Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/11448834

The Nature and Dimensionality of Organizational Citizenship Behavior: A


Critical Review and Meta-Analysis

Article  in  Journal of Applied Psychology · March 2002


DOI: 10.1037/0021-9010.87.1.52 · Source: PubMed

CITATIONS READS

1,613 17,192

3 authors, including:

Amir Erez
University of Florda
55 PUBLICATIONS   10,228 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Rudeness View project

All content following this page was uploaded by Amir Erez on 28 February 2017.

The user has requested enhancement of the downloaded file.


Journal of Applied Psychology Copyright 2002 by the American Psychological Association, Inc.
2002, Vol. 87, No. 1, 52– 65 0021-9010/02/$5.00 DOI: 10.1037//0021-9010.87.1.52

The Nature and Dimensionality of Organizational Citizenship Behavior:


A Critical Review and Meta-Analysis

Jeffrey A. LePine and Amir Erez Diane E. Johnson


University of Florida University of Alabama

This article reviews the literature on organizational citizenship behavior (OCB) and its dimensions as
proposed by D. W. Organ (1988) and other scholars. Although it is assumed that the behavioral
dimensions of OCB are distinct from one another, past research has not assessed this assumption beyond
factor analysis. Using meta-analysis, the authors demonstrate that there are strong relationships among
most of the dimensions and that the dimensions have equivalent relationships with the predictors (job
satisfaction, organizational commitment, fairness, trait conscientiousness, and leader support) most often
considered by OCB scholars. Implications of these results are discussed with respect to how the OCB
construct should be conceptualized and measured in the future.

Organizations have shifted away from the use of strict hierar- efforts. For example, is it appropriate to draw inferences about
chical structures and individualized jobs. Instead, somewhat au- OCB from a study that considers only one or two dimensions of
tonomous team-based work structures have been implemented, and OCB (e.g., altruism or sportsmanship)? As another example, what
this implementation has increased the importance of individual kinds of inferences can be made about OCB from a study in which
initiative and cooperation (Ilgen & Pulakos, 1999). As a result of the dimensions have different relationships with predictors, out-
this trend, organizational citizenship behavior (OCB), or behavior comes, or both?
that contributes indirectly to the organization through the mainte- Unfortunately, and despite the existence of three published OCB
nance of the organization’s social system (Organ, 1997), has been meta-analyses (Organ & Ryan, 1995; Podsakoff, MacKenzie, &
of increasing interest to both scholars and managers (Howard, Bommer, 1996a; Podsakoff, MacKenzie, Paine, & Bachrach,
1995; LePine, Hanson, Borman, & Motowidlo, 2000; Motowidlo, 2000), it is difficult to provide answers to such questions because
Borman, & Schmit, 1997; Motowidlo & Schmit, 1999; Organ & fundamental questions remain about the OCB construct itself and
Ryan, 1995). how it relates to its dimensions (Law, Wong, & Mobley, 1998;
Most of the research on OCB has been focused on identifying its Motowidlo, 2000). Although the three previous meta-analyses
predictors (Van Dyne & LePine, 1998). In conducting this re- examined relationships among dimensions of OCB and a variety of
search, scholars generally link predictors to an overall measure of correlates, there was no focus on how the dimensions related to
OCB, or they link predictors to the dimensions of OCB suggested one another, nor were there any systematic comparisons of how the
by Organ (1988). These dimensions are most often measured by different dimensions related to other variables in the broader
using scales such as those developed by Podsakoff and his col- nomological network.
leagues (e.g., Podsakoff, MacKenzie, Moorman, & Fetter, 1990). The purpose of this study was to critically evaluate the nature of
Although the research from both approaches has generated some the OCB construct and its dimensionality in the context of the
very interesting and important findings, there is the question of research that has taken place since its inception in the early 1980s.
which OCB operationalization is most appropriate. This question We attempted to use the existing research on OCB to describe how
is fundamental because it relates not only to how OCB should be investigators have implicitly defined OCB relative to its dimen-
measured but also to how results of research should be interpreted sions. We begin by briefly reviewing the literature on the OCB
and how OCB should be conceptualized in future theory-building construct and the development of the most popular dimensional
framework. Following this discussion, we present several meta-
analyses aimed at uncovering relationships among the dimensions
and between dimensions and predictors. We conclude with a
Jeffrey A. LePine and Amir Erez, Department of Management, War- discussion of the implications of our findings.
rington College of Business Administration, University of Florida; Diane
E. Johnson, Department of Management and Marketing, University of
Alabama. The OCB Construct
A preliminary version of this article was presented at the 16th Annual
Interest in work behavior that is discretionary, not directly or
Conference of the Society for Industrial and Organizational Psychology,
April 2001, San Diego, California.
explicitly recognized by the formal reward system, and that, in the
Correspondence concerning this article should be addressed to Jeffrey aggregate, promotes the effective functioning of the organization
A. LePine, Department of Management, Warrington College of Business (Organ, 1988, p. 4) can be traced back to Barnard (1938) and Katz
Administration, University of Florida, P.O. Box 117165, Gainesville, Flor- (1964). Greater scholarly interest in this type of behavior, how-
ida 32611-7165. E-mail: lepineja@notes.cba.ufl.edu ever, seemed to be triggered in the early 1980s after Organ and his

