Professional Documents
Culture Documents
Outcomes
Author(s): Hui Liao and Aichia Chuang
Source: The Academy of Management Journal, Vol. 47, No. 1 (Feb., 2004), pp. 41-58
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/20159559
Accessed: 16-02-2016 10:27 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/20159559?seq=1&cid=pdf-reference#references_tab_contents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management
Journal.
http://www.jstor.org
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
'
Academy ofManagement Journal
2?04, Vol. 47, No. 1, 41-58.
AICHIA CHUANG
National Taiwan University of Science and Technology
41
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
42 Academy of Management Journal February
factors at the other level are accounted for. Neither customer-driven employee performance" (1999:
would one know how factors across different levels 178).
interact with one another and jointly determine
service performance. Additionally, results obtained
at one level may not generalize to another level THEORETICAL BACKGROUND AND
without generation of specification errors (Kozlow HYPOTHESES
ski & Klein, 2000).
Employee Service Performance:
In fact, the only study that simultaneously exam
Conceptualization and a Multilevel Perspective
ined individual differences and contextual factors
provides an intriguing picture, indicating that per Employee performance, in general, refers to be
sonality traits are not related to employee customer haviors that are relevant to organizational goals and
service behavior once job characteristics are ac that are under the control of individual employees
counted for (Rogelberg, Barnes-Farrell, & Creamer, (Campbell, McCloy, Oppler, & Sager, 1993). In ser
1999). This study was limited in that it did not vice settings, customers have become an important
specify a conceptual framework for cross-level phe factor in how employee performance is defined
nomena; conceptualized job characteristics at the (Bowen & Waldman, 1999). Bowen and Schneider
individual level, thus measuring individual per (1988) noted three defining characteristics of
ception more than actual context; and contained service?intangibility, simultaneous production
hypothesis tests in which the hierarchical structure and consumption, and customer "coproduction"?
of the data was not considered. However, Rogelberg all of which imply that "the consumer experience
and colleagues' study did indicate that the whole is as important as, if not more important than, the
might not simply be the sum of its parts, thereby consumer good" (Bowen & Waldman, 1999: 164
underscoring the importance of examining the joint 165). Further, the quality of the interaction between
impact and the interactive effects of individual and employee and customer is critical in determining
situational factors. customer satisfaction. Therefore, the behavior of
The present study was an attempt to advance the employee plays an important role in shaping
knowledge in this area in several ways. First, it the customer's perception of service quality. Basing
bridged the macro and micro perspectives by de performance standards explicitly on customer ex
a multilevel framework and providing a pectations encourages employees' engagement in
veloping
more comprehensive picture of what kind of em behaviors that are particularly functional in achiev
ployees engage in good service performance and, at ing desirable customer outcomes (Bowen & Wald
the same time, what kind of organizational inter man, 1999). It is consistent with this customer
ventions facilitate service performance. Second, driven approach to employee performance that in
drawing on the theory of situational strength this study we defined employees' service perfor
(Mischel, 1977), we further integrated the two lev mance as their behaviors of serving and helping
els by investigating interactions across levels to see customers. Employee service performance hence is
whether the impact of individual on distinguished from service effectiveness, which re
personalities
service performance differed in different situations. fers to the results of service performance, such as
Finally, recognizing that organizations do not "per customer satisfaction and retention. Factors be
form" and that it is the individuals in an organiza yond employees' control influence variance in ef
tion who perform in ways that allow it to achieve fectiveness measures, but the behavioral measure
desirable customer outcomes (Kozlowski & Klein, of service performance we employed in this study
2000), we examined to what extent employee ser is less contaminated (Campbell et al., 1993).
vice performance, when aggregated to the store In what follows, we develop hypotheses regard
level, could explain between-stores differences in ing the antecedents and consequences of employee
observed customer outcomes. The current study is service performance. Implicit in the development
the first that we are aware of in which multilevel of our theoretical framework is the recognition that
were to employee ser an organization is an integrated and that
theory and method applied system
vice performance. Using hierarchical linear model individual and organizational characteristics inter
ing (HLM; Bryk & Raudenbush, 1992), we tested the act and combine to shape individual and organiza
proposed model using data on the employees, man tional outcomes (Kozlowski & Klein, 2000). The
agers, and customers of restaurants in a chain in the contribution of this multilevel perspective to or
U.S. Midwest. This study answers Bowen and ganizational science is twofold: Both top-down and
Waldman's call for research that pulls different bottom-up effects on organizational behavior are
sources of data together so that scholars can "better illuminated. A top-down approach establishes the
understand the requirements and consequences of need to conceptualize and assess organization, sub
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and 43
Chuang
unit, and group factors that can affect individual employed the "Big Five" personality traits to exam
attitudes, and behaviors. A bottom-up ine effects of personality on service performance
perceptions,
method, on the other hand, makes salient the pro for two reasons. First, convincing evidence of the
cesses that operate to reduce the variability of in validity of the Big Five taxonomy has accumulated
dividual perceptions and behaviors, thus facilitat over the last few decades across different theoreti
ing common interpretations of the emergence and cal frameworks, measures, occupations, cultures,
existence of collective phenomena (Kozlowski & and sources of ratings (De Raad & Doddema
Klein, 2000). Therefore, in addition to the individ Winsemius, 1999; John & Srivastava, 1999). Sec
ual differences factors that have been identified as ond, the use of the unifying Big Five taxonomy
important correlates of service performance in the instead of more specific personality traits facilitates
literature, we identified relevant contextual fea the accumulation of knowledge and comparison of
tures and expected that these factors would have findings across studies of personality. There were
top-down influences on employee service perfor theoretical and empirical reasons to expect that
mance via both a direct and a moderating effect. four of the Big Five personality dimensions would
Also consistent with this multilevel perspective be related to service performance. Two of the traits,
was our expectation that individual employees' conscientiousness and neuroticism, were expected
service performance would combine to form a col to be associated with performance in all jobs, and
lective phenomenon at the organizational level the other two traits, extraversion and agreeable
through bottom-up processes and would signifi ness, were expected to be particularly relevant
cantly relate to organizational effectiveness mea when performance involved interactions with
sures, including customer evaluation of service other people, as it does in a service context (Barrick
quality, customer satisfaction, and customer loy & Mount, 1991). In recent research, cognitive-moti
alty. In our multilevel theory building, we sought to vational work orientations have been proposed as
"connect the dots, making explicit the links be mediators between these personalities and job per
tween constructs previously unlinked within the formance; such mediation would provide addi
organizational literature" (Klein, Tosi, & Cannella, tional theoretical support for personality-service
1999: 243). Figure 1 depicts the theoretical frame performance relationships (Barrick, Stewart, &
work of this study. Piotrowski, 2002).
