of the Big Five personality factors, and the strength of these relationships varies across SJTs (McDaniel & Nguyen, 2001). Despite the fact that convergent validity evidence was weak, Ployhart and Ryan (2000) and Pulakos and Dorsey (2000) have contributed valuable early efforts targeting specific constructs with an SJT, and more work in this direction is needed.

Relationships with Job Experience It is reasonable to expect longer tenure and job knowledge to be associated with higher SJT scores. A recent meta-analysis, however, revealed a very small positive relationship between various measures of job experience and SJT scores (mean r = 0.05, corrected for sampling error only; McDaniel & Nguyen, 2001). Several subsequent studies also show low correlations between experience and SJT scores (Chan, 2001; Clevenger et al., 2001; Ployhart & Ryan, 2000; Weekley & Ployhart, 2002). The adaptability study described earlier did find that military rank correlated 0.30 with SJT scores (Pulakos & Dorsey, 2000). A likely interpretation of the generally low correlations between SJT scores and job experience is that time on the job, per se, may not be as important for enhancing job knowledge and jobrelated judgments as is what a job incumbent learns from this experience. Also, as McDaniel and Nguyen point out, there might be an asymptotic relationship between job experience and SJT scores, where notable gains in situational judgment are realized early on the job with subsequent experience contributing little to what the incumbent already knows about handling job-related situations. This will almost certainly depend on how complex and changeable the job is. What Do SJTs Measure? The persistent lack of agreement about what SJTs measure is not surprising in light of the fact that SJTs are most often designed to predict a broad criterion domain, not to reflect specific characteristics of individuals. Support is growing for the notion that the SJT is a measurement method rather than an indicator of a unitary construct, and as such, it might be used to assess various constructs (Chan & Schmitt, 1997; McDaniel et al., 2001; McDaniel & Nguyen, 2001; Motowidlo et al., 1990; Weekley & Jones, 1999). We agree that SJTs are best characterized as a measurement method and that what they measure is often multidimensional in nature. We can make some summary statements regarding the constructs most consistently related to SJTs, with the caveat that each of these relationships varies widely across different SJTs. Cognitive ability is the strongest correlate with SJTs (McDaniel et al., 2001),

Current Directions and Issues in Personnel Selection and Classification

331

but tends to be less prominent in video-based tests than in written tests (Chan & Schmitt, 1997; Swander, 2001). Notable non-cognitive components include emotional stability, conscientiousness, and agreeableness (McDaniel & Nguyen, 2001). Finally, SJTs typically have positive, but weak linear relationships with job experience (Chan, 2001; Clevenger et al., 2001; McDaniel & Nguyen, 2001; Ployhart & Ryan, 2000; Pulakos & Dorsey, 2000; Weekley & Ployhart, 2002). Future research should investigate possible nonlinear relationships between experience and situational judgment. Finally, recent attempts to measure specific personal characteristics with SJTs (Ployhart & Ryan, 2000; Pulakos & Dorsey, 2000), rather than simply measure their correlates after the fact should be continued.

Criterion-Related Validity A recent meta-analysis estimated the relationship between written SJTs and job performance to be 0.34 (corrected for sampling error and criterion unreliability; McDaniel et al., 2001). SJTs designed to target personality traits and SJTs administered via video, both excluded from this meta-analysis, demonstrate comparable levels of criterion-related validity (Ployhart & Ryan, 2000; Pulakos & Dorsey, 2000; Swander, 2001; Weekley & Jones, 1997). This evidence places the SJT among the best predictors in personnel selection, including general cognitive ability, conscientiousness, integrity, biodata, and assessment centers (Schmidt & Hunter, 1998). McDaniel et al.’s (2001) data were conclusive regarding one moderator: SJTs based on job analysis were more valid ( = 0.38) than SJTs that were not ( = 0.29). Sufficient data were available for independent assessment of three commercially available SJTs: Mean corrected validities for these ranged from 0.21 to 0.41. Results from the remaining moderator analyses suffered from small samples and were less conclusive. For example, the authors predicted that the general cognitive ability loading of SJTs would be positively associated with validity (based on the known association between g and job performance). SJTs with a high g-loading were indeed the most valid ( = 0.41), but SJTs with a medium g-loading were the least valid, and SJTs with a low g-loading were in-between. In addition to considering the relationship between SJTs and job performance, it is important to know whether SJTs are useful predictors over and above popular alternatives. One estimate is that an SJT plus a cognitive ability test yields = 0.40 for predicting job performance, compared to = 0.34 for the SJT alone or = 0.32 for g alone (McDaniel et al., 2001). This degree of incremental

332

WALTER C. BORMAN ET AL.

validity over g is comparable to that estimated for conscientiousness tests; slightly less than that estimated for integrity tests, work samples, and structured job interviews; and higher than that estimated for a wide variety of other predictors (Schmidt & Hunter, 1998). A few individual studies also provide tests of the incremental validity of SJTs, demonstrating SJTs’ incremental validity over combinations of some of the best performance predictors available, including cognitive ability, conscientiousness, job experience, and job knowledge (Chan, 2001; Clevenger et al., 2001; Weekley & Jones, 1997). It is important to remember that validities of SJTs and other predictors are affected by moderator variables, and thus to expect the degree of incremental validity offered by an SJT to vary across situations (McDaniel et al., 2001). One such situation might involve the nature of the performance criteria. For example, perhaps SJTs are differently related to citizenship versus task dimensions of job performance. Little research has addressed this question so far, but some data suggest that SJTs are good predictors of both citizenship and task performance (Chan, 2001; Pulakos & Schmitt, 1996). This issue is certainly worthy of further study.

