Professional Documents
Culture Documents
Correspondence to: Brad Shuck, University of Louisville, CEHD, ELEOD, Louisville, KY 40292, Phone: 502-852-7396,
E-mail: brad.shuck@louisville.edu.
Human Resource Management, November–December 2017, Vol. 56, No. 6. Pp. 953–977
© 2016 Wiley Periodicals, Inc.
Published online in Wiley Online Library (wileyonlinelibrary.com).
DOI:10.1002/hrm.21811
954 HUMAN RESOURCE MANAGEMENT, NOVEMBER–DECEMBER 2017
T
he psychological construct of employee Accordingly, the purpose of our work was to
engagement has garnered mounting present the development, method, and results
interest in the human resource (Saks & of a 12-item, three-dimensional employee
Gruman, 2014; Sarti, 2014) and man- engagement measurement tool. As a first step in
agement fields (Christian, Garza, & responding to Albrecht (2010), Shuck and Reio
Slaughter, 2011). Engaged employees are believed (2011), and Shuck, Twyford, Reio, and Shuck
to bring their full selves into their work roles— (2014) provided a detailed theoretical framework
they are cognitively attentive, emotionally vested, for employee engagement aimed at both scholars
and physically energetic in their work environ- and practitioners. Their work was well grounded
ment (Kahn, 1990; Rich, LePine, & Crawford, in, and built upon, Kahn’s (1990) original con-
2010). Currently, there exists a range of theories ceptualization of personal engagement, modified to
(i.e., social exchange theory; Saks, 2006; Saks & reflect the more employee-focused experience of
Gruman, 2014) and models (the job demands– employee engagement (cf. Shuck & Reio, 2011;
resources model; Bakker & Demerouti, 2008) that Shuck et al., 2014). We drew heavily from Shuck
assist researchers in conceptualizing the phenom- et al.’s (2014) existing framework (and invite
enon of engagement. interested readers to those articles for further
In the 2010 Handbook of Employee Engagement, exploration and detail) in addition to the defini-
Albrecht posed the question, “So, how best to tion previously forwarded by Shuck & Wollard
measure engagement” (p. 9)? Over the past (2010). Items that made up the final version of
decade, several measures of engagement-like the employee engagement scale (EES) were theo-
states have been forwarded (i.e., retically grounded in Kahn’s original conceptu-
Merely having choice work engagement, job engagement, alization of personal engagement as well as the
organizational engagement; for a more employee-focused experience of employee
was insufficient; comprehensive listing, see Christian engagement offered by Shuck and Reio (2011)
et al., 2011). Scholars and practitio- and Shuck et al. (2014). The remainder of the
rather, having scale ners have choices when it comes article unfolds as follows: (1) defining and posi-
items with a clear to measuring the many variants tioning employee engagement, (2) distinguishing
of engagement constructs (Cole, employee engagement from similar constructs,
linkage to an agreed- Walter, Bedeian, & O’Boyle, 2011; (3) development and validation of scales, (4) dis-
Saks & Gruman, 2014). Albrecht cussion, and (5) limitations and implications.
upon definition would (2010) suggested, however, that
be particularly useful merely having choice was insuffi- Defining and Positioning Employee
cient; rather, having scale items with
Engagement
in both scholarship a clear linkage to an agreed-upon
definition would be particularly Employee engagement is defined as an active,
and practice. useful in both scholarship and prac- work-related positive psychological state (Nimon,
tice (also see Fletcher & Robinson, Shuck, & Zigarmi, 2016; Parker & Griffin, 2011;
2014; Purcell, 2014). While most Shuck, Nimon, & Zigarmi, 2016; Shuck et al.,
engagement-like measures are grounded in lit- 2014) operationalized by the intensity and direc-
erature, few scales that purport to measure some tion of cognitive, emotional, and behavioral
aspect of an engagement-like state also combine energy (Shuck & Wollard, 2010). In addition to
the robustness of a parallel definition and pre- employee engagement, several other engage-
ceding conceptual framework to accompany the ment-like constructs are noted in the research
scale in practice. Notwithstanding, while scales literature including work engagement, job engage-
such as the job engagement scale (JES; Rich et al., ment, organizational engagement, and intellec-
2010), the Utrecht work engagement scale (9- and tual/social engagement. For each construct, the
17-item versions; Schaufeli, Bakker, & Salanova, originating citation, definition, distinguishing
2006; Schaufeli, Salanova, Gonzalez-Romá, & unique focal points, and example scale items are
Bakker, 2002), and the intellectual, social, and provided in Table I. We urge scholars to note the
affective engagement scale (Soane, Truss, Alfes, distinguishing unique focal points for each state
Shantz, Rees, & Gatenby, 2012) are choices schol- of engagement as a means to differentiate both
ars may choose to utilize, no measure with clear theoretical and practical distinction. For exam-
linkage to an agreed-upon definition and concep- ple, work engagement is operationalized as the
tual framework has been proposed for the more opposite of exhaustion, cynicism, and professional
unique, broad-spectrum experience of employee efficacy (Schaufeli et al., 2002, 2006). Schaufeli
engagement, specifically—despite appeals from et al. (2002) defined work engagement as the
scholars to do so (Saks & Gruman, 2014). “positive, fulfilling, work-related state of mind
Emotional Intensity and willingness to invest emotionality Working at <my current organiza-
Engagement toward positive organizational outcome tion> has a great deal of personal
meaning to me.
I feel a strong sense of belonging to
my job.
characterized by vigor, dedication, and absorp- encompassing how people experience and respond
tion” (Schaufeli et al., 2002, p. 74). Theoretically, to their coworkers, is also focused on a host of
work engagement is grounded in the burnout other important areas (as previously noted) and
literature (e.g., work engagement is the positive remains decidedly unique in conceptualization.
antithesis of burnout; Shuck, 2011) and has been Employee engagement has been routinely
positioned as the antipode to the experience of defined and theoretically positioned as a unique
burnout (Schaufeli et al., 2002). The primary construct (Saks & Gruman, 2014; Shuck et al.,
focus of work engagement is work activity (i.e., 2014). Resultantly, employee engagement is not a
“When I get up in the morning I feel like going to substitute for work engagement, job engagement,
work”; Schaufeli et al., 2002). Employee engage- organizational engagement, or intellectual/social
ment is nonetheless markedly different in both engagement. Rather differences between terms,
focus and definition. Specifically and grounded their focus, and those nuances that impact con-
in Shuck et al. (2014), employee engagement is ceptualization and precise measurement should
reflective of an active psychological state and be documented and differentiated (See Table I).
inclusive of the full spectrum of the immediate Employee engagement is also not an umbrella term,
work experience (i.e., work, job, team, and the but a distinct term, which has its own unique the-
active experience of working). oretical framework and definitional content (cf.
The focus of job engagement, defined as a Saks & Gruman, 2014; Shuck et al., 2014, 2016).
“multi-dimensional motivational concept reflect- At present, the specificity of current mea-
ing the simultaneous investment of an indi- surement tools falls decidedly short for measur-
vidual’s physical, cognitive, and ing employee engagement (Albrecht, 2010; Saks
emotional energy in active, full & Gruman, 2014; Shuck et al., 2014). As Kahn
Employees bring a full work performance” (Rich et al., (1990) aptly noted, employees bring a full range
2010, p. 619), is job activity. In the of cognitive, emotional, and physical energies
range of cognitive,
JES (Rich et al., 2010), participants into their work roles that combine to distinguish
emotional, and are asked to respond to a series of the experience of being engaged. Connected,
questions—each with the postfix at researchers (Christian et al., 2011; Rurkkhun &
physical energies my job (e.g., I feel energetic at my Bartlett, 2012) have openly advocated opera-
job; see Table I). Thus, job engage- tionalizing the experience of employee engage-
into their work
ment describes the degree to which ment as a three-dimensional construct grounded
roles that combine a person is engaged with his or her through Kahn’s personal engagement framework.
job only, whereas employee engage- Investment of cognitive, affective, and behavioral
to distinguish the ment focuses toward the fuller expe- energies is understood as an indicator of indi-
rience of employees active roles vidual performance representative of motivation
experience of being
within the experience of their work, (and movement) as a distinct, work-related posi-
engaged. including their work, job, team, tive, psychological experience (Shuck et al., 2014).
and organization. Organizational The intensity of simultaneously directed energy
engagement, defined as “the extent proportionately indicates the extent to which
to which an individual is psychologically present an employee is engaged (e.g., employee engage-
in a particular organizational role” (Saks, 2006, p. ment). In the following, we briefly detail each
604), is focused toward organizational identifica- subdimension of employee engagement (and we
tion. Organizational engagement describes how direct interested readers toward more comprehen-
captivating and exhilarating a person experiences sive theoretical and conceptual reviews including
their organization (cf. Saks, 2006 [italics appear Saks & Gruman, 2014; Shuck, 2011; Shuck et al.,
in the original]) while employee engagement is 2014).
inclusive of an attachment-like state to the orga-
nization, but is not specifically defined by it—a
Cognitive Engagement
limitation of the organizational engagement scale The subdimension of cognitive engagement is
(OES; Saks, 2006). defined as the intensity of mental energy expressed
Similar to the JES (Rich et al., 2010) and OES toward positive organizational outcomes (Rich
(Saks, 2006), the more condition-oriented ISA con- et al., 2010; Shuck et al., 2014). Cognitively
struct (intellectual, social, and affective engage- engaged employees are attentive and concen-
ment; Soane et al., 2012) is focused toward social trated in their workplace and they dispense men-
engagement, an often overlooked and relational tally laden energy toward work-related activities.
component of the engagement experience (Kahn Building from Kahn’s (1990) early conceptualiza-
& Heaphy, 2014). Employee engagement, while tion, cognitive engagement is characterized by
the ways in which an employee directs cognitive Distinguishing Employee Engagement from
energy in both direction and proportion, char- Similar Constructs
acterized by an employee’s expression of focus
To distinguish employee engagement from simi-
and attention as well as concentration toward
lar constructs, we reviewed previous research that
work-related tasks, experiences, and contexts. A
had examined the conceptual overlap of engage-
cognitively engaged employee would be propor-
ment in relation to similar constructs—mainly,
tionately concentrated, focused, and attentive
distinctions between engagement and job satis-
toward work-related experiences (i.e., his or her
faction, engagement and organizational commit-
work, his or her job, or within the active role of
ment, and engagement and job involvement (cf.
working).