52
DIMENSIONALITY OF OCB 53

colleagues (Bateman & Organ, 1983; Smith, Organ, & Near, 1983) employee. Factor analyses of these ratings indicated two factors.
used the term organizational citizenship behavior to describe it. The first factor, labeled Altruism, captured behavior directly in-
The focus of much of the early OCB research was in response to tended to help a specific person in face-to-face situations (e.g.,
Organ’s (1977) interest in the notion that job satisfaction might helping others who have been absent, volunteering for things that
influence organizational effectiveness through behaviors managers are not required, orienting new people even though it is not
want but cannot technically require (Motowidlo, 2000). required, helping others who have heavy workloads). The second
Over the past two decades, interest in behavior that generally fits factor, labeled Generalized Compliance, represented impersonal
the definition of OCB has increased dramatically (see Podsakoff et behaviors such as compliance with norms defining a good worker
al., 2000). Scholars, however, have not been completely consistent (e.g., being punctual, taking undeserved breaks [reverse scored],
with the terminology used to label it. Labels for domains of not spending time in idle conversation). Although measures of
behavior that overlap with OCB as described by Organ (1988) and these dimensions continue to be used by researchers, some have
others (e.g., Morrison, 1994; Van Dyne, Graham, & Dienesch, modified the response scales. Pond, Nacoste, Mohr, and Rodriguez
1994) include prosocial organizational behavior (Brief & Moto- (1997), for example, asked individuals to rate the frequency with
widlo, 1986; George, 1990, 1991; George & Bettenhausen, 1990; which they engaged in the behavior described by the item.
O’Reilly & Chatman, 1986), organizational spontaneity (George Five years later, Organ (1988) proposed an expanded taxonomy
& Brief, 1992; George & Jones, 1997), and extrarole behavior of OCB that included altruism (narrower than the altruism of
(Van Dyne, Cummings, & McLean Parks, 1995). Smith et al., 1983), conscientiousness (a narrower form of gener-
The label that is probably most relevant to human resource alized compliance), sportsmanship (e.g., not complaining about
management scholars and industrial and organizational psycholo- trivial matters), courtesy (e.g., consulting with others before taking
gists is contextual performance (Borman & Motowidlo, 1993, action), and civic virtue (e.g., keeping up with matters that affect
1997; Borman, White, & Dorsey, 1995; Motowidlo & Van Scotter, the organization). Podsakoff et al. (1990) were among the first
1994; Van Scotter & Motowidlo, 1996). Contextual performance researchers to operationalize Organ’s (1988) five dimensions.
emerged in response to the concern that selection scholars and They used the definitions provided by Organ and generated items
practitioners were neglecting a significant portion of the job per- that were then subjected to a Q sort and a confirmatory factor
formance domain (Motowidlo, 2000). That is, selection scholars analysis. The measure asked supervisors to rate the extent to which
and practitioners traditionally focused on activities that either they agreed that employees engaged in the behavior reflected in
supported or directly contributed to the transformation of the the items. The resulting OCB scales have served as the basis for
organization’s inputs to outputs while they also tended to ignore OCB measurement in a large number of empirical studies (e.g.,
activities that supported the social and psychological context in MacKenzie, Podsakoff, & Fetter, 1991; Moorman, 1991, 1993;
which the organization’s technical core was embedded. Borman Moorman, Niehoff, & Organ, 1993; Niehoff & Moorman, 1993;
and Motowidlo (1993, 1997) referred to the former set of activities Podsakoff & MacKenzie, 1994; Podsakoff, MacKenzie, & Bom-
as “task performance” and the latter set of activities as “contextual mer, 1996b; Podsakoff, MacKenzie, & Fetter, 1993; Tansky,
performance.” 1993).
As Motowidlo (2000) pointed out, although the behavioral do- Several other taxonomies of OCB-like behaviors have been
mains of OCB and contextual performance overlap a great deal, proposed and operationalized (e.g., Borman & Motowidlo, 1993;
there initially were some important definitional differences. Spe- Morrison, 1994; Van Dyne et al., 1994); however, the behavioral
cifically, Organ (1988) originally suggested that OCB must be domains of these taxonomies overlap with each other and with
discretionary and nonrewarded, which was not the case for con- Organ’s (1988) OCB domain in varying degrees (Coleman &
textual performance. Almost a decade later, Organ (1997) recog- Borman, 2000). For example, Van Dyne et al.’s (1994) OCB
nized the conceptual difficulties associated with these require- framework includes social participation, which overlaps with al-
ments (e.g., what is discretionary varies a great deal from person truism and courtesy; loyalty, which overlaps with sportsmanship
to person and from situation to situation). He then redefined OCB and a bit of civic virtue; and obedience, which overlaps with civic
as behavior that contributes “to the maintenance and enhancement virtue and conscientiousness. The final dimension, functional par-
of the social and psychological context that supports task perfor- ticipation, does not overlap with any of Organ’s dimensions, but it
mance” (Organ, 1997, p. 91). This modified definition of OCB is is very similar to Coleman and Borman’s notion of job–task
very similar to Borman and Motowidlo’s (1993, 1997) definition citizenship performance.
of contextual performance. Regardless of the specifics of the Morrison (1994) offered another OCB framework. Her altruism
definition, however, scholars have always conceived of OCB as dimension overlaps with Organ’s (1988) altruism and courtesy
consisting of several behavioral dimensions. dimensions. Her conceptualization of conscientiousness is a bit
narrower than Organ’s. She also presented sportsmanship and
Dimensionality of OCB involvement dimensions, the latter of which include components
of Organ’s sportsmanship and civic virtue dimensions. The last
Smith et al. (1983) were interested in predicting organizationally dimension, “keeping up with changes,” overlaps with Organ’s
beneficial behavior that was not formally rewarded and could not civic virtue and conscientiousness dimensions.
be enforced by the organization in terms of formal role expecta- As a final example, Van Scotter and Motowidlo (1996) mea-
tions or job requirements. They conducted structured interviews in sured two dimensions of contextual performance by asking super-
which managers identified “instances of helpful but not absolutely visors to rate employees on how likely they were to engage in
required . . . job behavior” (Smith et al., 1983, p. 656). Supervisors certain behaviors. The first dimension, interpersonal facilitation,
were then asked to rate how characteristic each behavior was of the overlaps with Organ’s (1988) altruism and courtesy dimensions,
54 LEPINE, EREZ, AND JOHNSON