Adverse Impact One of the reasons SJTs have become so popular is because of their potential to produce criterion-related validity comparable to that found for cognitive ability, while at the same time, producing smaller racial group differences. SJTs consistently produce smaller Black-White differences than do tests of g, but larger differences than do personality tests (e.g. Clevenger et al., 2001; Hough, Oswald & Ployhart, 2001; Pulakos & Dorsey, 2000). Whereas the d-scores are near zero for personality tests and 1.0 or even higher for g tests, there is considerable variability across SJTs, although the differences are always less than for g tests. For example, Hough et al. found mean d = −0.61 favoring Whites over Blacks for written SJTs, compared to mean d = −0.43 for video-based SJTs. In one study that compared Black and White participants’ scores on written and video-based SJTs that were identical in their content (Chan & Schmitt, 1997), the Black-White difference was much larger for the written version (d = −0.95) than for the video-based version (d = −0.21). Also, reading comprehension test scores were positively related to written SJT scores, but not related to video-based SJT scores. Finally, Black participants thought that the video-based test was substantially more job related than the written test, whereas White participants thought that both tests were highly face valid. Thus, both reading comprehension

Current Directions and Issues in Personnel Selection and Classification

333

and test perceptions might contribute to Black-White performance differences on written SJTs. Separate from the issue of group differences is the issue of differential prediction. That is, do SJTs have different relationships with criteria for different subgroups? One study of Black and White incumbents in three different organizations found no differential prediction for a written SJT (Clevenger et al., 2001). Using a video-based SJT, another study found no differential prediction based on race or sex (Weekley & Jones, 1997). In these two cases, SJTs showed no differential prediction despite the fact that they had moderate g loadings and did produce significant Black-White mean differences. More studies are needed on the topic of differential prediction before firm conclusions can be drawn. Turning to gender differences, women tend to score higher than men on both written (d = 0.26) and video-based (d = 0.19) SJTs (Hough et al., 2001). In the Army’s Project A, women scored higher than men (d = 0.33) on an SJT designed as a criterion measure of supervisory judgment and decision making (Hanson & Borman, 1993). Two other studies not included in Hough et al.’s summary, however, found no significant gender differences in SJT scores (Olson-Buchanan et al., 1998; Pulakos & Dorsey, 2000).

Conclusions Despite all of the advantages described above, important questions about SJTs remain. Perhaps the most important note of caution is that relationships between SJT scores and other constructs (e.g. general cognitive ability, personality) vary considerably and thus it is not clear what SJTs are measuring. At this point, it seems most reasonable to conclude that SJTs are inherently multidimensional, that constructs measured vary across tests, and that the SJT should be viewed as a measurement method rather than a measure of a particular construct or constructs. In fact, their multidimensional nature is probably part of the explanation for why SJTs tend to be such good predictors of job performance, which itself is often multidimensional. Accepting these conclusions, however, means also accepting that SJTs may never fit neatly into a nomological net of psychological variables of interest in industrial and organizational psychology. Perhaps it is possible, however, to create a test that has the benefits of an SJT and measures an identifiable set of constructs. One approach involves pre-specifying target constructs for the test, and developing situations to target those constructs. Two such attempts provide a good starting point (Ployhart & Ryan, 2000; Pulakos & Dorsey, 2000). This approach could lead to a new, less fakable, more face valid, and more criterion-valid means of measuring personality traits.

334

WALTER C. BORMAN ET AL.

PERSON-ORGANIZATION FIT
Conventional selection practices are geared toward hiring employees whose KSAs provide the greatest fit with clearly defined requirements of specific jobs. The characteristics of the organization in which the jobs reside, or those characteristics of the person relative to the organization as a whole, are rarely considered. The basic notion with person-organization fit (P-O fit) is that a fit between personal attributes and characteristics of the target organization contributes to important individual and organizational outcomes. There is considerable confusion, however, about what P-O fit is. Although researchers tend to define P-O fit as compatibility between individuals and organizations, how compatibility is conceptualized, operationalized, and measured varies greatly. We now provide an overview of a few major themes in this topic area.