Conscientiousness. Conscientious individuals
are described as responsible, organ
Individual-Level Antecedents of Service dependable,
ized, hardworking, and achievement-oriented (Bar
Performance: Personalities
rick &Mount, 1991). Because of these positive char
Certain employees may be predisposed to engage acteristics, conscientious people tend to do what is
in positive service-oriented behaviors. This study expected of them to accomplish work. Gellatly
FIGURE 1
A Multilevel Model of Service Performance
Agreeableness
Bottom-up
Process H4
Service Climate
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
44 Academy of Management Journal February
(1996) showed that conscientiousness related to ship between extraversion and job performance for
performance through expectancy and goal choice. sales representatives. Research has shown a posi
Barrick and colleagues (2002) also argued that con tive relationship between extraversion and the job
scientious individuals have higher intentions for performance of groups in occupations involving
achievement striving, which mediates the relation social interactions (Barrick & Mount, 1991, 1993).
ship between conscientiousness and job perfor Since most of the tasks of service employees con
mance. Indeed, conscientiousness has been found tain interactions with customers, we expected to
to positively associate with job performance in all observe that employees higher on extraversion
occupational groups tested in meta-analyses (Bar would demonstrate better service performance.
rick & Mount, 1991, 1993). Additionally, results of
Hypothesis lc. Individual-level extraversion
Frei and McDaniePs (1998) meta-analysis revealed
will be positively related to employee service
that conscientiousness was positively and strongly
related to customer service orientation, a personal performance.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and 45
Chuang
& West, 1998), safety climate formance. Our review of the literature on high
(e.g., Anderson (e.g.,
Hofmann & Stetzer, 1996), and transfer of training performance HR practices indicated employee
climate (e.g., Tracey, Tannenbaum, & Kavanagh, involvement, training, and performance incen
1995), have explained significant variance in spe tives as the most relevant for employee perfor
cific behavioral outcomes. Climate determines how mance in service settings. These practices also
individuals behave by influencing how they think closely capture the "foundation issues" specified
and feel about certain aspects of their environment by Schneider and coauthors (1998) that provide
(Salancik & Pfeffer, 1978). In particular, employees the fundamental support employees require to
we
rely on cues from their surrounding work environ deliver service effectively. In what follows,
ments to interpret events, develop appropriate atti offer the theoretical rationales and empirical
tudes, and understand expectations concerning findings associated with each of these practices.
their behavior and its consequences (Salancik & Involving employees by granting them discretion
Pfeffer, 1978). For example, when there exists a and inviting them to participate in decision making
climate for safety, employees are more committed is one way organizations can improve service per
to safety, more can meet a wide
likely to comply with safety rules formance. Empowered employees
and regulations, and less likely to be involved in range of customer demands and are able to share
accidents (e.g., Hofmann & Stetzer, 1996). the information they collect about customer behav
The current study examines climate and perfor iors, thereby serving customers better and helping
mance in the context of customer service. Service improve service quality. Research has indicated
climate is defined as employees' shared percep that service quality and customer satisfaction were
tions of the policies, and procedures that enhanced when
employees were involved in
practices,
are rewarded, supported, and expected concerning problem-solving idea generation (Schneider, Park
customer services (Schneider et al., 1998). When ington, & Buxton, 1980) and in sharing customer
there is a climate for service, employees have come evaluations (Johnson, 1996). Batt (1999) also found
to understand that superior customer service is ex that service quality and sales were positively re
pected, desired, and rewarded; other things being lated to employee discretion and group self-regula
equal, they are more likely to provide good service. tion. Other research has shown that high-involve
Some empirical evidence supports a relationship ment work systems improved performance,
between service climate and employee service per reduced costs, and increased productivity (e.g., Ap
formance. For example, service climate has been pelbaum, Bailey, Berg, & Kalleberg, 2000; Huselid,
shown to influence store-level service quality (e.g., 1995; Ichniowski, Shaw, & Prennushi, 1997). Thus,
Johnson, 1996; Schneider et al., 1998). We argue
Hypothesis 2b. Store-level employee involve
that favorable store-level service quality cannot be
ment in decision making will be positively re
achieved without elevated service performance on
lated to employee service performance after
the parts of individualemployees inspired by the individual-level are controlled for.
shared service climate and that a relationship be personalities
tween service climate and individual employee ser It is also reasonable to postulate that service
vice performance will exist. Additionally, Borucki training will increase employee service knowledge
and Burke (1999) found a significant, positive rela and skills and consequently improve employee ser
tionship between service climate and employee vice performance. Bishop (1990) documented that
service performance aggregated to the store level. the increase in the productivity of newly hired
Thus, we propose: employees was associated with their participation
in company training programs. Bartel (1994) found
Hypothesis 2a. Store-level service climate will
a positive effect of training on employee productiv
be positively related to employee service per
ity. Additionally, a meta-analytic review revealed
formance after individual-level personalities that training and instruction practices had a posi
are controlled for. tive effect on output quantity and quality and
Human resource
practices.