Morrison’s (1994) altruism dimension, and Van Dyne et al.’s not been used as often, and even when there are several studies,
(1994) social participation dimension. The second dimension, job there is less consistency with respect to the specific behaviors
dedication, includes elements of Organ’s sportsmanship, civic studied. For example, contextual performance scholars generally
virtue, and conscientiousness dimensions. In addition, job dedica- take a job analytic approach to identifying and measuring impor-
tion includes elements pertaining to persistence in completing tant job-relevant behavior. Accordingly, they tend to use instru-
one’s own job or task. This last element is similar to Van Dyne et ments that measure specific contextual behaviors identified by
al.’s (1994) functional participation and Coleman and Borman’s subject matter experts as making a positive (or a negative) contri-
(2000) job–task citizenship performance. bution to the organization in a specific setting or job. As a result,
Perhaps in recognition that the behavioral elements of OCB the contextual behaviors studied and the instruments used to mea-
overlap with each other (Coleman & Borman, 2000; Motowidlo, sure the behaviors may be more idiosyncratic than studies framed
2000; Organ, 1997), scholars have begun to consider whether the in terms of OCB.
elements should be combined into conceptually distinct subgroups.
Williams and Anderson (1991), for example, suggested that orga- Questions About the Definition and
nizational citizenship behavior directed toward individuals (OCBI) Dimensionality of OCB
is distinct from organizational citizenship behavior directed toward
the organization (OCBO). Altruism and courtesy are behaviors that Schwab (1980) defined construct validity as “the correspon-
fit in the former category, whereas sportsmanship, civic virtue, and dence between a construct (conceptual definition of a variable) and
conscientiousness fit in the latter category. the operational procedure to measure or manipulate that construct”
In a similar manner, Coleman and Borman (2000) noted that (p. 6). Construct validity, therefore, requires a definition that
there has been a proliferation of behavioral elements that fit the clearly specifies what a measure of it should and should not reflect
general definition of OCB, but scholars have not conducted re- and, thus, should also convey information about its dimensionality.
search aimed at identifying the extent to which these behaviors Information about the way the dimensions relate to the overall
represent broader underlying constructs. To address this issue, construct is also critically important because the results of model
Coleman and Borman used similarity data generated through con- testing vary a great deal depending on the way this relationship is
tent sorting of 27 citizenship behaviors and analyzed these data represented (Law & Wong, 1999). Misspecification of this rela-
using exploratory factor analysis, multidimensional scaling, and tionship increases the risk of making errors when parameter esti-
cluster analysis. Consistent with Williams and Anderson (1991), mates and tests of significance are being interpreted. Law et al.
the results of these analyses suggested categories of behaviors that (1998) suggested that scholars have not adequately defined OCB
vary with respect to the behaviors’ beneficiary. Their interpersonal with respect to its dimensions. In their article, Law et al. also
citizenship performance dimension refers to behavior that benefits described several ways by which the dimensions could relate to the
other organizational members and includes Organ’s (1988) altru- overall construct of OCB.
ism and courtesy dimensions and, thus, is similar to OCBI. Their If OCB conformed to a “latent” model, each dimension would
organizational citizenship performance dimension refers to behav- be a manifestation of OCB, and measures of the dimensions would
ior that benefits the organization and includes Organ’s sportsman- include some variance reflecting OCB, other systematic variance
ship, civic virtue, and conscientiousness dimensions and, thus, is not related to OCB, and error variance. The causal arrow in this
similar to OCBO. Their third dimension, called job–task citizen- model would be from OCB to the dimensions, and therefore, OCB
ship performance, refers to behavior that reflects extra effort and would be a latent variable that partially causes these dimensions.
persistence on the job, dedication to the job, and the desire to Law et al. (1998) used general cognitive ability, or g, as an
maximize one’s own job performance. Although this latter dimen- example of a latent construct because g is thought to be a cause of
sion seems to be outside the scope of Organ’s notion of OCB, it is scores on tests of more specific abilities. Motowidlo (2000) sug-
quite similar to the functional participation (Van Dyne et al., 1994) gested that if OCB were a latent construct, it would be similar to
and job dedication (Van Scotter & Motowidlo, 1996) dimensions. a personality construct. That is, OCB would be like a trait that
As the previous paragraphs suggest, the literature on OCB and causes the behaviors that are reflected in the dimensions. To our
related concepts is fairly diverse, with respect to both the nature of knowledge, scholars have not explicitly taken this approach in
the behavioral dimensions studied and perhaps more so the jargon defining OCB with respect to its dimensions. However, Organ
used by scholars to label the dimensions. By a wide margin, (1997) seemed to imply this perspective when he suggested that
however, Organ’s (1988) five-dimension framework has been the satisfaction would affect “people’s willingness to help colleagues
subject of the greatest amount of empirical research for at least and work associates and their disposition to cooperate in varied
three reasons. First, Organ’s framework has the longest history, and mundane forms to maintain organized structures that govern
and he and his colleagues have been very prolific with respect to work” (p. 92).
publishing OCB articles and book chapters. Second, Podsakoff and If OCB conformed to an “aggregate” model, OCB would be
his colleagues provided the field with a sound measure of Organ’s formed as a mathematical function of the dimensions. According
five dimensions (Podsakoff et al., 1990), and they conducted to this perspective, each of the dimensions would be part of the
several excellent empirical studies that included this measure (e.g., OCB construct. OCB would exist to the extent that systematic
Podsakoff et al., 1996b). Finally, OCB scholars generally assume variance from each dimension (common as well as specific) would
that over the long run, the behavioral dimensions are beneficial be captured and added (or perhaps multiplied) together. From this
across situations and organizations (Organ, 1997). As a result, perspective, the causal arrows (although they are not truly causal
OCB researchers typically measure all or most of the dimensions here) can be thought of as originating from the dimension and
in the same way across studies. Other behavioral frameworks have pointing toward the OCB construct. Law et al. (1998) used the
DIMENSIONALITY OF OCB 55