Shuck, Ghosh, Zigarmi, & Nimon, 2013, for addi-
Emotional Engagement tional information).
The subdimension of emotional engagement is
Employee Engagement and Job Satisfaction
defined as an employee’s intensity and willing-
ness to invest emotionality toward positive orga- Employee engagement and job satisfaction seem
nizational outcomes (Macey & Schneider, 2008; cut from the same cloth as the two are clearly
Shuck et al., 2014). Tangibly, emotional engage- linked in both practice and measurement. In
ment is the offering of emotionally connected, our view, however, employee engagement repre-
personal resources, such as believing in, feeling sents an active psychological state, whereas sat-
a sense of personal meaning toward, and being isfaction, a valuation of satiation “resulting from
emotionally connected, to a situation, person, the appraisal of one’s job or job experiences,”
or context within the full experience of work. is stationary (Locke, 1976, p. 1300). Certainly,
This subdimension demonstrates a deep, active, employee engagement is associated with job
and emotional connection to the active work- satisfaction, in that employees who are highly
ing experience. Emotionally engaged employees engaged are likely also to be satisfied. Job satisfac-
express affect that is directed toward a variety of tion, however, has often been focused at the job
work-focused targets that relate to the present level, consequently positioned as a temporal, gen-
momentary experience. For example, emotion- eral indicator of employee sentiment that is static
ally engaged employees would say they believe in rather than forward moving (Brief & Weiss, 2002).
the mission and purpose of their organization and That is, satisfaction may reflect an employee who
that the organization has a great deal of personal is satisfied in general but may not also be actively
meaning to them. motivated (Macey & Schneider, 2008). In distin-
guishing the two, scholars have juxtaposed con-
Behavioral Engagement ceptual and empirical distinctions between static
The subdimension of behavioral engagement is descriptions of satisfaction and active descriptions
defined as the psychological state of intention to of engagement (Cooper-Thomas, Leighton, Xu,
behave in a manner that positively affects per- Knight-Turvey, & Albrecht, 2010; Xanthopoulou,
formance (Macey & Schneider, 2008; Rich et al., Bakker, Demerouti, & Schaufeli, 2009) suggesting
2010). Tangibly, behaviorally engaged employees job satisfaction (when defined as satiation rather
are willing to put in extra effort, work harder for than confounding the construct with affect) is
their team and organization, and to do more than largely grounded in statements of the individual’s
is expected. Macey and Schneider (2008) referred sense of fulfillment with, and through, aspects of
to this state of engagement as proactive engage- the general work environment. This positions job
ment—because it was a forward moving, psycho- satisfaction as an indicator of fulfillment rather
logical state that manifested as behavior—but was than a motivational psychological state (Yalabik,
not yet materialized in action. In other words, Popaitoon, Chowne, & Rayton, 2013).
behavioral engagement represents a psychologi- Theoretically, job satisfaction may develop
cal state and is not yet action-related behavior under similar conditions as employee engagement.
(which differentiates it from performance, or In this regard, job satisfaction has been conceptu-
other related constructs such as organizational cit- alized in terms of a job attitude (Newman, Joseph,
izenship behavior, which is the actual behavior). Sparkman, & Carpenter, 2011) that can co-occur
It is not enough to just work harder; behaviorally within components of that attitude (intracom-
engaged employees see themselves as psychologi- ponent satisfaction) or alongside components of
cally willing to give more and often going above other attitudes such as job involvement, organi-
and beyond in a way that characterizes their for- zational commitment, and employee engagement
ward movement. (intercomponent satisfaction). Within the human
resource and management literature, however, the in some way toward the organization resultant
focus of job satisfaction has primarily served as an of attachment and is not an indicator of senti-
evaluative context, composed of conditional sum- ment toward the working experience itself, while
mary statements concerning the job or work (e.g., employee engagement describes an active motiva-
“I am satisfied with my work”). In other words, tional state encapsulating the full working expe-
satisfaction refers to a global, general, work-related rience. Conceptually, employee engagement may
perception focused toward satiation of the work be positively associated with higher levels of orga-
experience (Nimon et al., 2016; Shuck et al., 2013, nizational commitment, especially affective com-
2016). As noted previously, employee engagement mitment—a cycle that is likely to be self-fulfilling
is notably different as engagement manifests itself over time, but the constructs remain decidedly
as an active, motivational state contextualized by distinct in contextual positioning. We maintain
the psychological interpretation of the working that it is possible for organizational commitment
experience composed of meta-focused sources of and employee engagement to be mentalized dif-
cognitive, emotional, and behavioral energy (as ferently, yet practically co-occur simultaneously
described above). (cf. Shuck et al., 2016).
Employee Engagement and Organizational Employee Engagement and Job
Commitment Involvement
Organizational commitment may Job involvement and employee engagement likely
seem part and parcel to employee share similar conceptual space, yet the two con-
While engaged
engagement on the surface. In addi- structs focus toward different attentions in appli-
employees likely tion, organizational commitment cation. Job involvement for example is “the degree
has been routinely operational- to which a person is identified psychologically
identify with their ized as the far more affect-oriented with his work, or the importance of work in his
overlapping construct (Gruman & total self image” (Lodahl & Kejner, 1965, p. 24).
role in work, the
Saks, 2011; Guest, 2014) position- Consequently, job involvement is related to ego
limiting capacity ing employee engagement more involvement (Lawler & Hall, 1970; Vroom, 1962)
closely to affective organizational and positioned primarily as an identity-related
of job involvement commitment as opposed to norma- construct (Kanungo, 1982; Kühnel, Sonnentag,
tive or continuance commitment & Westman, 2009), whereas employee engage-
as a primarily
(although we note the potential ment is a motivational psychological state. This
cognitive function relation between all three forms of conceptual positioning is supported in theoretical
commitment). Researchers routinely propositions advocated by Shuck et al. (2013), and
demonstrates a suggest that engagement contains a research by Shuck et al. (2014 and 2016) suggest-
component of organizational com- ing job involvement and employee engagement
decided distinction
mitment embedded within the con- were tied primarily at the cognitive level, only. In
between the two. struct (Macey & Schneider, 2008; support, Macey and Schneider (2008) posited that
Shuck et al., 2013, 2016). When engagement and job involvement likely shared
experienced, engaged employees some level of construct overlap, citing definitions
are also likely to be organizationally committed that specifically mentioned facets of employee
employees. Scholars, however, have positioned engagement linked with job involvement. Scrima,
organizational commitment as “the relative Lorito, Parry, & Falgares (2013) further demon-
strength of an individual’s identification with strated this distinction empirically. In their work,
and involvement in a particular organization” job involvement and engagement (as well as
(Mowday, Steers, & Porter, 1979, p. 226). While affective commitment) were differentiated con-
often used in research as a focal perspective of ceptually in both focus and feature, and engage-
work (e.g., how committed an employee is), organi- ment fully mediated the relationship between
zational commitment is positioned as an outcome job involvement and affective commitment (cf.
of engagement, not a unique focal perspective of Scrima et al., 2013). Engagement, though, is not
the work experience (Shuck et al., 2014). only identity focused, but also a present-focused
The degree to which employees are orga- state looking toward the future encompassing
nizationally committed is representative of cognitive, emotional, and behavioral aspects.
their relative attachment to the organization. That is, while engaged employees likely identify
Organizational commitment is something that with their role in work, the limiting capacity of
happens to an employee (i.e., he or she becomes job involvement as a primarily cognitive function
affectively committed—e.g., attached). Thus, orga- demonstrates a decided distinction between the
nizational commitment represents a pledge to act two.
Development and Validation of Scales Criteria for Item Development and Support
As a first step in establishing a more comprehen-
for Item Inclusion
sive instrument, an initial research team made up First, a multiphase, content review process was
of two human resource professionals and research- employed to aid in the development of ini-
ers was used to generate items and establish con- tial items. Several databases were explored for
tent validity around an employee engagement content, to include the following: PsycInfo,
measurement tool. The first member was a human ABI/Inform, Proquest, Jstor, the Academy of
resource/organizational development scholar- Management database, and all four Academy of
practitioner at a large multinational hospitality Human Resource Development journals (Advances
company, and the second was a human resource in Human Resource Development, Human Resource
faculty member at a large, public, research- Development Review, Human Resource Development
intensive university. Both were located in the International, and Human Resource Development
southern portion of the United States. The research Quarterly). Additionally, Amazon.com and Google
team operationalized employee engagement in tri- Scholar were used as data collection sources for
part focus (as previously defined, e.g., cognitive, available scholarly books. The reviews of litera-
emotional, and behavioral), specifically as that ture helped each member of the team understand
energy an employee directed toward positive orga- the positionality and distinction of the employee
nizational outcomes during his or her experience engagement construct but did not
of work (i.e., see Shuck et al., 2014, for a compre- result in a list of items or previously
hensive overview). The EES was developed out of used scales that described the full The reviews of
the need to assess each dimension of the employee experience of employee engage-
literature helped
engagement experience as well as serve as a higher- ment, consistent with the detailed
order factor for the three lower-order factors. The theoretical grounding and defini- each member of the
sum of those dimensions was believed to frame the tional precision we sought. After
full experience of employee engagement. exhaustive and comprehensive team understand
The research team used the aforementioned reviews of existing literature and
the positionality
definition developed from the literature—specifi- following a similar protocol detailed
cally, that employee engagement was an active, by Nimon, Zigarmi, Houson, Witt, and distinction
work-related positive psychological state opera- and Diehl (2011), items were devel-
tionalized by the intensity and direction of cog- oped independently by each team of the employee
nitive, emotional, and behavioral energy—as a member to be included in the initial
engagement
starting point to assure content coverage of the version of the EES.
construct (Shuck et al., 2014; Shuck & Wollard, Second, post review, the construct.