Human resource cost effectiveness (Guzzo, Jette, & Katzell, 1985).
(HR) practices can play an important role in help Other research has studied a more direct link
ing employees achieve high-quality service. between training and service performance. Evi
These practices, on the one hand, provide em dence showed that new-employee formal training
ployees with the skills, resources, and discretion (Schneider & Bowen, 1985), general service train
they need to meet customer demands, making ing (Johnson, 1996), and gaining knowledge about
them able to deliver high-quality service. On the an organization's environment and about service
other hand, these practices may motivate em (Schneider et al., 1980) were helpful in achieving
ployees to be more willing to provide good per quality service and customer satisfaction. Batt
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
46 Academy of Management Journal February
(1999) also showed that the more training employ situations leave the person more discretion in de
ees were offered, the better was the service quality. termining which behavior to undertake?individ
On the basis of these previous research findings, we ual differences in personality are more to
likely
pose the following hypothesis: influence behavior in weak situations than in
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and 47
Chuang
Store-Level Service Performance and Customer egy but encourages the franchisees to maintain
Outcomes their individuality and grants them a large degree
of latitude concerning everyday management mat
Effectiveness is the bottom line of any organiza
ters, such as hiring, training, the degree of employ
tion. For a service customers' percep
organization, ees' involvement in decision making, incentive de
tions of service quality, customer satisfaction, and
customer are crucial indicators of effective signs, and so on. We sent survey packages to these
loyalty
ness because of their close relationship with sales
52 restaurants. Each package contained copies of an
and profits, as is evidenced in the liter employee questionnaire (equal to the number of
marketing in each restaurant, which was 25 on the
ature et al., 1998). We expected that employees
(see Schneider 3 copies of a manager 150
individual service performance, average), questionnaire,
superior employee of a customer and return en
when to the store level, would contrib copies questionnaire,
aggregated To ensure the of re
ute to achieving desirable customer outcomes. We velopes. anonymity employee
sponses, we instructed the managers to designate
examined the impact of store-level performance in
an employee representative to collect sealed enve
stead of individual performance on
employee's or set up a central collection
customer outcomes for two reasons. First, most ser lopes from employees,
box where employees could drop off their enve
vice encounters experienced by customers involve
lopes. Employees were also provided the option of
their interactions with and contributions from mul
sending their responses directly to the researchers.
tiple service employees. For example, in the cur
We received 52 manager surveys and 351 em
rent study, a customer's evaluation of his or her
ployee surveys from 30 locations, a number repre
dining experience was determined by the service
senting approximate response rates of 58 percent
performance of the hostess, the "busperson," the
for the restaurants, 56 percent for the managers,
server, the cook, the cashier, and so on. Thus, the
and 46 percent for the employees. We also received
in a store work together as a team to
employees 2,167 customer but were unable to calcu
for a cus surveys
create satisfactory service performance late a response rate, because we did not know how
tomer, and it is the overall level of employee ser
many customers had actually been approached. Fif
vice performance, not the performance of any teen dishwashers who did not speak English and
particular employee, that determines customer out
filled out the Spanish version of the questionnaire
comes. Second, the attraction-selection-attrition
were excluded from the analyses owing to their
(Schneider, 1975), socialization, and social infor
lack of interaction with customers. Thirty-seven
mation processing and learning processes that may
employees whose tenure was less than one month
operate in a store, as well as its shared organiza
were also eliminated from the analyses owing to
tional environment, tend to result in relatively
their lacking sufficient knowledge to provide accu
behaviors and performance across
homogenous rate evaluations of restaurant and proce
within the same a policies
employees store; therefore,
dures. "Listwise" of cases with missing
deletion
store-level service performance will emerge via
values on variables reduced
further the employee
bottom-up processes from individual employee
sample size to 264. Finally, we also excluded 7
and exist as a collective
performance phenomenon. two restaurants from the
employees from analyses,
Borucki and Burke (1999) provided empirical evi
because estimating the models of interest required
dence that aggregated employee service perfor at least 5 respondents per restaurant. The final us
mance customer outcomes. Thus, we
predicted able thus consisted of 257 employees, 44
sample
propose: and customers from 25 franchised
managers, 1,993
4. Store-level service performance restaurants, with the number of employees per res
Hypothesis
will be positively related to customer evalua taurant ranging from 5 to 21 (x = 10.3), the number
tion of service quality, customer of managers ranging from 1 to 3 (x = 1.8), and
satisfaction,
and customer the number of customers ranging from 3 to 147
loyalty. =
(x 81.8).
Eighty-nine percent of the employees in the final
METHODS sample were Caucasian, 31 percent were male, and
45 percent worked full-time; the average age was 26
Participants and Procedures
years old, and the average tenure was 35 months.