Conceptualization and Measurement of Fit Conceptualizations of Fit Muchinsky and Monahan (1987) suggested that “fit” can be conceptualized as either supplementary or complementary in nature. Supplementary fit is directed at finding individuals who possess characteristics similar to other individuals in a particular organization. This type of fit reflects vocational choice; people prefer to work in environments with others who are similar. Complementary fit is directed at filling gaps in the organization by finding individuals possessing the requisite KSAOs. Complementary fit based on person-job (P-J) match has traditionally driven employment selection practices. Both supplementary and complementary fit should be important in selection processes. Direct and Indirect Measures of Fit Researchers have studied direct judgments of perceived fit (e.g. interviewers’ perceptions of applicant fit), and indirect measures of actual fit (e.g. the fit between independent judgments of individual and organizational characteristics). Kristof (1996) suggested that direct measures are beneficial if fit is conceptualized as the judgment that a person fits well in an organization. Indirect measurement, on the other hand, is said to reflect actual fit because it allows independent verification without asking for input from the participants. Although there have been a variety of methods used to assess fit, in broad terms, the results have been discussed as investigations of the same construct. In truth, however, various conceptualizations of fit may differentially predict particular dependent variables. Consequently, several more recent studies have begun to

Current Directions and Issues in Personnel Selection and Classification

335

investigate the comparability of these multiple conceptualizations of fit (see, for example, Bretz & Judge, 1994). Indices of Fit In most P-O fit research, a profile comparison approach has been adopted for establishing fit, for example, using O’Reilly, Chatman and Caldwell’s (1991) Organizational Culture Profile. The Q-sort technique applied in these studies implies a holistic comparison of individuals and situations across multiple dimensions, rather than the comparison of persons and situations on specific dimensions. Difference scores are then used to contrast attributes of individuals with attributes of the environment, and most typically, for each attribute the difference is calculated and squared. These squared differences are then summed; larger sums represent poorer fit. Edwards (1993), however, noted that profile similarity indices discard information on direction of misfit and rely on the assumption that each dimension of fit contributes equally to outcome measures. Edwards (1991) recommended instead, the use of a polynomial regression procedure because it simultaneously considers not only main effects of the individual and the environment, but also interaction effects as potential correlates of a criterion of interest. We now turn to a brief summary of relevant research findings on P-O fit, followed by some recent developments in the area, and thinking about future directions.

Research Findings Job Selection/Job Choice Much of the research on P-O fit and job choice has examined the match between personality variables and peoples’ preferences for organizations. For example, Bretz, Ash and Dreher (1989) found that students with high need for achievement were attracted to organizations that encouraged and rewarded competitive individual effort and accomplishment. Turban and Keon (1993) found that subjects with low self-esteem preferred larger firms with more centralized organizational structures, and subjects with high need for achievement preferred organizations that rewarded performance rather than seniority. Cable and Judge (1994) showed that materialism and self-efficacy significantly predicted preferences for organizations with high pay levels and individual-based pay. Judge and Cable (1997) examined the relationship between job seekers’ personality traits and their attraction toward different employers with whom they had interviewed. Results suggested that the Big Five personality constructs (extraversion, neuroticism, openness, agreeableness, conscientiousness) were

336

WALTER C. BORMAN ET AL.

related to dimensions of cultural preferences (innovative, detail-oriented, aggressive, outcome-oriented, supportive, decisive, and team-oriented). Work Attitudes and Outcomes Strong support has also been found for the positive effects of P-O fit on individual work attitudes. For example, Chatman (1991) found organizational fit to be a determinant of job satisfaction and organizational commitment, using new hires in an accounting firm. In addition, she found that socialization activities, such as mentorship programs and attendance at firm-related social events, were positively related to fit within one year of organizational entry. O’Reilly et al. (1991) found similar results related to satisfaction and commitment, and suggested P-O fit was a significant determinant of employee turnover within two years of the assessment. Vancouver and Schmitt (1991), in a study of teachers and principals found that both teacher-principal and teacher-teacher goal congruence were positively related to satisfaction and commitment. Bretz and Judge (1994) examined fit as a predictor of career success; and found indirect effects of P-O fit on job promotions, and to a lesser degree salary level, in addition to direct effects on organizational tenure and satisfaction. Finally, Goodman and Svyantek (1999) noted that employees’ perceptions of the organization’s “warmth” and “competence” were the strongest predictors of the employee’s positive citizenship and task performance. Research on Different Aspects of Fit Kristof-Brown (2000) conducted two studies to assess whether recruiters form distinguishable perceptions of applicant person-job (P-J) and person-organization (P-O) fit. The first study demonstrated that knowledge, skills, and abilities are relied on more frequently to assess P-J fit, and values and personality traits more often to assess P-O fit. Study 2 also found that both types of perceived fit offer unique prediction of hiring recommendations. Taken together, these results present evidence that recruiters discriminate between applicants’ P-J and P-O fit during early interviews. Hollenbeck et al. (2002) introduced the notion of internal fit and external fit, and suggested that for an organization to be successful, the organization’s structure needs to be aligned both internally (in terms of who performs the work) and externally (in terms of the environment in which the work takes place). Thus, structure becomes the central link between internal and external fit. They examined external and internal person-team fit, and found that when confronted with a poor fit between their divisional structure and their task environment, participants’ emotional stability turned out to be the most critical personal attribute. In other words, poor external fit can attenuate the otherwise positive effects of cognitive ability. In addition, high levels of emotional stability can also offset, to

Current Directions and Issues in Personnel Selection and Classification

337

some degree, the negative effects of being in a situation characterized as a poor external fit.