2010). Care was taken to ensure that each item research team met in person to dis-
was consistent with the definition of the research cuss items generated, independent
variable (i.e., logical validity) in question and that dimensions, and total scale. This process yielded
the vocabulary was appropriate for the intended 45 items total, 15 for each dimension. Third, to
respondents. acquire another measure of logical validity, items
In a series of four studies, we first explored were pilot tested with a small sample of profes-
the factor structure of the EES and reliability of its sionals, faculty experts, and students (N = 12). The
scales (Study 1). After refining the instrument, in pilot group, which included seven women and
a new sample (Study 2), we confirmed the factor five men with a range of experience (i.e., scholars
structure and examined evidence of reliability and well established in their field, early career schol-
initial levels of nomological and convergent valid- ars, and practitioners), completed the 45-item
ity. Next (Study 3), we completed a final reduction measure that included a definition of employee
in scale items, confirmed the factor structure of engagement. Post pilot test, the research team
the tighter, more parsimonious scale, examined received verbal feedback on the scale and each
additional evidence of reliability and nomological item independently. Participants suggested that
validity, and examined evidence of discriminant many of the items overlapped, were redundant,
validity. In our final sample (Study 4), we tested or were too long. Participants also suggested the
for evidence of predictive incremental validity. We 45-items induced survey fatigue and that the
sought out different samples as a means to increase instrument was cumbersome.
external validity and demonstrate the behavior of Because possible overlap, divergence, and
the scale items across varying samples. In the fol- commonality were noted, to better ensure con-
lowing section, we detail the process used for item tent validity and concept coverage, rater agree-
generation and each of the four separate studies. ment for each item was subsequently evaluated
regarding whether each item was either essential, No significant issues were encountered with
useful but not essential, or not necessary (Lawshe, distribution.
1975). Under Lawshe’s system, a content validity
ratio (CVR) is used to generate a content valid- The EES
ity coefficient for each item. With less than eight The initial version of the EES contained 14 items:
panelists (we had two), for an item to be retained, 4 items for the cognitive engagement subscale, 5
there had to be a minimum CVR value of .99 (99 items for the emotional engagement subscale, and
percent interrater agreement) to ensure that agree- 5 items for the behavioral engagement subscale.
ment was unlikely due merely to chance. Content All items were measured on a 5-point Likert scale,
coverage and thus content validity related to the where 1 = strongly disagree and 5 = strongly agree.
development of new instruments has been shown
Analysis Procedure
to be viable with as few as two expert panelists
(Waltz, Strickland, & Lenz, 2005). This evaluation Following Nimon et al. (2011), exploratory fac-
process shortened the overall instrument, leav- tor analysis (EFA) was employed (Preacher &
ing 14 total items (4 for cognitive engagement, 5 MacCallum, 2003). We conducted the EFA using
for emotional engagement, and 5 for behavioral principal axis factoring because it is the “clas-
engagement). The initial 14 items were deployed sic factor analytic approach” (Pett, Lackey, &
in Study 1. Sullivan, 2003, p. 103). Prior research (e.g., Rich
et al., 2010; Wildermuth, Vaughan, & Christo-
Study 1 Baker, 2013) indicated the hypothesized factors
of cognitive engagement, emotional engagement,
Purpose and behavioral engagement were related; there-
The purpose of the first study was exploratory, fore, we used oblimin rotation, as it permitted
defining range and measurement reliability. Of correlations among factors. The measures of sam-
central importance was examining the underlying pling adequacy, specifically Barlett’s test of sphe-
psychometric structure of the 14-item EES using ricity (χ2 = 2125.55, df = 1, p < .001), the diagonals
exploratory factor analyses. of the anti-image correlation matrix (all > .80),
and the Kaiser-Meyer-Olkin measure of sampling
Method adequacy (.898) suggested the data were suitable
for EFA (Pett et al., 2003).
Participants and Procedure We considered several criteria to determine
The sample for Study 1 was recruited from a large, the number of factors to extract: the Kaiser-
diverse sampling of organizations of a regional Guttman rule of retaining as many factors as there
professional affiliation. The sample (N = 283) are eigenvalues greater than 1.0, the elbow of the
consisted of workers in the service (n = 97), man- scree plot, and parallel analysis with principal
ufacturing (n = 9), professional (n = 106), and non- components analysis (both the mean criterion
profit (n = 32) industries. Thirty-nine individuals and the 95th percentile criterion). All of these cri-
did not report their industry. Of the 283 partici- teria suggested three factors.
pants, 54 percent of the sample was female, and
most participants identified as Caucasian (n = Factor Analysis Results
134) or Hispanic (n = 91). Several indicated they Table II displays the pattern matrix, structure
worked as a frontline team member (n = 111), 27.2 matrix, and extraction communalities. We used
percent indicated being in a supervisory or man- the following criteria for retaining items: (1) pat-
agement capacity, and 18.4 percent were at an tern coefficient was at least .4 (a criterion fre-
executive level. Finally, the largest age group was quently used to determine salience of variables;
between 30 and 39 (n = 70), followed closely in Thompson, 2004), (2) pattern coefficients for non-
size by the 50–59 (n = 67), and 40–49 (n = 66) age relevant factors were less than .3, and (3) pattern
groups. Tenure and size of company varied across coefficients for nonrelevant factors were at least
the sample. .2 less than the pattern coefficient on the relevant
An Internet-based self-report survey was factor. Item 3 (“I am rarely distracted mentally
used as the data collection tool. Dillman, Smyth, when I am working”) did not meet the first cri-
and Christian’s (2009) four-stage method and terion (pattern coefficient = .382) or the third cri-
interval scheduling framework was used to pre- terion (pattern coefficient for nonrelevant factor
pare the survey for distribution. The survey was = .188). Item 4 (“Time passes quickly when I am
placed online for administration. Participation working”) did not meet the second criterion (pat-
was strictly voluntary. Each survey was assigned tern coefficient for nonrelevant factor = .399) or
a unique URL linked to the survey instrument. the third criterion (pattern coefficient for relevant
factor = .490). The pattern coefficients for all other and the simultaneous challenge of developing
criteria were met. cognitively grounded questions without the bias
As displayed in Table II, the proportion of vari- of affect (Nimon et al., 2011). Because the scores
ance in the items that was explained by the fac- on this scale did not exhibit adequate reliability or
tors (h2) was stronger for items loading on Factor adequate interitem correlations, new items were
1 (emotional engagement) than the other factors, generated.
with the items on Factor 3 (cognitive engagement) To refine the cognitive engagement measure,
having the lowest communalities (in addition to the original two-member research team revisited
the two problematic items). The items loading on findings from the initial review and researched
the three factors operationally defined the three databases for updated literature. Following the
intended constructs of emotional engagement (five protocol from Study 1, the researchers gener-
items), behavioral engagement (five items), and ated refined items independently, then met to
cognitive engagement (four items, two of which discuss overlap and divergence using the results
were problematic). These three factors explained from Study 1 to develop a more informed set of
58.93 percent of the common variance. The two items. Care was taken to develop items that were
scales that did not have problematic items, emo- grounded in the essence of cognitive engagement
tional engagement and behavioral engagement, (i.e., being attentive, focused, and concentrating)
had a moderate correlation of .49. as well as to include statements that emphasized
Cognitive engagement had a weaker the full experience of work and thinking about
The researchers correlation with the other scales (.32 how employees might direct their energy at work.
generated refined with emotional engagement and .29 Seven additional items were agreed upon unani-
with behavioral engagement). mously (Lawshe, 1975) that covered the cogni-
items independently, tive engagement construct and pilot tested with
a small sample (N = 9). The makeup of the pilot
then met to discuss Internal Consistency Reliability group included three women and six men with
Analyses for Study 1 and a range of professional and academic experience
overlap and Resulting Decisions (i.e., scholars well established in their field and
divergence using the We examined the internal consis- currently practicing human resource practitio-
tency reliability for each of the three ners). Post pilot test, the research team received
results from Study verbal feedback on the scale and each item inde-
factors. Cronbach’s alpha was above
1 to develop a more the acceptable level of .80 recom- pendently. No significant changes were advised.
mended by Clark and Watson (1995) The seven refined items were deployed in Study 2
informed set of items. for the emotional engagement scale alongside the emotional engagement and behav-
(α = .89) and the behavioral engage- ioral scales used in Study 1 (17 items total).
Care was taken to
ment scale (α = .92). However,
develop items that Cronbach’s alpha for the cogni- Study 2
tive engagement scale was below
were grounded in the an acceptable threshold (α = .63).