Fifty-two stores of a family franchise restaurant To lessen concern about possible sampling bias,
chain operating in several states in the U.S. Mid we first compared sample means for the usable
west were invited to participate in the study. The cases and the cases
dropped on the basis of incom
franchiser designs signature menu items and pro plete information on all relevant variables in the
vides a centralized purchasing and marketing strat employee, manager, and customer samples. Results
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
48 Academy of Management Journal February
of ?-tests indicated the two groups were not statis prepared" and "Imake a mess of things" (reverse
tically significantly different from each other, ex coded) for conscientiousness; "I worry about
on means for the conscientiousness variable and "I mood a lot" for neuroti
cept things" change my
obtained from the employee sample and on means cism; "I start conversations" and "I don't talk a lot"
for the age variable obtained from the customer (reverse-coded) for extraversion; and "I am inter
sample. Specifically, employees with incomplete ested in people" and "I sympathize with others'
information had lower conscientiousness scores feelings" for agreeableness. The coefficient alphas
than those with information = were
complete [d 0.30, .77, .82, .85, and .81, respectively, for these
p < .05), and customers with incomplete responses scales.
were on the average older than those with complete Store-level antecedents. We constructed the
=
responses [d 5.07, p < .01). Further, we calcu store-level constructs by aggregating the individual
lated the binary correlations between the response employee or manager scores to the store level and
rates at restaurants and all variables specified in the testing the within-store agreement. Additionally, as
study for both the employee data and the manager Sirotnik (1980) suggested, we computed the inter
data and found none of the relationships was sta nal consistency reliability estimates for these vari
tistically significant at .05 level. Therefore, we con ables at the store level.
cluded that sampling bias should not be a problem. Employees were asked to rate, on the basis of
theirpersonal observation, their restaurant's cus
tomer service climate on a seven-item ser
Measures global
vice climate scale (Schneider et al., 1998; 1, "poor,"
Variables relevant to the current study as well as to 5, "excellent"; a = .95). An example item is
their corresponding sources of information are de "efforts to measure and track the quality of the
scribed below. We list in the Appendix the com work and service in your restaurant." Managers
plete scales for which we have obtained the per were asked to rate the level of employee involve
missions to reproduce the scale items, and below ment, or the extent to which their employees had
we provide example items for the other scales. influence over decisions at work on a five-item
Employee service performance. Employee ser scale (1, "not at all," 5, "a great deal") modified
vice performance was assessed using the sales per from Haynes, Wall, Bolden, Stride, and Rick (1999).
sonnel service performance measure from Borucki The scale items are listed in the Appendix. The
and Burke (1999). To adapt the measure, prior to coefficient alpha for this scale was .85.
the survey period we both consulted these authors Managers provided information concerning ser
and discussed specific items with restaurant man vice
training by rating the extent to which various
agers at a bimonthly chain gathering they attended. topics related to service performance were empha
We determined that 7 of the original 13 items could sized in the training or orientation of employees (1,
adequately capture the nature of restaurant service "not at all"; 2, "to a moderate extent"; 3, "to a great
performance and the domain of the construct at the extent"). This 13-item scale was based on Stevens,
same time. The Appendix lists these items. The Knutson, and Patton's (1995) DINESERV, a mea
employees were asked to rate their own perfor sure of restaurant service quality. We provide these
mance on an 11-point Likert-type scale with scale items in the Appendix. The scale coefficient alpha
anchors ranging from "completely unsatisfactory" was .91.
(1) to "extremely good" (11). The coefficient alpha Managers provided information about perfor
was .88 for this scale. mance incentives by answering three items we gen
Personalities. Conscientiousness, neuroticism, erated. A restaurant was considered to provide in
extraversion, and agreeableness were each mea centives for good service and coded with a 1 if they
sured by a ten-item scale from the International answered yes to the following: "Some monetary
Personality Item Pool (IPIP)developed by Goldberg rewards, not related to employees' regular pay, are
(1999). The average correlation between "domain provided (e.g., bonus or store coupon)," "Wages are
markers" for the Revised NEO Personality Inven tied directly to
employees' performance," and
tory (Costa & McCrae, 1992) and the corresponding "Good employees are promoted to a higher level
scales in the IPIP is .77, which rises to .94 when position." An internal consistency estimate was not
corrected for attenuation due to the unreliabilities relevant for this dummy-coded variable.
of both scales (Goldberg, 1999). Employees were Customer outcomes. We measured three cus
asked to rate how accurately each item described tomer outcome variables: customer evaluation of
them as they generally were on a five-point Likert service quality, customer satisfaction, and cus
type scale (1, "very inaccurate," 5, "very accurate"). tomer loyalty. Customer evaluation of service
Sample items include the following: "I am always quality was assessed via the 29-item DINESERV
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and Chuang 49
(Stevens et al., 1995). Customer satisfaction was using procedures described in Bryk and Rauden
measured with Gotlieb, Grewal, and Brown's bush (1992).
(1994) 3 customer satisfaction items, which were Finally, we examined to what extent store-level
service performance translated into desirable cus
adapted from Oliver (1980). Customer loyalty to the
restaurant the customer visited was as tomer outcomes. Since customers were nested in
particular
sessed with Webster and Sundaram's (1998) 5-item restaurants, we conducted HLM using in
analyses
customer scale. We the items of dividual-level customer-evaluated service quality,
loyalty provide
these scales in the Appendix. The scale anchors customer satisfaction, and customer loyalty as the
for the three scales ranged from 1, "strongly dis outcome variables. At 1, we
level controlled for
to 7, The coefficient al individual customers' age and gender. At level 2,
agree," "strongly agree."
customer we included the average employee service perfor
phas for service quality, satisfaction, and
customer were mance at the store level while controlling for the
loyalty .97, .96, .73, respectively.
level of local competition (assessed as the number
of competing restaurants within a ten-minute
drive).