example of job satisfaction as an aggregate construct because job Research Questions


satisfaction can be thought of as a sum of scores on instruments
measuring satisfaction with pay, satisfaction with coworkers, sat- Our intent in pointing out deficiencies in OCB theory and
isfaction with supervision, and so forth. research is not to suggest that the results of past studies of OCB
Motowidlo (2000) recently pointed out that contextual perfor- should be written off. Indeed, we believe that the existing literature
mance is a clear example of an aggregate multidimensional con- base may be very helpful in guiding the development of OCB
struct. He and his colleagues (e.g., Motowidlo et al., 1997) defined theory. This belief is consistent with that of Schwab (1980), who
job performance as the aggregated value to the organization of the described how the results of empirical studies often lead scholars
behavioral episodes performed by individuals over time that have to modify measures of constructs and construct definitions. In fact,
positive or negative consequences for the organization. Contextual Schwab explicitly suggested that all of these modifications are a
performance, in turn, is defined as the aggregated value to the natural part of the construct validity “process.” Because a large
organization of all the behavioral episodes that have effects on the number of studies have separately considered the dimensions of
social, organizational, and psychological context of the organiza- OCB, for example, it is possible to examine relationships among
tion’s technical core. dimensions as well as relationships between a common set of
Many OCB scholars view the behavioral dimensions as being predictors and the different dimensions. It certainly is not optimal
related (e.g., somewhat discretionary behaviors intended as posi- to begin with data and then infer a construct definition; however,
tive contributions to the organization) but distinct (e.g., the most when construct definitions are assumed or ill specified, it is pos-
proximal beneficiary of the behaviors differs). Consistent with this sible to examine empirical relationships in an effort to infer con-
viewpoint, scores on measures of the OCB dimensions should struct meaning post hoc (Schwab, 1980, p. 12). This was our focus
reflect common variance as well as specific variance, and as we in the present study.
pointed out previously, the partitioning of variance in this way is We suggest that if the behavioral dimensions covary only mod-
erately and if there are clear differences between dimensions with
consistent with an aggregate definition of OCB. Indeed, many
respect to relationships in the broader nomological network, then
OCB scholars have combined scores on the behavioral dimensions
it may be appropriate to draw inferences about the behavioral
into a composite score (e.g., Allen & Rush, 1998; Chen, Hui, &
dimensions as if they were separate constructs. If this is the case,
Sego, 1998; Deckop, Mangel, & Cirka, 1999; Hui, Law, & Chen,
the existing dimensions may also be thought of as deficient indi-
1999; Netemeyer, Boles, McKee, & McMurrian, 1997). However,
cators of OCB, and accordingly, inferences about OCB should be
the creation of OCB composites has not been guided by theory or
made only after theory is developed that states how the dimensions
construct definition. Instead, researchers created these composites
combine together as in the aggregate model. However, if the
because they recognized that the behavioral dimensions of OCB
relationships between dimensions approach the magnitude of typ-
covary rather strongly and that combining the scores makes sense
ical reliabilities and if there are no differences between dimensions
with respect to promoting parsimony.
with respect to the relationships with a common set of correlates,
Of course, there is the possibility that OCB is not really a
then it may be that OCB (at least as conceptualized by Organ,
construct at all but instead a useful label for sets of behaviors that
1988) should be thought of as a latent construct. If this is the case,
conceptually belong together (Motowidlo, 2000). This approach then the behavioral dimensions should be thought of as imperfect
seems to have been used in developing early OCB scales. Smith et indicators of OCB and treated as such. Drawing inferences about
al. (1983), for example, began with the definition of OCB and OCB from a score on a single dimension in this case may be
asked managers to identify instances of behavior that fit the problematic to the extent that contamination and unreliability are
definition. The two subsequent dimensions were generated empir- present. Just as important, the tendency to provide substantive post
ically using factor analysis. OCB as a label may also be implied by hoc explanations for differences in findings across dimensions
those who state hypotheses in terms of OCB and draw inferences should be avoided (e.g., Findley, 1995; MacKenzie et al., 1991;
in terms of OCB in the discussion but conduct tests of the hypoth- Moorman & Blakely, 1995; Moorman, Blakely, & Niehoff, 1998;
eses by using measures of individual dimensions (e.g., Konovsky Moorman et al., 1993; Tansky, 1993; VanYperen, van den Berh, &
& Organ, 1996; Podsakoff, Niehoff, et al., 1993). Willering, 1999). If the dimensions of OCB are simply reflections
Finally, there are other studies that consider a specific OCB of the same underlying construct, the true underlying explanation
dimension in isolation. This research generally acknowledges the for differential relationships across dimensions in studies is likely
more common concept of OCB; however, the focus is on devel- to be sampling error, or perhaps differences in the extent to which
oping a better understanding of a behavior that is thought to be the dimensions are imperfect indicators of OCB and not anything
important in a specific work setting. LePine and Van Dyne (1998, substantive. Organ (1994), for example, observed that the empir-
2001a, 2001b; Van Dyne & LePine, 1998), for example, have ical literature suggests that satisfaction may be more strongly
published several articles on specific types of OCB-like behavior. related to the altruism dimension of OCB than to the conscien-
They have studied helping, which is similar to the altruism dimen- tiousness dimension of OCB. He then offered a theoretical expla-
sion of OCB and the interpersonal facilitation dimension of con- nation for this relationship. He specifically suggested that the
textual performance. They have also studied voice, which is sim- personality trait of conscientiousness likely suppresses the rela-
ilar to the civic virtue dimension of OCB, although voice is more tionship between satisfaction and conscientiousness as a form of
about providing suggestions for change rather than keeping abreast OCB. He suggested that the reason for this suppression effect is
of change. The primary focus of their research to date has been to that trait conscientiousness has a direct positive effect on consci-
distinguish voice from cooperative forms of OCB and from task entiousness OCB and a negative effect on satisfaction. We suggest
performance. that if OCB is a latent construct and if the differences in relation-
56 LEPINE, EREZ, AND JOHNSON