Purpose
Removing any single item would The purpose of Study 2 was to confirm the fac-
essence of cognitive tor structure and improve the overall model as
not increase alpha, and according
engagement. to the Spearman-Brown prophecy well as establish initial levels of nomological and
formula, we would need to add six convergent validity of the revised scale. Factor
comparable items to the scale (for and reliability analyses were conducted with the
a total of 10 items) to increase the reliability to intent to reduce the item set from 17 items to 15
the minimally acceptable level of .80—too many (5 items per scale). Nomological validity is estab-
for practical deployability. Although the aver- lished when scores on a given measure behave as
age interitem correlations for the emotional and expected with respect to other constructs to which
behavioral engagement scales were .62 and .72, it is theoretically related (Kelley, 1927; Shuck
respectively, the average interitem correlation for et al., 2016), while evidence of convergent valid-
the cognitive engagement scale was only .30, with ity is established when scores on a measure corre-
a range of .23 to .36. The standard deviation of the late with scores on another measure evaluating a
interitem correlations was .05, suggesting homo- similar construct. Nomological validity was tested
geneity of item variance and that the interitem by correlating EES scores to a measure of cogni-
correlations were all similarly weak. This could tive work appraisal developed for use in Study 2.
have been due to the complexity and latent struc- Convergent validity was tested by correlating EES
ture of assessing the phenomenon of cognition scores to the JES (Rich et al., 2010).
& Muthén, 1998), which allowed us to treat the model with the same model fit. The difference
data as categorical (using the WLSM-V estimation between the models was that rather than correla-
method), to test the CFA model with three latent tions among the three factors, we now had regres-
variables representing the three factors identified sion weights (path coefficients) for each factor
by the EFA: cognitive engagement, emotional measuring employee engagement. All three were
engagement, and behavioral engagement. The 17 statistically significant and strong (standardized
observed variables were the survey items, and we regression weights: emotional engagement = .68,
permitted the factors to correlate with each other. behavior engagement = .84, cognitive engage-
All the unstandardized regression weights (or path ment = .87).
coefficients) were statistically significant, and the Finally, based on factor loadings, overlap, and
standardized regression weights were all very expert review of our scale, we eliminated 2 of the 7
strong. The three factors were statistically signifi- cognitive engagement scale questions (questions
cantly, moderately, and positively related to one 2 and 3, respectfully). This brought the final scale
another (emotional and behavioral engagement: to 15 questions (5 total questions for each sub-
r = .57; emotional and cognitive engagement: r scale). See Table III for the standardized regression
= .59; behavioral and cognitive engagement: r weights for the final 15 items.
= .73). Both the comparative fit index (CFI, .98)
Internal Consistency Reliability Analyses for the
and the Tucker-Lewis index (TLI, .98) indicated
Final Full Instrument
good model fit (Preacher & MacCallum, 2003).
The mean chi-squared (χ2) was 422.40 with 116 All three of the final scales had strong internal
degrees of freedom (p < .001), although Yu (2002) consistency reliability. Alpha for the cognitive
found that this statistic has inflated Type I error engagement scale was .93, .91 for the emotional
rate with WLSM-V estimation. engagement scale, and .92 for the behavioral
We also tested a higher-order factor model engagement scale.
with employee engagement as the higher-order Scores on items in each scale were summed
latent construct. Because the model had three to obtain the scale score. When interpreting scale
lower-order factors previously correlated, this scores, it should be kept in mind that a 5-point
was a statistically equivalent model to the initial Likert scale was used; as such, the range of possible
TABLE III Final Standardized Regression Weights for Confirmatory Factor Analysis for Study 2
Cognitive Emotional Behavioral
Item Engagement Engagement Engagement
CE1. I am really focused on my job when I am working. .960 — —
CE2. I concentrate on my job when I am at work. .958 — —
CE3. When working, I think a lot about how I can give my best. .788 — —
CE4. At work, I am focused on my job. .962 — —
CE5. When I am at work, I give my job a lot of attention. .981 — —
EE1. Working at my current organization has a great deal of — .896 —
personal meaning to me.
EE2. I feel a strong sense of belonging to my job. — .943 —
EE3. I am proud to tell others that I work for my current — .890 —
organization.
EE4. I believe in the mission and purpose of my company. — .821 —
EE5. I care about the future of my company. — .942 —
BE1. I do more than is expected of me. — — .895
BE2. I really push myself to work beyond what is expected of me. — — .883
BE3. I am willing to put in extra effort without being asked. — — .972
BE4. I often go above what is expected of me to help my team be — — .928
successful.
BE5. I work harder than expected to help my company be — — .850
successful.
Note: Parameter estimates “fixed” to be 0 are reported as dashes (“—”).
TABLE IV Final Standardized Regression Weights for the CWAS-11 for Study 2
Item Estimate
CWAS1. The work I do on this job is important. .944
CWAS2. I feel safe at work. .642
CWAS3. The work I do is meaningful to me. .882
CWAS4. I have the resources I need to complete my work. .781
CWAS5. I feel a sense of responsibility to complete my work. .825
CWAS6. I feel supported by my supervisor. .771
CWAS7. The work I do makes a contribution to the organization I work for. .825
CWAS8. I enjoy doing my work. .877
CWAS9. I have the time I need to do my job well. .457
CWAS10. I feel supported by my coworkers. .577
CWAS11. My work has just the right amount of challenge. .545
scores for the cognitive, emotional, and behav- model fit was adequate, with a CFI of .95 and a TLI
ioral engagement scales (five items each) was 5 to of .94 (Preacher & MacCallum, 2003). The mean
25. On the cognitive engagement scale, the CFA χ2 was 247.68 with 44 degrees (p < .001), although
sample had a mean of 12.68 (SD = 3.63; range = this statistic has inflated Type I error rate with
5–25). On the emotional engagement scale, the WLSM-V estimation (Yu, 2002). The reliability of
CFA sample had a slightly lower mean score of 9.27 the CWAS-11 was strong (α = .87).
(SD = 3.37; range = 5–25), indicating moderate lev- When we ran a model in which employee
els of emotional engagement. On the behavioral engagement (the higher-order factor measured by
engagement scale, the CFA sample had a mean of our three scales) was regressed on
9.02 (SD = 3.06; range = 5–25), indicating similar CWAS (a factor measured by the 11
reporting patterns as emotional engagement. items on the scale), the standardized When we ran a model
regression weight was .78 (SE = .03;
Nomological and Convergent Validity p < .001), indicating that for every in which employee
To provide further validity evidence, we exam- 1-SD increase in cognitive work engagement (the
ined relationships between our scale scores and appraisal, employee engagement is
two other measures. First, we explored the CWAS- predicted to increase .78 SDs. This higher-order factor
11, which was developed for use in the second was theoretically and practically
study as a general indicator of the antecedents of consistent with previous research measured by our
employee engagement and used to test for nomo- between antecedents of engagement three scales) was
logical validity. After establishing acceptable psy- and the experienced psychologi-
chometric properties for the CWAS-11 (see below), cal state of employee engagement regressed on
in keeping with previous research findings (cf. (cf. Kahn, 1990; Wollard & Shuck,
Shuck, Adelson, & Reio, 2015; Wollard & Shuck, 2011), demonstrating initial levels CWAS (a factor
2011), we hypothesized that the CWAS-11 would of nomological validity. measured by the 11
predict employee engagement (the higher-order The JES was used as a paral-
factor that our scale measured). Second, to test lel measure of job engagement items on the scale),
for convergent validity, we hypothesized that our grounded in Kahn’s (1990) opera-
scales of cognitive, emotional, and behavioral tionalization of personal engage- the standardized
engagement from the EES would moderately cor- ment. The JES exhibited strong regression weight
relate with the cognitive, emotional, and physical internal consistency reliability is
engagement scales, respectively, of the JES (Rich each of its scales (cognitive engage- was .78 (SE = .03;
et al., 2010)—one of the more recent measures of ment: α = .93; emotional engage-
job engagement (e.g., Fletcher & Robinson, 2014; ment: α = .90; physical engagement: p < .001).
Shuck & Reio, 2013). α = .91). To determine whether our
Because we had developed the CWAS-11 for instrument demonstrated con-
this study, we examined construct validity by run- vergent validity, we attempted a CFA to exam-
ning a CFA using WLSM-V estimation to account ine the correlations among the scales in the two
for the categorical nature of the data. Table IV surveys. We first attempted to analyze a CFA in
provides the standardized regression weights. The which each of the six scales: cognitive, emotional,
TABLE V Correlations among the Six Scales from the Employee Engagement Scale (EES) and the Job
Engagement Scale (JES) (Study 2)
EES JES
CE EE BE CE EE PE
EES CE — .591 .727 .832 .800 .849
EE — — .573 .493 .721 .549
BE — — — .707 .695 .779
JES CE — — — — .770 .811
EE — — — — — .819
PE — — — — — —
Note: EE = emotional engagement; BE = behavioral engagement; CE = cognitive engagement; PE = physical engagement. All scales were
modeled as latent factors except the CE scale from the JES, which was modeled as an observed sum score due to a nonpositive definite
residual covariance matrix.
others). Roughly 74 percent (n = 788) of the sam- Avolio, Avey, and Norman (2007). A higher total
ple was female, and most participants identified as score on the JSS represented heightened levels of
Caucasian. The majority of participants indicated participants’ satisfaction with their job. All items
that they worked in some form of production or were measured on a 5-point Likert scale, where 1 =
service role (n = 333) followed closely by manager strongly disagree and 5 = strongly agree.