Data Analysis
Because the key dependent variable of this study,
service performance, was measured via RESULTS
employee
employee self-reports, we assessed the construct of Measures
Validity
validity of this measure by examining its dimen
Construct validity of service performance. We
sionality, criterion-related validity, and discrimi
conducted the following analyses to demonstrate
nant validity. We then checked the viability of the
the validity of the service performance measure in
store-level constructs by examining the within
this data. We first examined the dimensionality of
& Wolf,
group agreement [rwg; James, Demaree, this measure a
by conducting principle compo
1984), intraclass correlation (ICC[1]), and reliability
nents factor analysis with "varimax" rotation and
of the mean (ICC[2]).
obtained a one-factor solution in which all the
The service performance model to be tested was =
items had high "loadings" (average loading .85)
hierarchical, with the dependent variable, em
on the 73 percent
service an individual single factor, which explained of
ployee performance, being the variance. We then examined the criterion-re
level construct, and the predicting variables
span lated validity of the service performance measure
ning the individual and store levels. The data were
by examining its relationship to other measures
also hierarchical, since employees were "nested" in
that should be theoretically related. As reported in
restaurants. We therefore adopted the hierarchical
Table 1, the pattern of correlations was consistent
linear modeling (HLM; Bryk & Raudenbush, 1992) with the aforementioned nomological network. At
method and tested the model in four steps. First,
the individual level, employee service performance
we estimated a null model that had no predictors at was significantly correlated with conscientious
either level 1 (the individual level) or level 2 (the = =
ness (r .33, p < .01), extraversion (r .26, p <
store level) to partition the service performance = <
.01), neuroticism (r -.21, p .01), and agreeable
variance into within- and between-stores compo =
ness (r .29, p < .01); at the store level, service
nents. Second, in a level 1 analysis, within each was significantly ser
performance correlated with
restaurant, service
performance
was
regressed
on
vice climate = < involve
(r .47, p .05), employee
individual-level of = =
grand-mean-centered predictors ment (r .50, p < .01), service training (r .50, p <
personality. A regression line was estimated for customer = <
.01), and satisfaction (r .42, p .05). As
each of the 25 stores in this step. In the third step, these relationships were consistent with
largely
or the level 2 analysis, we used the intercept esti theories about and empirical on service
evidence
mates obtained from level 1 as outcome variables and some of these measures were ob
performance,
and regressed these on the store-level predictors, tained from sources other than the stud
employees
including service climate and HR practices, to as ied here, coworkers, and
including supervisors,
sess the main effects of the store-level factors. In the customers, the results criterion-related
provided
last step, we regressed the slope estimates obtained evidence for the service
validity performance
from level 1 on the store-level factors to detect measure.
cross-level interaction effects. We also computed Further, we examined the discriminant validity
the proportion of variance in service performance of this measure its relationship with
by assessing
explained by individual-level factors lR2Within-store) theoretically unrelated variables gauged by the
as well as by store-level factors same source
(#2between-stores) (employees). No compelling theory or
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
50 Academy of Management Journal February
TABLE 1
Descriptive Statistics and Correlationsa
Individual-level, employee
1. Conscientiousness 3.74 0.56
2. Extraversion 3.49 0.71 .18**
3. Neuroticism 2.70 0.66 .33** .21*
4. Agreeableness 3.96 0.58 .55** .26* .18**
5. Employee service 8.31 1.75 .33** .26* .21** .29*
performance
Individual-level, customer
1. Age 44.46 17.16
2. Genderb 0.53 0.50 .01
3. Customer evaluation 5.77 0.85 i -i* * * .05*
of service quality
-i***
4. Customer satisfaction 6.09 1.03 1 .06* .69***
5. Customer loyalty 5.75 1.10 .19*** .06* .46*** .55**
Store-level
1. Service climate 3.28 0.43
2. Employee 3.72 0.71 .24
involvement
3. Service training 2.79 0.27 .29 .46*
4. Performance 0.80 0.41 .11 .33 .14
incentives
5. Store-level service 9.35 0.83 .47* .50** .50** .10
performance
6. Customer evaluation 5.77 0.24 .50** .24 .24 .05 .36+
of service quality
7. Customer satisfaction 6.10 0.31 .34+ .28 .21 .07 .42* .89**
8. Customer loyalty 5.77 0.29 .44* .09 .14 .36+ .34+ .81** .83*
a
n = 257; Customers n 1,993; Stores n = 25.
Employees
b
Coded as male, 1; female, 0.
+
p < .10
*
p < .05
< .01
**p
< .001
***p
Two-tailed tests.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and 51
Chuang
ally acceptable value of .70. We also obtained the tween stores, and 88 percent of the variance resided
following ICC(l) and ICC(2) values: employee within stores.