Schneider’s Attraction-Selection-Attrition Framework Much of the recent interest in the concept of P-O fit can be traced to the attraction – selection-attrition (ASA) framework proposed by Schneider (e.g. Schneider, 1987, 1989; Schneider, Smith & Goldstein, 2000). Schneider outlined a theoretical framework of organizational behavior based on the mechanism of person-environment fit that integrates both individual and organizational theories. It suggests that certain kinds of people are attracted to, and prefer, particular types of organizations; organizations formally and informally seek certain types of employees to join the organization; and attrition occurs when employees who do not fit a particular organization leave. Those people who stay with the organization, in turn, define the structure, processes, and culture of the organization. Schneider, Goldstein and Smith (1996) suggested that, although positive consequences of good fit include such outcomes as harmony, positive affect, effective socialization, and adjustment, good fit may lead to so much homogeneity that the organization becomes unable to adapt and change when the environment requires it. Thus, the organization may lack needed innovation, fail to challenge existing norms, and reflect “group think” at an organizational level. Responding to Schneider’s ASA notion that “people make the place,” Van Vianen (2000) argued that although many aspects of organizational life may be influenced by the attitudes and personality of employees in the organization, this does not necessarily require that the culture of a work setting originate in the characteristics of people. He suggested instead, that cultural dimensions that reflect the human side of organizational life are more adaptable to characteristics of people whereas cultural dimensions that reflect the production side of organizational life are more determined by organizational goals and the external environment. Van Vianen (2000) suggested that both the strength of organizational culture and the existence of subcultures are neglected issues in P-O fit research. Likewise, Schaubroeck, Ganster and Jones (1998) proposed that a more complex conceptualization of the ASA process that incorporates the distinction between occupational and organizational influences should be examined more closely. These researchers investigated the role of personality facets and P-O fit, and found that personality homogenization occurs differently and more strongly within particular occupational subgroups within an organization. Similarly, Haptonstahl and Buckley (2002) suggested that as work teams become more widely used in the corporate world, person-group fit becomes an increasingly relevant construct.

338

WALTER C. BORMAN ET AL.

In fact, Furnham (2001) argued that the fit concept should be viewed as dynamic and flexible because people both adapt to and change their jobs soon after they start them, and because of rapid changes in technology, jobs evolve and change over time. Furnham goes on to note that people can show dramatic changes in attitudes, beliefs, and work-related behavior because of the incentives and requirements of the job. In addition, most organizations attempt to mold employee behavior into some acceptable pattern. Individuals also change their jobs, without leaving them. They “personalize” different aspects of the job, and soon they are doing the job differently than did their predecessors. Finally, jobs themselves evolve and change. Organizational restructuring, the introduction of new technology, changing marketplace demands, etc., all force jobs to evolve fairly rapidly. So, over time, it may be that the degree of fit varies substantially as a function of the person, the organization, or job evolution. Thus, the concept of fit is itself far from clear or simple.

Toward an Expanded Model of Fit and a Broader Perspective of Selection The research and theorizing reported in this section has suggested that selection theory should consider making fit assessments based on person-job fit, personorganization fit, and person-group fit. Traditional selection theory considers person-job fit as the basis for selecting job applicants, with the primary predictor measures being KSAs, and the criterion targets being job proficiency and technical understanding of the job. Werbel and Gilliland (1999) have suggested that challenges to the predominant use of person-job fit in the selection process stem from new research and conceptualizing aimed at expanding both the predictor and the criterion domains. Borman and Motowidlo (1993) called for an expansion of the criterion domain and suggested that selection practices should be based on factors associated with both technical job performance and non-technical performance. Expansion of the predictor domain comes from activities such as: (1) global competition, employee empowerment, development of work teams, and increased use of technology; (2) an understanding that organizations possess certain cultures, visions, and values, and it may be important to select people who share the values and visions of the organization; and (3) new thinking about the nature of jobs, and a reliance on more fluid and rapidly changing role responsibilities. To include P-O fit as a component of the selection process, one would evaluate applicants’ needs, goals, and values. The assumption here is that the greater the match between the needs of the applicant and organizational reward systems, the greater the willingness to perform for the organization; and the greater the match