(n = 246) and administrative/clerical (n = 129). The
Discretionary Effort
largest age group was between age 50 and 59 (n =
242), followed by 40–49 (n = 233) and 30–39 (n = Discretionary effort was measured using a six-item
242). Tenure and size of work unit varied across discretionary effort scale (DES; Lloyd, 2008). A
the 16 independent work units. Similarly to Study higher total score on the DES represented a partic-
2, to ensure representation of the sample, we com- ipant’s intention to go above normal job-related
pared the research sample to the general popula- duties. All items were measured on a 5-point
tion of our study group. Gender, position within Likert scale, where 1 = strongly disagree and 5 =
the organization, and age was roughly equivalent strongly agree.
for the targeted population.
Psychological Well-Being
Using the same procedures followed in the
first two studies, Dillman et al.’s (2009) four-stage Psychological well-being was measured using the
method and interval scheduling framework was Schwartz outcome scale-10 (Blais et al., 1999). A
used to prepare the survey for distribution. The higher total score represented heightened levels
survey was placed online for administration. of participants’ general well-being. All items were
Participation was strictly voluntary, and partic- measured on a 5-point Likert scale, where 1 =
ipants were able to opt out of the study at any strongly disagree and 5 = strongly agree.
point. No significant issues were encountered
Intent to Turn Over
with distribution.
Following the same protocol as Study 2, based Intention to turn over was measured using the
on factor loadings, overlap, and expert review of three-item intention to turnover scale (ITS;
our scale, we eliminated one question from each Colarelli, 1984). A higher total score on the ITS
of the three subscales to reduce the overall scale represented heightened levels of intention to
to 12 items from 15 to deploy the most parsimo- leave one’s place of work. All items were measured
nious and psychometrically robust scale possible. on a 5-point Likert scale, where 1 = strongly dis-
This brought the final scale to 12 questions (4 agree and 5 = strongly agree.
total questions for each subscale).
CFA Models
The EES We tested the three-factor solution with the
Using factor and reliability analysis as well as reduced scales (four questions per scale) using
expert evaluation, we reduced the 15-item version Mplus version 7.1 (Muthén & Muthén, 1998)
of the EES from Study 2 to 12 final items to be and WLSM-V estimation. The CFA model had
deployed in Study 3. The final version of the scale three latent variables (cognitive, emotional, and
was composed of 4 items for each subscale (cog- behavioral engagement), which we permitted to
nitive engagement, emotional engagement, and correlate with each other. Both the CFI (.99) and
behavioral engagement). Slight semantic tweaks the TLI (.99) indicated good model fit (Preacher
were made in the cognitive engagement scale & MacCallum, 2003). The mean χ2 was 459.89
items for ease of readability. All items were mea- with 51 degrees of freedom (p < .001), although
sured on a 5-point Likert scale, where 1 = strongly Yu (2002) found that this statistic has an inflated
disagree and 5 = strongly agree. Type I error rate with WLSM-V estimation. All the
unstandardized regression weights (or path coeffi-
CWAS-11 cients) were statistically significant, and the stan-
The 11-item CWAS from Study 2 was used. Each dardized regression weights (Table VI) were all very
question within the scale was designed to tap into strong. The three factors were statistically signifi-
an antecedental dimension of employee engage- cantly, moderately, and positively related to one
ment. All items were measured on a 5-point Likert another (emotional and behavioral engagement:
scale, where 1 = strongly disagree and 5 = strongly r = .67; emotional and cognitive engagement: r
agree. = .59; behavioral and cognitive engagement: r =
.74).
Job Satisfaction As in Study 2, we also tested a higher-order
Job satisfaction was measured using a three-item factor model with employee engagement as the
job satisfaction scale (JSS) developed by Luthans, higher-order latent construct. Because the model
TABLE VI Final Standardized Regression Weights for Confirmatory Factor Analysis for Study 3
Emotional Behavioral Cognitive
Item Engagement Engagement Engagement
CE1. I am really focused when I am working. .948 — —
CE2. I concentrate on my job when I am at work. .963 — —
CE3. I give my job responsibility a lot of attention .994 — —
CE4. At work, I am focused on my job. .904 — —
EE1. Working at <my current organization> has a great deal of — .885 —
personal meaning to me.
EE2. I feel a strong sense of belonging to my job. — .863 —
EE3. I believe in the mission and purpose of <my company>. — .902 —
EE4. I care about the future of <my company>. — .942 —
BE1. I really push myself to work beyond what is expected of me. — — .932
BE2. I am willing to put in extra effort without being asked. — — .973
BE3. I often go above what is expected of me to help my team be — — .928
successful.
BE4. I work harder that expected to help <my company> be — — .868
successful.
Note: Parameter estimates “fixed” to be 0 are reported as dashes (“—”).
had three lower-order factors previously corre- Nomological and Discriminant Validity
lated, this was a statistically equivalent model to To provide further validity evidence for this short-
the initial model with the same model fit. The dif- ened format of the survey, we examined relation-
ference between the models was that rather than ships between our scale and several other measures.
correlations among the three factors, we once First, we again explored the CWAS-11. After again
again had regression weights (path coefficients) establishing acceptable psychometric properties
for each factor measuring employee engage- for the CWAS-11 within the sample for Study 3
ment. All three were statistically significant and (see below), in concert with findings detailed in
strong (standardized regression weights: cognitive Study 2, we hypothesized that the CWAS-11 would
engagement = .81, emotional engagement = .73, again predict employee engagement (the higher-
and behavior engagement = .92). order factor that our scales measured). We also
hypothesized that employee engagement would
Internal Consistency Reliability Analyses for the
be positively related to job satisfaction, discretion-
Final Full Instrument
ary effort, and well-being and negatively related
All three of the final subscales had strong to intent to turn over in keeping with longstand-
internal consistency reliability. Alpha for the ing, well documented research on engagement
emotional engagement scale was .88, for the (cf. Bhatnagar, 2012; Rich et al, 2010; Saks, 2006;
behavioral engagement scale was .91, and for the Shuck & Reio, 2013; Shuck, Zigarmi et al., 2015).
cognitive engagement scale was .94. Because the CWAS-11 had been developed
Scores on items in each scale were summed for use in Study 2 as an exploratory tool, we
to obtain the scale score. The range of possible examined construct validity for a second time
scores for the cognitive, emotional, and behav- by running a CFA using WLSM-V estimation to
ioral engagement scales (four items each) was 5 account for the categorical nature of the data.
to 20, as each scale had four items measured on The model fit was adequate, with a CFI of .94
a 5-point Likert scale. The sample for this study and a TLI of .93 (Preacher & MacCallum, 2003).
had high levels of engagement. On the cogni- The mean χ2 was 1229.19 with 44 degrees (p <
tive engagement scale, the mean was 17.94 (SD .001), although this statistic has inflated Type I
= 2.15; range = 4–20); on the emotional engage- error rate with WLSM-V estimation (Yu, 2002).
ment scale, the mean score was 16.92 (SD = 2.78; Additionally, the internal consistency for the
range = 4–20); and on the behavioral engagement CWAS was .91.
scale, the mean score was 17.84 (SD = 3.06; range When we ran a model in which employee
= 5–25). engagement (the higher-order factor measured by
TABLE VII Latent Construct Correlations between Employee Engagement Scale (EES) and Four Validity
Measures (Study 3)
Job Satisfaction Intent to Turn Over Discretionary Effort Well-Being
Engagement .77 -.54 .89 .62
Emotional Eng. .75 –.62 .55 .54
Cognitive Eng. .65 –.32 .68 .50
Behavioral Eng. .52 –.29 .89 .51
Note: All correlations significant at p < .001. “Engagement” was a higher-order factor measured by the three engagement subscales.
our three scales) was regressed on CWAS (a factor misfit than Model 1, evidence would indicate
measured by the 11 items on the scale), the stan- that the deployed measures were not correlated,
dardized regression weight was 0.83 (SE = 0.01; p whereas if Model 1 were to have less model-data
< .001), indicating that for every 1-SD increase in misfit than Model 1, evidence would indicate that,
cognitive work appraisal, engagement is predicted as hypothesized, those measures are correlated.
to increase 0.83 SDs, which was very similar to the Finally, to test whether employee engagement was
findings in Study 2. a distinct construct from the other
Next, guided by research, we correlated factors in the model (i.e., CWAS,
employee engagement as a higher-order factor job satisfaction, discretionary effort, We correlated
measured by our three scales and each individual psychological well-being, and inten-
employee
scale with job satisfaction, intent to turn over, tion to turn over), we compared
discretionary effort, and well-being, treating each our hypothesized model (Model 1) engagement as a
construct as a latent variable. Table VII provides with Model 3, which introduced
correlations for each of the constructs and our one higher-order factor measured by higher-order factor
measure of employee engagement. As hypoth- all the constructs. For both model
measured by our
esized, employee engagement had a moderate to comparisons, we used the DIFFTEST
large positive correlation with job satisfaction, option in Mplus. Results suggested three scales and
discretionary effort, and well-being (r = .77, .89, that employee engagement was in
and .62, respectively). As might be expected, the fact correlated with the other factors each individual scale
behavioral engagement subscale had the greatest (χ2 correlated w, p < .001, favoring
with job satisfaction,
correlation with discretionary effort (r = .89) and Model 1), which was in line with pre-
emotional engagement had the greatest correla- vious research, yet the factors were intent to turn over,
tion with job satisfaction (r = .75).1 All three scales distinct and could not be treated as
had nearly the same correlation with well-being one construct (χ2(5) = 271.48, p < discretionary effort,
(r = .50 to .54). Also as hypothesized, employee .001, favoring Model 1). To provide
and well-being,
engagement had a moderate negative correlation further evidence of the distinctness
with intent to turn over (r = –.54), such that the of the construct (i.e., discriminant treating each
more engaged an employee was, the less likely the validity), we compared our hypothe-
employee would intend to leave his or her place of sized model with one factor predict- construct as a latent
work. The most strongly correlated scale with ITS ing all items. Again, results indicated
variable.
was emotional engagement (r = –.62), with only that the model with distinct factors
small correlations for the cognitive and behav- exhibited better model fit (χ2(1035)
ioral subscales (r = –.32 and –.29, respectively). = 49607.25, p < .001, favoring Model 1).