perceived service climate, .12 and .56; employee Individual-level
predictors only. Hypotheses la,
involvement, .63 and .70; service training, .50 and lb, lc, and Id predict that individual personalities
.58; performance incentive., .17 and .24; and store will be associated with individual employees' ser
level service performance, .12 and .56. All of these vice performance. We estimated a level 1 model
were comparable to the median or recommended including these variables, with no predictors spec
ICC values reported in the literature (see Schneider ified for the level 2 model. As a block, the person
et al., 1998). We thus concluded aggregation was ality variables explained 24 percent of the within
for these variables. store variance. conscientiousness =
justified Specifically, (y
=
.58, p < .001) and extraversion [y .37, p < .001)
had significantly positive relationships with em
HLM Results for the Antecedents of Employee
ployee service performance. Therefore, Hypotheses
Service Performance
la and lc were supported. Contrary to the predic
Null model. hypotheses Our predict that both tions of Hypotheses lb and Id, neuroticism and
individual- and
store-level variables would be sig agreeableness were not significantly related to em
ployee service performance. Thus, using HLM, we in which the personality variables were the level 1
estimated a null model in which no predictors were predictors and then regressed the intercept coeffi
specified for either the level 1 or level 2 function to cients obtained from level 1 on the measures of
test the
significance level of the level 2 residual store-level service climate and HR practices at level
= =
variance of the intercept (f00 .35, p < .001). The 2. As reported in Table 2, both service climate [y
=
ICC(l) was .12, indicating 12 percent of the vari .45, p < .01) and employee involvement (y .39,
ance in employee service performance resided be p < .05) demonstrated significant relationships
TABLE 2
Hierarchical Linear Modeling Results for Employee Service Performance0
Null Model
Variable Individual-Level Predictors Adding Group-Level Predictors
Level 1
Intercept 9.33(0.35***) 9.35 (0.35***) 5.66** (0.25***)
Conscientiousness 0.58*** (0.00) 0.51** (0.03)
Neuroticism -0.15 (0.14**) -0.08 (0.18**)
Extraversion 0.37*** (0.06) 0.39*** (0.04)
Agreeableness 0.23 (0.25) 0.32+ (0.19)
Level 2
Service climate 0.45**
a
n = 257, Stores n = 25. Entries are estimations of the fixed effects (ys) with robust standard errors. Estimations of the
Employees
random variance components (ts) are in parentheses. The ts for the also the between-stores variance in employee
intercepts represented
service performance.
b
Proportion of within-store variance explained by level 1 predictors.
c
Proportion of between-store variance explained by level 2 predictors (after level 1 variables are controlled for).
+
p < .10
*p < .05
< .01
**p
< .001
***p
One-tailed tests.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
52 Academy of Management Journal February
= =
with serviceperformance, after we had accounted satisfaction (7 .12, p < .01) and loyalty (7 .13,
for individual-level predictors. However, service p < .05). At the store level, the higher the level of
training and performance incentives did not have local competition was, the higher the customers'
significant relationships with service performance. ratings on service quality, satisfaction, and loyalty
As a group, the specified store-level variables ac = .002 for all < .01 for service
(7 outcomes; p
counted for 29 percent of the between-stores vari quality and customer satisfaction, and p < .05 for
ance in service performance. Hence, Hypotheses 2a customer loyalty). Store-level service performance
and 2b were supported, while Hypotheses 2c and and level of local competition explained 40 percent
2d were not. of the between-stores variance in service quality, 50
Testing cross-level interactions. Hypothesis 3 percent in customer satisfaction, and 50 percent in
posits that the store-level variables will moderate customer loyalty.
the relationship between personalities and individ
ual employees' service performance. A prerequisite
DISCUSSION
for testing these cross-level interactions was that
there be significant random variance for the person Prior research on service performance has been
ality variables in the intercepts-as-outcomes mod rather fragmented and has focused on either an
els estimated in the previous step. As reported in organization level or individual level of analysis.
Table 2, in which estimates of the random-variance This study bridged the gap between the macro and
components appear in parentheses, only neuroti micro We proposed
approaches. and tested a mul
- framework
of employee service performance
cism had significant random variance (f22 .18, tilevel
p < .01), suggesting significant variability in the and examined individual-level (that is, personality
level 1 neuroticism-service performance relation variables) and store-level (that is, service climate
ship across stores. We then examined whether this and HR practices) antecedents of service perfor
variance could be explained by store-level factors; mance as well as the impact of cross-level interac
none of these variables was significantly related to tions on service performance. We found that signif
the neuroticism slopes. Therefore, Hypothesis 3 icant variance in employee service performance
was not supported. existed both within and between stores and that
some of the individual factors (conscientiousness
and extraversion) and contextual factors (service
HLM Results for the Impact of Aggregated
climate and employee involvement) specified in
Service Performance on Customer Outcomes
this study explained a moderate amount of this
We further investigated whether employee ser variance. We also found significant between-store
vice performance aggregated to the store level was variance in customer outcomes and that employee
related to desirable customer outcomes. Three service
performance, when aggregated
to the store
HLM analyses were performed, with service qual level through bottom-up processes, contributed to
ity, customer satisfaction, and customer loyalty as the explanation of significant variance in customer
the dependent variables, individual customer age satisfaction and customer loyalty. The results pro
and gender as the level 1 variables, and aggregated vide a "deeper, richer portrait of organizational
service performance and level of local competition life?one that acknowledges the influence of the
as the level 2 variables. The results revealed signif organizational context on individuals' actions and
icant between-stores variance in customer evalua perceptions and the influence of individuals' ac
=
tion of service quality (f00 .03, p < .001), cus tions and perceptions on the organizational con
= <
tomer satisfaction (f00 .02, p .001), and text" (Klein et al., 1999: 243). The present study is
customer = < At the store thus a compelling extension of the previous
loyalty (f00 .02, p .001). ap
level, aggregated store service performance was sig proach to research on service quality, in which
related to customer satisfaction = is limited to micro-only or macro
nificantly (7 .07, investigation
=
p < .05) and customer loyalty (7 .06, p < .05), but only analysis, an approach that ignores influences
itwas insignificantly related to service quality = from other levels. our use of hierarchical
(7 Further,
.03, p > .05). Thus, Hypothesis 4 was partially linear modeling facilitated taking a multilevel ap
at us to investigate
upheld. Of the control variables, age, measured proach, which allowed the impact
the individual level, was significantly related to all of the predictors at different levels on the individ
of the three types of customer evaluations: the older ual-level service performance outcome while main
the customer, the higher the evaluations they gave taining the appropriate level of analysis for these
to the restaurants = < .001 for all out
(7 .01, p predictors (Hofmann, 1997).