Current Directions and Issues in Personnel Selection and Classification

339

between a person’s goals and values and an organization’s expectations and culture, the greater the satisfaction and commitment. Consequently, the criteria associated with P-O fit should include organizational citizenship behaviors, organizational commitment, organizational satisfaction, and retention. Citizenship performance is important to organizations because higher levels of such activities facilitate the meeting of organizational goals and organizational performance, and help to shape the internal environment of the organization. Finally, at a more detailed level of fit, there is the expectation that suborganizational units such as groups may have different norms and values than the organization in which they are embedded. Thus, the degree of fit between an individual and group may differ significantly from the fit between the person and the organization. P-G fit has not received as much research attention as either P-J fit or P-O fit, but it is clearly different from these other types of fit (Kristof, 1996). Werbel and Gilliland (1999) suggested the following tenets about the three types of fit and employee selection: (1) the greater the technical job requirements, the greater the importance of P-J fit; (2) the more distinctive the organizational culture, the greater the need for P-O fit; (3) the lengthier the career ladder associated with an entry-level job, the greater the importance of P-O fit; (4) the more frequent the use of team-based systems within a work unit, the greater the importance of P-G fit; and (5) the greater work flexibility within the organization, the greater the importance of P-O fit and P-G fit.

Conclusions When jobs and tasks are constantly changing, the process of matching a person to some fixed job requirements becomes less relevant. Whereas traditional selection models focused primarily on P-J fit, several have argued that with new organizational structures and ways of functioning, individual-organizational fit and individual-group fit become more relevant concepts. In our judgment, selection models in the future should incorporate all three types of fit as is appropriate for the target job and organization. We can conceive of a hybrid selection model where two or even all three kinds of fit are considered simultaneously in making selection decisions. For example, consider the possibility of a special multiple hurdle application in which applicants must have above a level of fit for the initial job, the team they will first be assigned to, and the target organization in order to be hired. Or, depending on the particular selection context for an organization, the three different types of fit might be weighted differentially in selecting applicants. Obviously, the details of hybrid selection models such as these have yet to be worked out. However, the notion of using more than one fit

340

WALTER C. BORMAN ET AL.

concept seems to hold promise for a more flexible and sophisticated approach to making selection decisions.

OVERALL CONCLUSIONS
The world of work is in the midst of profound changes that will require new thinking about personnel selection. Workers will likely need to be more versatile, handle a wider variety of diverse and complex tasks, and have more sophisticated technological knowledge and skills. The aim of this paper was to provide a review of the state-of-the-science on criterion and predictor research and thinking, with an emphasis on recent developments that seem especially relevant. Accordingly, we completed a detailed review of predictor space issues, including predictors that tap such domains as ability, personality, vocational interests, and biodata. A similar review was undertaken for the criterion space, and included both technical and non-technical performance. In addition, other relevant criterion domains, such as adaptability and work commitment were examined as potentially important. These reviews also covered all available literature that provides empirical relationships between these predictor and criterion domains. Finally, we felt that it was important to expand this individual-level perspective to encompass broader organizational issues related to person-job match and person-organization fit. On the predictor side, advances in the last decade or so have shown that we can reliably measure personality, motivational, and interest facets of human behavior and that under certain conditions these can add substantially to our ability to predict turnover, retention, and job performance. The reemergence of personality and related volitional constructs as predictors is a positive sign, in that this trend should result in a more complete mapping of the KSAO requirements for jobs and organizations, beyond general cognitive ability. One particularly promising approach to measuring individual differences in the interpersonal and personality areas is the situational judgment test (SJT). These tests are based on the premise that there are important and often subtle differences between responding effectively and responding ineffectively to problems or dilemmas confronted in the course of carrying out job responsibilities. Further, such differences are reflected in responses to situations presented in written, video, or computer form. Recent research with SJTs offer hope that further work might lead to a less fakable, more criterion-valid measure of personality. In terms of criterion measurement, research has demonstrated that short-term, technical performance criteria, are best predicted by general cognitive ability, whereas longer term criteria such as non-technical job performance, retention, and

Current Directions and Issues in Personnel Selection and Classification

341

promotion rates are better predicted by other measures, including personality, vocational interest, and motivation constructs. In order to select and retain the best possible applicants, it would seem critical to understand, develop, and evaluate multiple measures of short-and long-term performance, as well as other indicators of organizational effectiveness such as turnover/retention. Our review suggests that organizations should explore in greater detail, criterion measurement domains such as citizenship performance, adaptability, and work commitment, as important dependent measures against which to validate predictors of long-term success. Finally, we would recommend, that organizations consider ways of expanding the predictor and criterion domains that result in selecting applicants with a greater chance of long-term career success, and when doing so, it will be important to extend the perspective to broader implementation issues that involve classification of personnel and person-organization (P-O) fit. As organizational flexibility in effectively utilizing employees increasingly becomes an issue, the P-O fit model may be more relevant compared to the traditional person-job match approach.

NOTE
1. Unless otherwise noted, represents true operational validity, an estimate corrected for sampling error, range restriction, and criterion unreliability.