Due to concerns regarding the distinctiveness To go further, because model-data misfit
between the outcomes related to the deployed might be at play in one factor and not others,
scales, and as further evidence of employee engage- we conducted similar analyses with each factor
ment performing as a distinct construct, we com- independently. We first specified a model with
pared three structural models. Model 1 was our employee engagement as a higher-order factor
hypothesized model (employee engagement as a measured by our three scales and with one of the
higher-order factor correlated with job satisfac- other individual scales (job satisfaction, intent to
tion, intent to turn over, discretionary effort, and turn over, discretionary effort, or well-being) as
well-being). To test our hypothesis that those mea- a latent variable. Then we compared that model
sures would be correlated, we compared Model 1 with: (1) a model with one factor for all items on
with Model 2, which was the same as our hypoth- the two scales and (2) a model with the separate
esized model except that factor correlations were factors but uncorrelated. For all eight model com-
set at 0. If Model 2 were to have less model-data parisons, we found that the original hypothesized
model had better model fit (p < .001), indicating incremental validity is established when scores
the employee engagement was correlated to job on a measure explain variance over and above
satisfaction, discretionary effort, psychological a second measure evaluating a similar construct
well-being, and intent to turn over but performed or with a construct that is theoretically related.
as a distinct construct, as measured by those Incremental validity was tested looking at the
instruments deployed in this study. variance explained by the EES compared to the
As an additional assessment of discrimi- variance explained by the engaged teacher scale
nant validity, we calculated several heterotrait- (ETS; Klassen, Yerdelen, & Durksen, 2013)—a mea-
monotrait (HTMT) ratios of the correlations sure of teacher-specific engagement within the
(Henseler, Ringle, & Sarstedt, 2015), which is an context of the education field alongside measures
alternative approach to the Fornell-Larcker crite- of job satisfaction and intention to turn over, in
rion and the examination of cross-loadings and keeping with the theory that engagement would
is based on the multitrait-multimethod matrix. be positively related to job satisfaction and nega-
The HTMT is “the average of the heterotrait-het- tively related to intention to turn over.
eromethod correlations (i.e., the correlations of
indicators across constructs measuring different
Method
phenomena), relative to the average of the mono-
Participants and Procedure
trait-heteromethod correlations (i.e.,
the correlations of indicators within For our final sample (N = 490), participants were
Evidence of the same construct)” (Henseler et al., recruited from a sampling of employees in the
2015, p. 121). We used the HTMT to education field and who were participating in a
incremental validity
examine the discriminant validity of larger study on teacher retention. Of the 490, 67
is established when the three constructs within the EES percent (n = 333) of the sample was female and
and also to examine the discrimi- most participants identified as Caucasian (n =
scores on a measure nant validity between the EES and 382). The largest age groups fell between 36 and
job satisfaction, intent to turn over, 50 years in age (n = 231), followed by 18–35 (n
explain variance over
discretionary effort, well-being, and = 137). Similarly to the previous three studies, to
and above a second cognitive work appraisal. ensure representation of our sample, we compared
The HTMT is an estimate of the the research sample to the general population of
measure evaluating correlation between constructs, so if our study group. Gender and age were roughly
the value is smaller than 1.0, then equivalent for the targeted population.
a similar construct
the true correlation between the As deployed in the previous studies, the
or with a construct constructs is most likely different survey was placed online for administration.
from one, indicating they are differ- Participation was strictly voluntary and partic-
that is theoretically ent constructs (i.e., they exhibit dis- ipants were able to opt out of the study at any
criminant validity). As our criterion, point. As with the previous studies, no significant
related.
we used a threshold of .85 because it issues were encountered with distribution.
is the most likely approach to indi-
The EES
cate a lack of discriminant validity (Henseler et al.,
2015). When comparing the three subscores of the The 12-item version of the EES from Study 3
EES, all three had an HTMT value less than .85, was used. The final scale consisted of three sub-
with emotional having the lowest value (.48) and scale dimensions with four items each: cognitive
cognitive having the highest value (.78). When engagement (e.g., “I am really focused when I am
comparing the EES to job satisfaction, intent to working”), emotional engagement (e.g., “I feel a
turn over, discretionary effort, well-being, and strong sense of belonging to my job”), and behav-
cognitive work appraisal, the HTMT values ranged ioral engagement (e.g., “I am willing to put in
from .51 (for intent to turn over) to .76 (for cog- extra effort without being asked”). Each subscale
nitive work appraisal). Thus, all values indicated is composed of four questions. All items were mea-
discriminant validity among the subscores of the sured on a 5-point Likert scale, where 1 = strongly
EES and between the EES and the other measures disagree and 5 = strongly agree.
deployed in this study.
The Engaged Teachers Scale (ETS)
Study 4 The ETS was used as a specific measure of teacher
engagement due to the occupational context of
Purpose this study. Deployed as a 16-item measure spe-
The purpose of Study 4 was to test for evidence cific to the engagement level of educators and
of predictive incremental validity. Evidence of those in the education field, the ETS was used a
context specific measure of engagement to deploy TABLE VIII Proportion of Variance Explained (R2) in
alongside the EES to explore incremental valid- Three Outcomes to Demonstrate Incre-
ity. The ETS is composed of four subscales: cog- mental Validity (Study 4)
nitive engagement, emotional engagement, social
CWAS-11 JSS ITS
engagement with students, and social engage-
ment with colleagues. In previous work (Klassen ETS scale only .58 .36 .16
et al., 2013), the ETS was found to correlate highly ETS and EES scales .82 .54 .25
with the UWES (Schaufeli et al., 2006), a widely EES scale only .81 .53 .24
utilized measure of work engagement. All items Note: CWAS-11 = cognitive work appraisal scale-11; ITS = intention to turno-
were measured on a 5-point Likert scale, where 1 = ver scale; ETS = engaged teacher scale; EES = employee engagement scale.
strongly disagree and 5 = strongly agree.
Implications for Theory and Practice in the complete work experience. To be clear, our
Human Resources position is that the JES is a sound, psychomet-
rically established measure of job engagement.
The major contribution of our work is that
Scholars looking for evidence about levels of job
the EES is the first measure designed to mea-
engagement should be encouraged to use mea-
sure employee engagement—not job engage-
sures such as the JES. However, the JES may not
ment (JES; Rich et al., 2010; Saks, 2006), work
capture the more broad, unique experiences of
engagement (UWES; Schaufeli et al., 2006), orga-
employee engagement. Notwithstanding, in
nizational engagement (Saks, 2006), or intel-
the Saks (2006) measures of both job and orga-
lectual and/or social engagement (Soane et al.,
nizational engagement, participants are asked
2012)—grounded in previously documented
to respond to a series of questions concerning
theory and definition. While the JES, UWES,
“my job” and “this organization” (p. 617; e.g.,
ISA, and Saks (2006) measures of job and orga-
“I really ‘throw’ myself into my job and being a
nizational engagement are robust measures for
member of this organization is very captivating”
nuanced foci, the EES was specifically grounded [italics added for emphasis]). While the EES does
in a definition and theoretical framework (cf. ask questions about the job and the organiza-
Shuck et al., 2014) focused for unpacking and tion, it also seeks information about the team,
explaining the more full spectrum experience doing more than expected, and how someone cares
of employee engagement, specifi- about their organization. See Table VI.
cally (Shuck et al., 2014; Shuck & In support for developing the EES, Saks and
The EES provides
Wollard, 2010). As we have previ- Gruman (2014) lamented, “a top priority is to
a psychometrically ously noted, scholars have called for develop a valid measure of employee engage-
a more detailed understanding of ment that is truly distinct from other constructs”
strong measurement employee engagement as a means
(p. 178). They go on to suggest “future research
to refine and improve understand- should focus on the development of new mea-
operationalization
ing of the employee engagement sures that have their basis in Kahn’s (1990) con-
of that phenomenon construct. The EES provides a psy- ceptualization and theory of engagement” (Saks
chometrically strong measurement & Gruman, 2014, p. 178). Purcell (2014) sug-
with strong, grounded operationalization of that phenom-
gested that to move the concept of employee
enon with strong, grounded theo- engagement forward, we must “build on the
theoretical roots,
retical roots, directly responding to key advantage of the focus on employees, their
directly responding calls in both theory and practice. In beliefs, values, behaviors, and experiences at
light of other measures of engage- work in a way not seen before the mainstream”
to calls in both theory ment-like states, the EES advances
(Purcell, 2014, p. 251). The development of the
the field forward and contributes EES responds directly to these calls for the further
and practice.
to emerging engagement theory by refinement, as well as those of Albrecht (2010)
way of precise measurement and and Cole et al. (2011), who both advocated for
construct focus (instead of construct entangle- further psychometric exploration of Kahn’s origi-
ment with those measurement tools that may nal conceptualization of engagement and the
not fully capture the employee engagement three-dimensional model of engagement (i.e.,
experience, cf. Shuck, 2011). cognitive, emotional, and behavioral; Shuck &
For example, as we have mentioned, the JES, Wollard, 2010). The EES is theoretically grounded
which we used to show convergent validity in in Kahn’s (1990) original conceptualization and
Study 2, uses the postfix “my job” at the end rooted in theory that supported Kahn’s “theory
of every question on the scale (i.e., I work with generating ethnographic research” (Rich et al.,
intensity on my job [italics added]). The postfix 2010, p. 618).