ships between dimensions are due to statistical artifacts, attempts theory but did not include measures of OCB that we could include in our
to provide a theoretical explanation for these differences are un- meta-analyses. For example, Eastman (1994) focused on attributions for
necessary. In fact, these post hoc explanations probably hold back OCB, and Crant (1995) validated a proactive personality scale.
the field to the extent that scholars expend resources by empiri- For each of the 113 remaining studies, we listed the relevant OCB
variables along with their correlates. Only 31 of these studies reported
cally assessing them.
relationships among at least two of the five dimensions under consideration
here, and 12 of these studies included relationships among all five dimen-
Summary and a Caveat sions. There are many labels for the behavioral dimensions of OCB, and
when possible, we examined items to ensure that they corresponded to
Although research on OCB has nearly a 20-year history, there is dimension definitions. We excluded from our analysis effects for which, in
ambiguity concerning the definition of OCB and how OCB should our judgment, the items overlapped across dimensions. For example, Smith
be measured. In this study, we attempted to use the existing body et al.’s (1983) altruism scale overlaps with Organ’s (1988) notion of
of research on OCB to provide some insight into how scholars altruism and courtesy. Likewise, although Smith et al.’s generalized com-
have implicitly defined OCB with respect to its dimensions. Our pliance is similar to conscientiousness, several items overlap with Organ’s
hope is that this insight will translate into implications for how definition of civic virtue.
OCB ought to be defined and measured in future theoretical and We decided a priori to consider a correlate in the meta-analysis if it was
related to each of the five dimensions in at least 3 studies. Consequently,
empirical efforts. Next, we describe how we used meta-analytic
the correlates that we considered in this study were satisfaction, organiza-
techniques to assess (a) the relationships among behavioral dimen-
tional commitment, fairness, leader support, and conscientiousness. When
sions of OCB and (b) the relationships between the behavioral a correlate was assessed with multiple measures, we used Hunter and
dimensions and a common set of predictors. Before moving on, Schmidt’s (1990, pp. 451– 463) recommendation for conceptual replica-
however, we need to address an additional issue. We specifically tion. That is, when the measures were highly related, we formed a com-
need to explain why we focused on research that uses the OCB posite measure using the average of the correlations. In addition to the 31
dimensions as suggested by Organ (1988) and not others. studies noted above, there were 6 studies that reported a relationship
It is very difficult to say much about how the dimensions from between a single dimension of organizational citizenship and at least one of
different behavioral frameworks relate to one another in a con- the correlates. Thus, our study considered effect sizes from a total of 37
struct validity sense because scholars have not assessed the extent studies. These studies included 395 effect sizes from a total sample size
of 16,330.
to which newly developed constructs are distinct from similar
Before continuing, we note that of the remaining 76 studies, 26 included
constructs of other existing behavioral frameworks. In fact, be-
a measure of overall organizational citizenship (operationalized as some
cause there are no studies that include similar constructs from composite of the dimensions of OCB). Six other studies considered OCBI
alternative frameworks (e.g., sportsmanship and loyalty), we sim- (altruism and courtesy) and OCBO (sportsmanship, civic virtue, and con-
ply cannot tell how highly the constructs relate to one another. scientiousness) composites, and 14 other studies considered altruism and
Thus, because Organ’s (1988) OCB framework is the only one that generalized compliance as operationalized by Smith et al. (1983). Most of
has been treated consistently over a fairly large number of studies, the remaining studies considered constructs related to OCB, but they did
and because it is highly difficult to confidently map behavioral not correspond precisely to the five dimensions considered here. For
elements from another framework onto Organ’s five dimensions, example, several studies considered the typology of citizenship proposed
we addressed our research questions by using studies grounded in by Van Dyne et al. (1994; social participation, loyalty, obedience, and
functional participation), whereas others considered isolated behaviors
his framework.
such as voice and helping (e.g., LePine & Van Dyne, 1998, 2001b; Van
Dyne & LePine, 1998).
Method
We assessed our questions regarding the dimensionality of OCB first by Meta-Analysis
conducting a comprehensive literature search and then by applying an
appropriate method of meta-analysis. The purpose of meta-analysis is to combine information from various
studies to provide information about relationships of interest. In combining
Literature Search information from independent but similar studies, meta-analysis “borrows
power” from multiple studies to improve on parameter estimates that are
We began our literature search by collecting the articles reviewed in obtained from a single study (Erez, Bloom, & Wells, 1996). As such,
Organ and Ryan’s (1995) OCB meta-analysis. We then conducted a search meta-analysis serves as a powerful tool to model relationships between
of the PsycINFO database for the years 1983 to 1999 to ensure that we had variables of interest that goes beyond those estimates that are obtained
a comprehensive set of published articles. For this search, we used key- from one sample or just a few samples.
words such as organizational citizenship behavior, contextual perfor- One of the questions that this study tries to answer is whether the various
mance, organizational spontaneity, prosocial organizational behavior, and dimensions of OCB are differentially related to other variables of interest
extra-role behavior. We also searched dissertation abstracts for the same (e.g., job satisfaction). To answer this question, a model that can test the
key phrases. We then contacted 23 authors who have published widely in significance of differences in relationships should be used. The random
this area and asked them for unpublished manuscripts or raw data. Finally, effects (RE) method of meta-analysis provides this type of tool by allowing
we manually searched the references in all of these articles for studies that us to expressly model and test for significant differential relationships
we may have missed. Overall, this search yielded 133 empirical articles between the OCB dimensions and a predictor variable. The RE approach,
(including 12 dissertations). Twenty of these studies did not include which parallels regression, uses the entire data set (instead of breaking it
information that could be coded, and they were immediately eliminated into several homogeneous data sets) to explicitly model potential moder-
from further consideration. For example, there were reports of factor ators and test for their significance.
analyses that included only factor loadings (e.g., Becker & Randall, 1994). The general RE meta-analysis model follows the form of ri ! ! " ui "
There were also studies on constructs that used citizenship terminology and ei. In this model, ri represents the observed correlation, ! is analogous to
DIMENSIONALITY OF OCB 57