REFERENCES
Aamodt, M. G., & Pierce, W. L., Jr. (1987). Comparison of the rare response and vertical percent methods for scoring the biographical information blank. Educational and Psychological Measurement, 47, 505–511. Abrahams, N. M., Alf, E. F., Keickhaefer, W. F., Pass, J. J., Cole, D. R., & Walton Paxton, E. (1994). Classification of test composites from the ASVAB, CAT-ASVAB, and ECAT batteries (Contract N66001-90-D-95 02, Delivery Order 7J16). San Diego: Navy Personnel Research and Development Center. Abrahams, N. M., Neumann, I., & Githens, W. H. (1971). Faking vocational interests: Simulated versus real-life motivation. Personnel Psychology, 24, 5–12. Alderton, D. L., & Larson, G. E. (1994). Dimensions of ability: Diminishing returns? In: M. Rumsey, C. Walker & J. Harris (Eds), Personnel Selection and Classification (pp. 137–144). Hillsdale, NJ: Erlbaum. Alf, E. R., & Abrahams, N. M. (1996). The impact of number of jobs and selection ratio on classification efficiency: An extended table of Brogden’s Allocation Average. Educational & Psychological Measurement, 56(6), 951–956. Allen, N. J., & Grisaffe, D. B. (2001). Employee commitment to the organization and customer reactions: Mapping the linkages. Human Resource Management Review, 11, 209–236.

342

WALTER C. BORMAN ET AL.

Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective, continuance and normative commitment to the organization. Journal of Occupational Psychology, 63, 1–18. Allen, T. D., & Rush, M. C. (1998). The effects of organizational citizenship behavior on performance judgments: A field study and a laboratory experiment. Journal of Applied Psychology, 83, 247–260. Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart, & Winston. Allworth, E., & Hesketh, B. (1999). Construct-oriented biodata: Capturing change-related and contextually relevant future performance. International Journal of Selection & Assessment, 7, 97–111. Asher, J. J. (1972). The biographical item: Can it be improved? Personnel Psychology, 25, 251– 269. Ashworth, S. D. (1989, April). The distinctions that Industrial/Organizational psychologists have between biodata and personality measurement are no longer meaningful. In: T. W. Mitchell (chair), Biodata vs. Personality: The Same or Different Classes of Individual Differences. Symposium presented at the 4th annual conference of the Society for Industrial and Organizational Psychology, Boston, MA. Barge, B. N., & Hough, L. M. (1988). Utility of biographical data for predicting job performance. In: L. M. Hough (Ed.), Literature Review: Utility of Temperament, Biodata, and Interest Assessment for Predicting Job Performance (pp. 91–130). Alexandria, VA: U.S. Army Research Institute for Behavioral and Social Sciences. Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimensions and job performance: A meta-analysis. Personnel Psychology, 44, 1–26. Barrick, M. R., Mount, M. K., & Judge, T. A. (2001). Personality and performance at the beginning of the new millennium: What do we know and where do we go next? International Journal of Selection and Assessment, 9, 52–69. Beck, K., & Wilson, C. (2001). Have we studied, should we study, and can we study the development of commitment? Methodological issues and the developmental study of work-related commitment. Human Resources Management Review, 11, 257–278. Becker, T. E. (1992). Foci and bases of commitment: Are they distinctions worth making? Academy of Management Journal, 35, 232–244. Benet-Mart´nez, V., & John, O. P. (1998). Los cinco grandes across cultures and ethnic groups: Multitrait ı multimethod analyses of the big five in Spanish and English. Journal of Personality and Social Psychology, 75, 729–750. Blake, R. J., & Sackett, S. A. (1993, August). Holland’s typology and the Five-Factor Model: A rationalempirical analysis. Paper presented at the 101st annual meeting of the American Psychological Association, Toronto. Blakley, B. R., Qui˜ ones, M. A., Crawford, M. S., & Jago, I. A. (1994). The validity of isometric n strength tests. Personnel Psychology, 47, 247–274. Block, J. (1995). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 117, 187–215. Bobko, P. (1994). Issues in operational selection and classification systems: Comments and commonalities. In: M. Rumsey, C. Walker & J. Harris (Eds), Personnel Selection and Classification (pp. 443–456). Hillsdale, NJ: Erlbaum. Borman, W. C., Buck, D. E., Hanson, M. A., Motowidlo, S. J., Stark, S., & Drasgow, F. (2001). An examination of the comparative reliability, validity, and accuracy of performance ratings made using computerized adaptive rating scales. Journal of Applied Psychology, 86, 965– 973.