potentially influences a participant to answer We also note that previous work on employee
only to the context of their job—an important engagement has almost exclusively relied on the
but limited context of the full work experi- UWES for the measurement of engagement as evi-
ence. In comparison, only two questions on the denced by continued mounting citations and ref-
EES end with the postfix “my job.” While job erences in emerging research and doctoral work
engagement is important, the employee engage- across the globe (see for example, Sarti, 2014).
ment phenomenon is presented as a broader, This is troubling given that several independent
more full experience influenced by environ- research reports have called into question the use
ment conditions, experiences, and interpreta- of the UWES on the grounds of concept redun-
tions of in-the-moment occurrences that capture dancy (Cole et al., 2011; Nimon et al., 2016; Saks
& Gruman, 2014). As for the EES, we believe that Limitations and Implications for Future
employee engagement represents something Research
unique—an independent phenomenon—that
No measure in its initial presentation is devoid of
connotes positive energy directed toward some
limitations. In the following, we highlight three
target and that this happens cognitively, emo-
major limitations of our work and potential impli-
tionally, and behaviorally, not as the opposite
cations for future research.
of something. Certainly, scholars and practitio-
ners who wish to study the burnout perspective
of engagement, or work engagement specifically, Issues of Validity and Reliability
should be encouraged to use a measure such as First, there is a need for further psychometric test-
the UWES if it fits their purpose, scope, and beliefs ing of the EES including additional evidence of
about what engagement is. Notwithstanding, the convergent, discriminant, and nomological valid-
UWES measures work engagement, not employee ity. For example, although we established a level
engagement and the EES measures employee of nomological and convergent validity with the
engagement, not work engagement (in the same CWAS-11 and the JES (Rich et al., 2010), we wonder
way that the JES measures job engagement). how the EES might perform alongside other mea-
Further, based on the results of Study 4 and the sures of engagement (the ISA; Soane et al., 2012;
discriminant validity issues we encountered with organizational engagement; Saks,
the ETS, we recommend administering the EES 2006), as well as measures of satisfac-
in place of and not in addition to the ETS when tion, commitment, and involvement The CWAS-11 is
studying engagement within the occupational (as a means to test for nomologi-
not meant to be
contexts of education. cal validity). We would advocate
Finally, we note the introduction of the the use of the EES alongside other an exhaustively
CWAS-11, a secondary focus of our work, meant constructs commonly associated
to connect more broadly to those resources and with employee engagement, such comprehensive tool,
antecedents of engagement alluded to by Saks as affective commitment, job fit,
but rather an easily
and Gruman (2014) in their developing theory well-being, and psychological work-
of employee engagement and connected to pre- place climate, among others. There deployable measure
vious work by Wollard and Shuck (2011). Before is an exhaustive list of variables that
moving forward, we note the much more com- could be explored alongside the EES of those antecedental
prehensive work cognition inventory (WCI; presenting fruitful implications for
conditions often
Nimon et al., 2011) and work cognition inven- future research.
tory-revised (WCI-R; Nimon & Zigarmi, 2014) Also, we recommend that future referred to in the
as a robust, more nuanced battery of anteceden- studies focus on developing addi-
tal factors connected to the influencing condi- tional levels of criterion evidence research literature
tions of employee engagement. The CWAS-11 is related to the validity of the EES
but hard to capture
not meant to be an exhaustively comprehensive scale. This could include further
tool, but rather an easily deployable measure of validation work (e.g., examining for due to survey fatigue
those antecedental conditions often referred to evidence of measurement invari-
in the research literature but hard to capture due ance across groups; Nimon & Reio, and increasingly
to survey fatigue and increasingly complex mod- 2011), reliability, and test-retest
complex models.
els. Similar in function to other short, anteceden- applications. Future research might
tal scales (e.g., the Gallup Q12; Harter, Schmidt, also focus on the development of a
& Hayes, 2002), the CWAS-11 scale connects short form of the EES. Such a short form should
conceptually and theoretically with the WCI be administered stand-alone to examine whether
(Nimon et al., 2011) as well as research on the the psychometric properties as well as exter-
antecedental conditions of engagement targeted nal validity evidence uphold. Further, although
in research (Brown & Leigh, 1996; Kahn, 1990; consistent with prior instrument development
Saks, 2006; Wollard & Shuck, 2011). More spe- research (Waltz et al., 2005), an additional limita-
cifically, the coupling of the CWAS-11 and the tion regarding future scale development may have
EES connects the first two pieces of the engage- been the use of only two expert panelists for the
ment puzzle that Nimon et al. (2011) discussed CVR (Lawshe, 1975) technique used in Study 1.
in their work—the appraisal and the phenom- Lawshe (1975) suggested that five or more par-
enon. The presentation of these data represents ticipants was ideal for using the CVR and conse-
one of the first empirical testings of this puzzle quently, future research that works to shorten the
simultaneously. EES could deploy a more robust practice of the
(2016), the Child-Life Council Research-to-Practice Award (2015), the E. F. Holton III Outstand-
ing Article of the Year (2013 and 2015) and the Advances in Developing Human Resources
Outstanding Issue of the Year (2011). He holds faculty affiliate status with the Department
of Counseling and Human Development (University of Louisville) and the US Army Cadet-
Command. He is director of the Leadership, Engagement, and Development Lab (LEADLab),
a multidisciplinary research team focused on understanding how human beings interact with
their places of work.
THOMAS G. REIO JR. is professor of adult education and human resource development and
associate dean of graduate studies at Florida International University in Miami, Florida. He
is a past editor of Human Resource Development Review and is currently associate editor
of Human Resource Development Quarterly. His research concerns curiosity and risk-taking
motivation, workplace socialization processes, workplace incivility, and workplace learning.
His work has been published in leading journals in education, business, and psychology,
including Personality and Individual Differences, Journal of School Psychology, Educational
and Psychological Measurement, Journal of Business and Psychology, Journal of Interper-
sonal Violence, Human Resource Development Quarterly, Human Resource Development
International, Journal of Managerial Psychology, and the Journal of School Psychology.
and practice (pp. 87–97). Northhampton, MA: Edward Locke, E. A. (1976). The nature and causes of job satisfac-
Elgar. tion. In M. D. Dunnette (Ed.), Handbook of industrial and
Dalal, R. S., Baysinger, M., Brummel, B. J., & LeBreton, J. M. organizational psychology (pp. 1297–1343). Chicago, IL:
(2012). The relative importance of employee engagement, Rand McNally.
other job attitudes, and trait affect as predictors of job Lodahl, T. M., & Kejner, M. (1965). The definition and meas-
performance. Journal of Applied Social Psychology, 42, urement of job involvement. Journal of Applied Psychol-
295–325. doi: 10.1111/j.1559-1816.2012.01017.x ogy, 49, 24–33. doi: 10.1037/h0021692
Dillman, D. A., Smyth, J. D., & Christian, L. M. (2009). Inter- Luthans, F., Avolio, B. J., Avey, J. B., & Norman, S. M. (2007).
net, mail, and mixed-mode surveys: The tailored design Positive psychological capital: Measurement and rela-
method (3rd ed.). Hoboken, NJ: Wiley. tionship with performance and satisfaction. Personnel
Fletcher, L., & Robinson, D. (2014). Measuring and under- Psychology, 60, 541–572. doi: 10.1111/j.1744-6570.2007
standing engagement. In C. Truss, K. Alfes, R. Delbridge, .00083.x
A. Shantz, & E. Soane (Eds.), Employee engagement in Macey, W. H., & Schneider, B. (2008). The meaning of
theory and practice (pp. 273–290). New York, NY: Rout- employee engagement. Industrial and Organizational
ledge. Psychology, 1, 3–30. doi: 10.1111/j.1754-9434.2007.0002.x
Gruman, J., & Saks, A. (2011). Performance management May, D. R., Gilson, R. L., & Harter, L. M. (2004). Psychological
and employee engagement. Human Resource Manage- conditions of meaningfulness, safety and availability and
ment Review, 21, 123–136. doi: 10.1016/j.hrmr.2010.09.004 the engagement of the human spirit at work. Journal of
Guest, D. (2014). Employee engagement: Fashionable fad or Occupational and Organizational Psychology, 77, 11–37.
long-term fixture? In C. Truss, K. Alfes, R. Delbridge, A. doi: 10.1348/096317904322915892
Shantz, & E. Soane (Eds.), Employee engagement in the- Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The meas-
ory and practice (pp. 221–235). New York, NY: Routledge. urement of organizational commitment. Journal of Voca-
Harter, J. K., Schmidt, F. L., & Hayes, T. L. (2002). Business- tional Behavior, 14, 224–247. doi: 10.1016/0001-8791(79)
unit-level relationship between employee satisfaction, 90072-1
employee engagement, and business outcomes: A meta- Muthén, L. K., & Muthén, B. O. (1998-2010). Mplus Users
analysis. Journal of Applied Psychology, 87, 268–279. doi: Guide. Sixth Edition. Los Angeles, CA: Muthén & Muthén.