the grand mean in an analysis of variance and represents the overall variable of interest included in that particular model (e.g., job satisfaction).
population value, ui is the effect of between-study differences on the A statistically significant chi-square, in contrast, would indicate that there
correlation coefficient of study i, and ei is the within-studies error. The is variation in the magnitudes of the effect sizes. At this point, we would
three parameters that are estimated are !, the grand-mean correlation; V, use the second stage of this analysis and test whether the dimensions of
the error variance of the eis; and " 2, the between-studies variance of the uis. OCB moderated the relationship between the grand-mean correlation (!)
A Cochran chi-square test (Cochran, 1937) is used to explicitly assess the and the predictor variable of interest. Stated somewhat differently, we
assumption that all studies are homogeneous with respect to effect size would assess the extent to which the dimensions of OCB explained
(i.e., H0: between-studies variance is zero; " 2 ! 0). Study characteristics variability in the magnitudes of the effect sizes. In a nomological network
that cause variation in population values are represented by the between- sense, such a finding would help support the assertion that the dimensions
studies variance term " 2, and if these characteristics are theoretically are distinct.
important, they can be modeled. This is accomplished by using a covariates To conduct this analysis, we dummy coded the dimensions of OCB, and
model that follows the form of ! ! WiT#. In this model, the Wis represent as in an ordinary regression analysis, the parameter estimates capture the
the different characteristics of the studies that are under investigation as differences in effects between the groups that are coded 1 and a reference
moderators (the superscript T indicates that the matrix of indicator char- group that is coded 0. We used four dummy variables to capture the five
acteristics is transposed). The # represents the estimated effect size of each dimensions, and the altruism dimension served as the reference group. A
Wi characteristic. significant t test for a regression coefficient implies that a specific dimen-
Because raw correlations come from a skewed distribution, we per- sion (e.g., sportsmanship) has a correlation with the variable of interest
formed Hotelling’s and Fisher’s z transformations before conducting any (e.g., job satisfaction) that is different from the correlation of the base
meta-analyses (see Erez et al., 1996, for details). Transforming the corre- group (e.g., altruism) with this variable of interest.
lations normalizes their distribution and stabilizes their variance, which
allowed us to use maximum-likelihood estimation procedures to analyze Results
the data (Bickel & Doksum, 1977). The transformed correlations follow
approximately a normal distribution with mean $ and a known stable error Relationships Among Dimensions
variance of 1/(n # 3) (Bobko, 1995). The meta-analyses of the transformed In the first analysis, we estimated the magnitude of the relation-
correlations were then conducted using Hierarchical Linear Modeling ships among the OCB dimensions at the population level. We
(HLM) software (Byrk, Raudenbush, & Congdon, 1996). The transformed
analyzed uncorrected correlations and correlations corrected for
correlations are entered into HLM as a “variance known” problem struc-
ture. Although there are many different ways to conduct an RE meta-
unreliability of both the predictor and the criterion. We corrected
analysis (e.g., Erez et al., 1996; Raudenbush, 1994; Wagner & LePine, correlations for unreliability before we transformed them to nor-
1999), the HLM software includes convenient features that allow research- malize their distributions. The HLM program produced the esti-
ers to assess the relative fit of alternative models. mated population correlation (!), the standard error of this corre-
In the first set of analyses, an unconditional model (without covariates) lation (SE), the between-studies variance (" 2), and the chi-square