comes). Female customers also rated service qual Some insignificant findings of this study are par
=
ity higher (7 .06, p < .05) and reported higher ticularly thought-provoking. Although they might
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and Chuang 53
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
54 Academy of Management Journal February
common method variance, we followed Podsakoff and that existing customers are an avenue to bring
and Organ's (1986) recommendations. In particu ing in new customers and a potential base for
lar, we operationally defined all of the store-level "cross-selling"1 (see Schneider et al., 1998). Thus,
constructs with measures from multiple raters and customer retention is critical for a service organi
used different sources of information, including the zation's survival and success. We found that better
managers, and customers. We also con employee service performance was associated with
employees,
ducted confirmatory factor analyses and found the higher customer satisfaction and increased cus
one-factor structure fitted the data poorly in all tomer loyalty, both of which
determine customer
three data sources. Further, if common method retention. Therefore, it pays for an organization to
variance were high, the pattern of relationships we emphasize high-quality service performance and
observed, which included some absences of rela enhance service performance among employees.
tionship, would be unlikely. Finally, most of the Our study also provides specific recommenda
relationships we saw were consistent with previ tions for improving employee service performance.
ous and theoretical work on service per First, fostering a service-oriented climate helps.
empirical
formance. Therefore, it seems that common method Employees do not work in a vacuum; their perfor
effects did not significantly influence the findings. mance is influenced by the messages management
Nonetheless, those conducting future research sends and by the perceptions employees share
should strive to obtain
service perfor among themselves. Organizations can use this
employee
mance evaluationsfrom multiple raters, including mechanism to guide and educate their employees
themselves, supervisors, peers, and cus as to how the organizations value excellent service
employees
tomers, as each of these sources of ratings can ex and to get policies and procedures implemented.
plain unique variance in performance measures When organizations demonstrate poor management
(Atwater, Ostroff, Yammarino, & Fleenor, 1998). of service, employees may feel that and start to
Drawing on existing research, for the present shirk on their duties. Second, our results suggest it
study we identified a set of key individual and pays to involve employees in decision making by,
contextual correlates of employee service perfor for instance, allowing employees to participate in
mance. However, there are likely to be other factors decisions that affect them and letting them resolve
that have an impact on service performance. For customer complaints on their own. This approach
individual ability and experience, em is consistent with the idea of employee empower
example,
ployees' emotional displays during service encoun ment used in total quality management (TQM). Re
ters, group demographic composition, leadership search on TQM has well documented the value of
style, and tip-sharing schemes among restaurant employees' opinions. When employees have a say
staff are additional individual- and store-level fac in how work is done, they assume responsibility
tors to be considered in the future. and return more effective work. Third, in their em
Another limitation of the current study concerns ployment selection procedures, managers may con
generalizability. While restricting our sider applicants' levels of conscientiousness and
potential
sample to a single occupation from the same organ extraversion, among other selection criteria, to im
ization ruled out superfluous factors associated prove customer service performance.
In sum, the
with different occupations and organizations, the results suggest that having the right employees,
of our results to other service or enforcing a positive service climate, and involving
generalizability
ganizations might be limited. However, the results employees in service management each adds incre
were largely consistent with prior theoretical and mental utility to the others. The fact that these three
work, that are not sam sources of influence were simultaneously signifi
empirical suggesting they
ple-specific. Nonetheless, future replication and cantly related to service performance indicated that
extension of this multilevel are war they didnot merely act as substitutes for each
investigation
ranted. this study employed a cross other, but functioned jointly to achieve superior
Finally,
sectional design, making causal inferences impos employee service performance.
sible. Future research should examine how these In conclusion, the research presented here con
over time. tributes to knowledge on service performance. This
relationships develop
is the first study to bring the micro and macro
perspectives together and to specify and test a
Managerial Implications
The of this study have substantial
results impli
cations for service organizations. It has long been 1
Cross-selling is the strategy of selling new products
recognized that it costs five to eight times more to to current customers on the basis of their previous pur
a one
acquire a new customer than to retain current chases.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and Chuang 55
multilevel model of the antecedents and conse Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager,
of service The C. E. A theory of performance. 1993. In N. Schmitt &
quences employee performance.
of putting ser W. C. Borman (Eds.), Personnel selection in organi
findings underscore the importance
zations: 35-70. San Francisco:
vice performance back into its organizational con Jossey-Bass.
text, which is inherently multilevel and integrated. Costa, P. T., & McCrae, R. R. 1992. Revised NEO Per
sonality Inventory (NEO PI-R) and NEO Five
Factor Inventory (NEO FFI) professional manual
REFERENCES Odessa, FL: Psychological Assessment Resources.
Beaty, J.C, Cleveland, J.N., &Murphy, K. R. 2001. The Psychology, 38: 275-291.
relation between personality and contextual perfor C. E., Wall,
Haynes, T. D., Bolden, R. I., Stride, C, & Rick,
mance in "strong" versus "weak" situations. Human
J. E. 1999. Measures of perceived work characteris
Performance, 14(2): 125-148. tics for health services research: Test of a measure
Bowen, D. E., Siehl, C, & Schneider, B. 1989. A frame tigation of factors influencing unsafe behaviors and
work for analyzing customer service orientations in accidents. Personnel Psychology, 49: 307-339.
manufacturing. Academy of Management Review, Huselid, M. A. 1995. The of human resource
impact
14: 75-95.
on turnover,
management practices productivity,
Bowen, D. E., & Waldman, D. A. 1999. Customer-driven and corporate financial performance. Academy of
employee performance. InD. A. Ilgen & E. D. Pulakos Management Journal, 38: 635-672.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
56 Academy of Management Journal February
James, L. R., Demaree, R. G., & Wolf, G. 1984. Estimating Salgado, J. F. 1997. The five factor model of personality
within-group interrater reliability with and without and job performance in the European Community.
response bias. Journal of Applied Psychology, 69: Journal of Applied Psychology, 82: 30-43.