Current Directions and Issues in Personnel Selection and Classification

343

Borman, W. C., Hanson, M. A., & Hedge, J. W. (1997). Personnel selection. In: J. T. Spence, J. M. Darley & D. J. Foss (Eds), Annual Review of Psychology (Vol. 48, pp. 299–337). Palo Alto, CA: Annual Review. Borman, W. C., Hanson, M. A., Oppler, S. H., Pulakos, E. D., & White, L. A. (1993). The role of early supervisory experience in supervisor performance. Journal of Applied Psychology, 78, 443–449. Borman, W. C., Kubisiak, U. C., & Schneider, R. J. (1999). Work styles. In: N. Peterson, M. Mumford, W. Borman, P. Jeanneret & E. Fleishman (Eds), The Occupation Information Network (O∗ NET) (pp. 213–226). Washington, DC: American Psychological Association. Borman, W. C., Logan, K. K., Hedge, J. W., Hanson, M. A., Bruskiewicz, K. T., Schneider, R. J., & Houston, J. S. (1996). Basic research evaluating the reliability of the Situational Test of Aircrew Response Styles and its ability, personality, and leadership correlates (Institute Report No. 282). Minneapolis, MN: Personnel Decisions Research Institutes. Borman, W. C., & Motowidlo, S. M. (1993). Expanding the criterion domain to include elements of contextual performance. In: N. Schmitt & W. C. Borman (Eds), Personnel Selection (pp. 71–98). San Francisco: Jossey-Bass. Borman, W. C., Motowidlo, S. J., & Hanser, L. M. (1983). A model of individual performance effectiveness: Thoughts about expanding the criterion space. Paper presented as part of symposium, Integrated Criterion Measurement for Large Scale Computerized Selection and Classification, the 91st Annual American Psychological Association Convention. Borman, W. C., Penner, L. A., Allen, T. D., & Motowidlo, S. J. (2001). Personality predictors of citizenship performance. International Journal of Selection and Assessment, 9, 52–69. Borman, W. C., Rosse, R. L., & Abrahams, N. M. (1980). An empirical construct validity approach to studying predictor-job performance links. Journal of Applied Psychology, 65, 662–671. Borman, W. C., White, L. A., & Dorsey, D. W. (1995). Effects of ratee task performance and interpersonal factors on supervisor and peer performance ratings. Journal of Applied Psychology, 80, 168–177. Borman, W. C., White, L. A., Pulakos, E. D., & Oppler, S. H. (1991). Models of supervisory job performance ratings. Journal of Applied Psychology, 76, 863–872. Boyatzis, R. E. (1982). The competent manager: A model for effective performance. New York: Wiley. Bretz, R. D., Ash, R. A., & Dreher, G. F. (1989). Do people make the place? An examination of the attraction-selection-attrition hypothesis. Personnel Psychology, 42(3), 561–581. Bretz, R. D., & Judge, T. A. (1994). The role of human resource systems in job applicant decision process. Journal of Management, 20, 531–551. Brief, A. P., & Motowidlo, S. J. (1986). Prosocial organizational behaviors. Academy of Management Review, 11, 710–725. Buss, A. H., & Plomin, R. (1975). A temperament theory of personality development. New York: Wiley. Cable, D. M., & Judge, T. A. (1994). Pay preferences and job search decisions: A person-organization fit perspective. Personnel Psychology, 47, 317–348. Campbell, D. P. (1971). Handbook for the strong vocational interest blank. Stanford, CA: Stanford University Press. Campbell, J. P. (1990). An overview of the army selection and classification project (Project A). Personnel Psychology, 43, 231–239. Campbell, J. P., Gasser, M. E., & Oswald, F. L. (1996). The substantive nature of job performance. In: K. R. Murphy (Ed.), Individual Differences and Behavior in Organizations (pp. 258–299). San Francisco: Jossey-Bass.

344

WALTER C. BORMAN ET AL.