was assessed for each relationship among the OCB dimensions. These test for the significance of this variance. Table 1 shows the
analyses were conducted to estimate the effect sizes for the relationships uncorrected (r) and corrected (rc) population correlations among
among the OCB dimensions at the population level (grand-mean $). The dimensions as well as the 95% confidence intervals around the
grand-mean $s were then back-transformed into normal correlations (!s) population correlations. The confidence intervals were created by
using the formula recommended by Erez et al. (1996, p. 289). In addition, adding $1.96 SE to the population correlations.
the analyses produced estimates of the variance across effect sizes (" 2) for
As Table 1 shows, all but one of the corrected correlation
each relationship and a chi-square that tested the hypothesis that this
estimates (r ! .40 for sportsmanship and civic virtue) were strong.
variance was zero (" 2 ! 0). A significant chi-square indicates the presence
of variability in the effect sizes that could potentially be explained by
In addition, most of the corrected correlations approached the
theoretically relevant moderators (e.g., the proportion of female partici- magnitude of acceptable internal consistency reliability (average
pants in the study samples, type of job or task). Although the information rc ! .67). In fact, all but two of the 95% confidence intervals (the
needed to assess these types of moderators was often not provided in the correlations for sportsmanship and civic virtue and for sportsman-
original studies, we report the " 2 and chi-square statistics for the sake of ship and conscientiousness) included values that exceeded .70, the
completeness and in the hope that scholars begin to more completely generally accepted minimum value for internal consistency reli-
describe their sample and setting in the future. ability (Nunnally, 1978). Thus, with the possible exception of
In the second set of analyses, we examined relationships between OCB sportsmanship, it appears that the behavioral dimensions were
dimensions and predictors that have been considered in empirical studies of highly related to one another and that the relationships closely
OCB. These analyses were conducted in two stages. In the first stage, the approached or exceeded the typical levels of reliability.
transformed correlations representing the relationship between a predictor
We do note that for each relationship, the " 2 was statistically
(e.g., job satisfaction) and all the OCB dimensions were submitted to a
significant, which suggests that the correlations among the OCB
single unconditional model. The purpose of this analysis was to examine
whether there was variation in the effect sizes among the different OCB dimensions may not be homogeneous. It certainly would be inter-
dimensions and the variable of interest. As with the previous analysis, the esting to identify theoretically relevant moderators that might
variance of the effect sizes (" 2) was evaluated using the chi-square test of explain this variability. However, this type of analysis was not
significance. If the effect sizes did not differ significantly from one feasible because the studies did not consistently include enough of
another, there would be no variance in the correlations to be explained by the needed information (e.g., percentage of female participants in
the dimensions of OCB. This would mean that the data do not support the sample, interdependence of the tasks, task complexity).
distinctions among the OCB dimensions with respect to relationships with
the other variable of interest (e.g., job satisfaction). The grand-mean Relationships With Predictors
correlation (!), in this case, would represent the overall relationship be-
tween all the dimensions combined (which could be thought of as overall In the second analysis, we assessed whether the OCB dimen-
OCB as measured by somewhat imperfect dimensional indicators) and the sions were related differentially to predictors that have been used
58 LEPINE, EREZ, AND JOHNSON