85-98.
Schneider, B. 1975. climates: An essay.
Organizational
John, O. P., & Srivastava, S. 1999. The Big Five trait Personnel Psychology, 28: 447-480.
Lucas, R. E., Diener, E., Grob, A., Suh, E. M., & Shao, L. Sirotnik, K. A. 1980. Psychometric implications of the
2000. Cross-cultural evidence for the fundamental unit of analysis problem (with examples from the
features of extraversion. Journal of Personality and measurement of organizational climate). Journal of
79: 452-468. Educational Measurement, 17: 245-282.
Social Psychology,
Solomon, M. R., Surprenant, C, J. A., & Gutman,
Mischel, W. 1977. The interaction of person and situa Czepiel,
tion. In D. Magnusson & N. S. Endler Person E. G. 1985. A role theory perspective on dyadic in
(Eds.),
teractions: The service encounter.
ality at the crossroads: Current issues in interac Journal of Mar
333-352. 49(winter): 99-111.
tional psychology: Hillsdale, NJ: Erlbaum. keting,
M. M. & G. L. 1998. Stevens, P., Knutson, B., & Patton, M. 1995. DINESERV:
Mount, K., Barrick, R., Stewart,
Five-factor model of personality and performance in A tool for measuring service in restaurants.
quality
interactions. Human Cornell Hotel and Restaurant Administration
jobs involving interpersonal
Performance, 11: 145-165. Quarterly, 36(2): 56-60.
R. L. 1980. A model of the antecedents Tett, R. P., Jackson, D. N., & Rothstein, M. 1991. Person
Oliver, cognitive
and of satisfaction decisions. measures as of A
consequences Journal ality predictors job performance:
17: 460-469. meta-analytic review. Personnel 44:
of Marketing Research, Psychology,
703-742.
Podsakoff, P. M., & Organ, D. W. 1986. in
Self-reports
research: Problems and Tracey, J. B., Tannenbaum, S. I., Kavanagh, M. J. 1995.
organizational prospects.
Journal of Management, 12: 531-544. Applying trained skills on the job: The importance of
the work environment. Journal of Applied Psychol
Rogelberg, S. G., Barnes-Farrell, J. L., & Creamer, V. 1999.
ogy, 80: 239-252.
Customer service behavior: The interaction of ser
vice predisposition and job characteristics. Journal Vroom, V. H. 1964. Work and motivation. New York:
processing approach to job attitudes and task design. tion criticality in failure recovery. Journal of Busi
Administrative Science Quarterly, 23: 224-253. ness Research, 41: 153-159.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
2004 Liao and Chuang 57
Customer Satisfaction01
Performance Incentives 1. I am happy about my decision to come to this
1. Some monetary rewards, not related to restaurant.
employees' regular pay, are provided (for 2. I believe I did the right thing when I came to this
store coupon or a bonus). restaurant.
example,
2. Wages are tied directly to employees' 3. Overall, I am satisfied with the decision to come
performance. to this restaurant.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions
58 Academy of Management Journal February
Customer Loyalty6
Mr
1. Iwill recommend this restaurant to others.
2. I am sure that Iwill not visit this restaurant
again.
3. Iwill dine at another similar restaurant instead Hui Liao (huiliao@smlr.rutgers.edu) received her Ph.D.
of this particular
one. in human resources and industrial relations from the
4. I consider this restaurant to be University of Minnesota. She is an assistant professor of
reputable.
5. I definitely will not come to this restaurant again. human resource
management in the School of Manage
ment and Labor Relations at Rutgers, the State University
a
Adapted from Borucki and Burke (1999). Reproduced with of New Jersey. Her current research interests include
permission from the Journal of Organizational Behavior. service customer service workforce
b performance, quality,
from Wall, Bolden, Stride, and Rick and work
Adapted Haynes, diversity, job behaviors, group dynamics.
(1999). Reproduced with permission from the British Journal of
Health Psychology, published by the British Psychological So Aichia Chuang earned her Ph.D. at the University of
Minnesota; she is currently an assistant of hu
ciety. professor
c
Adapted from Stevens, Knutson, and Patton (1995). Repro man resource and behavior
management organizational
duced with permission from the Cornell Hotel and Bestaurant in the Department of Business Administration at the Na
Administration Quarterly. tional Taiwan University of Science and Technology. Her
d
Adapted from Gotlieb, Grewal, and Brown(1994) and
research interests include service climate and perfor
Oliver (1980). Reprinted with permission the Journal
from of
mance, fit, and assess
and from the Journal Be person-organization employee
Applied Psychology of Marketing
ment and selection.
search, published by the American Marketing Association,
Richard L. Oliver/Gilbert Churchill, November 1980, XVII,
page 463.
e
Adapted from Webster and Sundaram (1998). Reproduced
A
with permission from the Journal of Business Besearch.
This content downloaded from 14.139.210.68 on Tue, 16 Feb 2016 10:27:22 UTC
All use subject to JSTOR Terms and Conditions