10.1037//0021-9010.87.2.268 Newman, D. A., Joseph, D. L., Sparkman, T. E., & Carpenter,
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new crite- N. C. (2011). Invited reaction: The work cognition inven-
rion for assessing discriminant validity in variance-based tory: Initial evidence of construct validity. Human Resource
structural equation modeling. Journal of the Academy of Development Quarterly, 22, 37–47. doi: 10.1002/hrdq.20065
Marketing Science, 43, 115–135. Nimon, K., & Reio, T. G., Jr. (2011). Measurement invariance:
Kahn, W. (1990). Psychological conditions of personal A foundational principle for quantitative theory building.
engagement and disengagement at work. Academy of Human Resource Development Review, 10, 198–214. doi:
Management Journal, 33, 692–724. doi: 10.2307/256287 10.1177/1534484311399731
Kahn, W., & Heaphy, E. (2014). Relational contexts of Nimon, K., Shuck, B., & Zigarmi, D. (2016). Construct overlap
personal engagement at work. In C. Truss, K. Alfes, between employee engagement and job satisfaction: A
R. Delbridge, A. Shantz, & E. Soane (Eds.), Employee function of semantic equivalence? Journal of Happiness
engagement in theory and practice (pp. 82–96). New York, Studies, 17(3), 1149-1171. doi: 10.1007/s10902-015-9636-6.
NY: Routledge. Nimon, K., & Zigarmi, D. (2014). The work cognition inven-
Kanungo, R. N. (1982). Measurement of job and work tory: Initial evidence of construct validity for the revised
involvement. Journal of Applied Psychology, 67, 341–349. form. Journal of Career Assessment. (Advance online
doi: 10.1037/0021-9010.67.3.341 publication.) doi: 1069072714523241
Kelley, T. L. (1927). Interpretation of educational measure- Nimon, K., Zigarmi, D., Houson, D., Witt, D., & Diehl, J.
ments. New York, NY: World Book. (2011). The work cognition inventory: Initial evidence of
Klassen, R. M., Yerdelen, S., & Durksen, T. L. (2013). Measur- construct validity. Human Resource Development Quar-
ing teacher engagement: Development of the engaged terly, 22, 7–35. doi: 10.1002/hrdq.20064
teachers scale (ETS). Frontline Learning Research, 1(2), Parker, S. K., & Griffin, M. A. (2011). Understanding active
33–52. psychological states: Embedding engagement in a wider
Kühnel, J., Sonnentag, S., & Westman, M. (2009). Does work nomological net and closer attention to performance.
engagement increase after a short respite? The role of job European Journal of Work and Organizational Psychol-
involvement as a double-edged sword. Journal of Occu- ogy, 20, 60–67. doi: 10.1080/1359432X.2010.532869
pational and Organizational Psychology, 82, 575–594. Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making
Lawler, E. E., & Hall, D. T. (1970). Relationship of job char- sense of factor analysis: The use of factor analysis for
acteristics to job involvement, satisfaction, and intrinsic instrument development in health care research. London,
motivation. Journal of Applied Psychology, 54, 305–312. England: Sage.
doi: 10.1037/h0029692 10.1037/h0029692 Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff,
Lawshe, C. H. (1975). A quantitative approach to content N. P. (2003). Common method biases in behavioral
validity. Personnel Psychology, 28, 563–575. doi: 10.1111/ research: A critical review of the literature and recom-
j.1744-6570.1975.tb01393.x mended remedies. Journal of Applied Psychology, 88,
Lloyd, R. (2008). Discretionary effort and the performance 879–903. doi: 10.1037/0021-9010.88.5.879
domain. Australian and New Zealand Journal of Organi- Preacher, K. J., & MacCallum, R. C. (2003). Repairing Tom
zational Psychology, 1, 22–34. Swift’s electric factor analysis machine. Understand-
ing statistics: Statistical Issues in Psychology, Educa- Shuck, B., & Reio, T. (2011). The employee engage-
tion, and the Social Sciences, 2(1), 13–43. doi: 10.1207/ ment landscape and HRD. Advances in Devel-
s15328031us0201_02 oping Human Resources, 13, 419–428. doi:
Purcell, J. (2014). Disengaging from engagement. Human 10.1177/1523422311431153
Resource Management Journal, 24, 241–254. doi: Shuck, B., & Reio, T. (2013). Employee engagement and well-
10.1111/1748-8583.12046 being: A moderation model and implications for practice.
Reio, T. G., Jr. (2010). The threat of common method variance Journal of Leadership and Organizational Studies, 21,
bias in theory building. Human Resource Development 43–58. doi: 10.1177/1548051813494240
Review, 9, 405–411. doi: 10.1177/1534484310380331 Shuck, B., & Reio, T., & Rocco, T. (2011). Employee engage-
Rich, B. L., LePine, J. A., & Crawford, E. R. (2010). Job ment: An examination of antecedent and outcome vari-
engagement: Antecedents and effects on job perfor- ables. Human Resource Development International, 14,
mance. Academy of Management Journal, 53, 617–635. 427–445. doi: 10.1080/13678868.2011.601587
doi: 10.5465/AMJ.2010.51468988 Shuck, B., Twyford, D., Reio, T. G., & Shuck, A. (2014). Human
Rurkkhun, S., & Bartlett, K. R. (2012). The relation- resource development practices and employee engage-
ship between employee engagement and organi- ment: Examining the connection with employee turnover
zational citizenship behavior in Thailand. Human intentions. Human Resource Development Quarterly, 25,
Resource Development International, 15, 157–174. doi: 239–270. doi: 10.1002/hrdq.21190
10.1080/13678868.2012.664693 Shuck, B., & Wollard, K. (2010). Employee engage-
Saks, A. M. (2006). Antecedents and consequences of ment and HRD: A seminal review of the founda-
employee engagement. Journal of Managerial Psychol- tions. Human Resource Development 9, 89–110. doi:
ogy, 21, 600–619. doi: 10.1108/02683940610690169 10.1177/1534484309353560
Saks, A. M., & Gruman, J. A. (2014). What do we really know Shuck, B., Zigarmi, D., & Owen, J. (2015). Psychological
about employee engagement? Human Resource Devel- needs, employee engagement, and work intentions: A
opment Quarterly, 25, 155–182. doi: 10.1002/hrdq.21187 Bayesian multi-measurement mediation approach and
Sarti, D. (2014). Job resources as antecedents of engage- implications for HRD. European Journal of Training and
ment at work: Evidence from a long-term care setting. Development, 39, 2–21. doi: 10.1108/EJTD-08-2014-0061
Human Resource Development Quarterly, 25, 213–237. Soane, E., Truss, C., Alfes, K., Shantz, A., Rees, C., & Gatenby,
doi: 10.1002/hrdq.21189 M. (2012). Development and application of a new meas-
Schaufeli, W. B., Bakker, A. B., & Salanova, M. (2006). The ure of employee engagement: The ISA engagement scale.
measurement of work engagement with a short question- Human Resource Development International, 15, 529–547.
naire. Educational and Psychological Measurement, 66, doi: 10.1080/13678868.2012.726542
701–716. doi: 10.1177/0013164405282471 Thompson, B. (2004). Exploratory and confirmatory fac-
Schaufeli, W. B., Salanova, M., González-Romá, V., & Bak- tor analysis: Understanding concepts and applications.
ker, A. B. (2002). The measurement of engagement and Washington, DC: American Psychological Association.
burnout: A two sample confirmatory factor analytic Vroom, V. H. (1962). Ego-involvement, job satisfaction, and
approach. Journal of Happiness Studies, 3, 71–92. doi: job performance. Personnel Psychology, 15, 159–177. doi:
10.1023/A:1015630930326 10.1111/j.1744-6570.1962.tb01858.x
Scrima, F., Lorito, L., Parry, E., & Falgares, G. (2013). The Waltz, C. F., Strickland, O. L., & Lenz, E. R. (2005). Measure-
mediating role of work engagement on the relationship ment in nursing and health research (3rd ed.). New York,
between job involvement and affective commitment. NY: Springer.
International Journal of Human Resource Management, Wildermuth, C., Vaughan, A. G., & Christo-Baker, E. A. (2013).
25, 2159–2173. doi: 10.1080/09585192.2013.862289 A path to passion: Connecting personality, psychological
Shuck, B. (2011). Four emerging perspectives of employee conditions, and emotional engagement. Journal of Psy-
engagement: An integrative literature review. Human chological Issues in Organizational Culture, 3, 18–45. doi:
Resource Development Review, 10, 304–328. doi: 10.1002/jpoc.21082
10.1177/1534484311410840 Wollard, K. K., & Shuck, B. (2011). Antecedents to employee
Shuck, B., Adelson, J., & Reio, T. G., Jr. (2015). The employee engagement. Advances in Developing Human Resources,
engagement scale: Initial evidence of construct validity 13, 429–446. doi: 10.1177/1523422311431220
and implications for HRD. In J. Moats (Ed.), Academy of Xanthopoulou, D., Bakker, A. B., Demerouti, E., & Schaufeli,
Human Resource Development Conference, St. Louis, W. (2009). Work engagement and financial returns: A
MO. diary study on the role of job and personal resources.
Shuck, B., Ghosh, R., Zigarmi, D., & Nimon, K. (2013). The jin- Journal of Occupation and Organizational Psychology,
gle jangle of employee engagement: Further exploration 82, 183–200.doi: 10.1348/096317908X285633
of the emerging construct & implications for workplace Yalabik, Z. Y., Popaitoon, P., Chowne, J. A., & Rayton,
learning and performance. Human Resource Develop- B. A. (2013). Work engagement as a mediator between
ment Review, 12, 11–35. doi: 10.1177/1534484312463921 employee attitudes and outcomes. International Journal
Shuck, B., Nimon, K., & Zigarmi, D. (2016). Untangling the of Human Resource Management, 24, 2799–2823. doi:
predictive nomological validity of employee engage- 10.1080/09585192.2013.763844
ment: Decomposing variance in employee engagement Yu, C. Y. (2002). Evaluating cutoff criteria of model fit indices
using job attitude measures. Group and Organiza- for latent variable models with binary and continuous
tional Management. (Advance online publication.) doi: outcomes (Doctoral dissertation). University of California
10.1177/1059601116642364 Los Angeles.