Campbell, J. P., & Knapp, D. J. (2001). Exploring the limits in personnel selection and classification. Mahweh, NJ: Lawrence Erlbaum. Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager, C. E. (1993). A theory of performance. In: N. Schmitt & W. C. Borman (Eds), Personnel Selection in Organizations (pp. 35–70). San Francisco: Jossey-Bass. Carretta, T. R., & Ree, M. J. (1995). Air Force officer qualifying test validity for predicting pilot training performance. Journal of Business and Technology, 9(4), 379–388. Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytic studies. New York: Cambridge University Press. Carson, K. D., & Bedeian, A. G. (1994). Career commitment: Construction of a measure and examination of its psychometric properties. Journal of Vocational Behavior, 44, 237–262. Cattell, R. B. (1943). The description of personality: Basic traits resolved into clusters. Journal of Abnormal and Social Psychology, 38, 476–506. Chan, D. (2001). Practical intelligence and job performance. Paper presented at the 16th annual conference of the Society for Industrial and Organizational Psychology, San Diego, CA. Chan, D., & Schmitt, N. (1997). Video-based versus paper-and-pencil method of assessment in situational judgment tests: Subgroup differences in test performance and face validity perceptions. Journal of Applied Psychology, 82, 143–159. Chatman, J. (1991). Matching people and organizations: Selection and socialization in public accounting firms. Administrative Science Quarterly, 36, 459–484. Clevenger, J., Pereira, G. M., Wiechmann, D., Schmitt, N., & Harvey, V. S. (2001). Incremental validity of situational judgment tests. Journal of Applied Psychology, 86, 410–417. Coleman, V. I., & Borman, W. C. (2000). Investigating the underlying structure of the citizenship performance domain. Human Resource Management Review, 10, 25–44. Conley, J. J. (1984). Longitudinal consistency of adult personality: Self-reported psychological characteristics across 45 years. Journal of Personality and Social Psychology, 47, 1325– 1333. Conway, J. M. (1999). Distinguishing contextual performance from task performance for managerial jobs. Journal of Applied Psychology, 84, 3–13. Costa, P. T., Jr., & McCrae, R. R. (1988). From catalog to classification: Murray’s needs and the five-factor model. Journal of Personality and Social Psychology, 55, 258–265. Costa, P. T., Jr., & McCrae, R. R. (1995). Solid ground in the wetlands of personality: A reply to Block. Psychological Bulletin, 117, 216–220. Darley, J. B., & Hagenah, T. (1955). Vocational interest measurement: Theory and practice. Minneapolis: University of Minnesota Press. Dawis, R. V. (1980). Measuring interests. New Directions in Testing and Measurement, 7, 77–91. Dawis, R. V., & Lofquist, L. H. (1984). A psychological theory of work adjustment. Minneapolis: University of Minnesota Press. Day, D. V., & Silverman, S. B. (1989). Personality and job performance: Evidence of incremental validity. Personnel Psychology, 42, 25–36. Devlin, S. E., Abrahams, N. M., & Edwards, J. E. (1992). Empirical keying of biographical data: Cross-validity as a function of scaling procedure and sample size. Military Psychology, 4, 136–199. Digman, J. M., & Shmelyov, A. G. (1996). The structure of temperament and personality in Russian children. Journal of Personality and Social Psychology, 71, 341–351. Drasgow, F., & Olson-Buchanon, J. B. (Eds) (1999). Innovations in Computerized Assessment. Hillsdale, NJ: Erlbaum.

Current Directions and Issues in Personnel Selection and Classification

345

Dror, I. E., Kosslyn, S. M., & Waag, W. L. (1993). Visual-spatial abilities of pilots. Journal of Applied Psychology, 78(5), 763–773. Durhart, D. T. (2001). Violence in the workplace, 1993–1999. (DOJ Publication No. 190076). Washington, DC: U.S. Department of Justice. Dye, D. A., Reck, M., & Murphy, M. A. (1993). The validity of job knowledge measures. International Journal of Selection and Assessment, 1(3), 153–517. Eberhardt, B., & Muchinsky, P. (1982). Biodata determinants of vocational typology: An integration of two paradigms. Journal of Applied Psychology, 67, 714–727. Eberhardt, B. J., & Muchinsky, P. M. (1984). Structural validation of Holland’s hexagonal model: Vocational classification through the use of biodata. Journal of Applied Psychology, 69, 174–181. Edwards, J. R. (1991). Person-job fit: A conceptual integration, literature review, and methodological critique. International Review of Industrial and Organizational Psychology (Vol. 6, pp. 283–357). London: Wiley. Edwards, J. R. (1993). Problems with the use of profile similarity indices in the study of congruence in organizational research. Personnel Psychology, 46, 641–665. Edwards, J. E., & Morrison, R. F. (1994). Selecting and classifying future Naval officers: The paradox of greater specialization in broader areas. In: M. G. Rumsey, C. B. Walker & J. H. Harris (Eds), Personnel Selection and Classification (pp. 69–84). Hillsdale, NJ: Lawrence Erlbaum Associates. Erez, A., & Judge, T. A. (2001). Relationship of core self-evaluations to goal setting, motivation, and performance. Journal of Applied Psychology, 86, 1270–1279. Eysenck, H. J. (1991). Dimensions of personality: 15, 5, or 3? – Criteria for a taxonomic paradigm. Personality and Individual Differences, 12, 773–790. Farmer, W. L. (2001). A brief review of biodata history, research, and applications. Unpublished manuscript, Navy Personnel Research, Studies, & Technology. Fleishman, E. A., Constanza, D. P., & Marshall-Mies J. (1999). Abilities. In: N. G. Peterson, M. D. Mumford, W. C. Borman, R. P. Jeanneret & E. A. Fleishman (Eds), An Occupational Information System for the 21st Century: The Development of ONET. Washington, DC: American Psychological Association. Furnham, A. (2001). Vocational preference and P-O fit: Reflections on Holland’s theory of vocational choice. Applied Psychology, 50, 5–29. Gallagher, D. G., & McLean Parks, J. (2001). I pledge thee my troth . . . contingently: Commitment and the contingent work relationship. Human Resources Management Review, 11, 181–208. Goldberg, L. R. (1990). An alternative description of personality: The big-five factor structure. Journal of Personality and Social Psychology, 59, 1216–1229. Goldberg, L. R