Table 1
Population Correlations Among Organizational Citizenship Behavior Dimensions

Dimension 1 2 3 4 5

1. Altruism —
2. Civic virtue —
r (95% CI) .58 (.52–.63)
rc (95% CI) .75 (.66–.83)
K (N) 16 (5,160)
"2 .023**
3. Conscientiousness —
r (95% CI) .58 (.52–.63) .49 (.41–.57)
rc (95% CI) .72 (.63–.79) .64 (.52–.74)
K (N) 26 (7,356) 15 (4,901)
"2 .037** .035**
4. Courtesy —
r (95% CI) .67 (.62–.72) .54 (.49–.58) .57 (.51–.63)
rc (95% CI) .87 (.67–.95) .68 (.62–.74) .70 (.62–.77)
K (N) 15 (5,085) 15 (5,085) 15 (5,086)
"2 .031** .012** .023**
5. Sportsmanship —
r (95% CI) .47 (.39–.55) .34 (.25–.42) .45 (.38–.50) .58 (.50–.65)
rc (95% CI) .68 (.33–.87) .40 (.24–.54) .55 (.48–.61) .70 (.60–.78)
K (N) 18 (5,718) 20 (6,694) 17 (5,423) 15 (5,085)
"2 .048** .045** .019** .046**

Note. r ! uncorrected population correlation; 95% CI ! 95% confidence interval around the population
correlation; rc ! population correlation corrected for unreliability; K ! number of studies; N ! total sample size;
" 2 ! between-studies variance.
** p % .01.

most often in the literature. Table 2 shows the results of the An Alternative Framework
unconditional RE models. The coefficients in columns 2 and 3 in
Table 2 can be thought of as the relationships (uncorrected and As we mentioned earlier, some researchers have proposed mod-
corrected for attenuation due to unreliability) between each pre- els of OCB that include behavioral dimensions that are somewhat
dictor (i.e., satisfaction, commitment, fairness, leader support, and broader than those included in Organ’s (1988) framework. For
conscientiousness) and overall OCB. As shown in Table 2, the example, Coleman and Borman’s (2000) interpersonal citizenship
corrected population estimates ranged from .20 for commitment to performance dimension includes altruism and courtesy elements,
and their organizational citizenship performance dimension in-
.32 for leader support. We note that these effect sizes are slightly
cludes sportsmanship, civic virtue, and conscientiousness ele-
smaller than those reported by Organ and Ryan (1995) in their
ments. These broader dimensions are consistent with those pro-
meta-analysis of the altruism and generalized compliance dimen-
posed by Williams and Anderson (1991), who believe that it is
sions of OCB. However, this difference is understandable given
important to distinguish OCB with respect to the target of the
our consideration of dimensional measures that are narrower in behavior. They specifically distinguished OCBI (altruism and
breadth than those considered by Organ and Ryan. courtesy) from OCBO (sportsmanship, civic virtue, and conscien-
The right-hand column of Table 2 shows that for all the predic- tiousness). We thought that perhaps these broader dimensions
tors, the " 2 was statistically significant, suggesting situational might explain the variability between studies in the predictor–
specificity. In other words, for each predictor variable of interest criterion relationships. Accordingly, we created a dummy variable
(i.e., satisfaction, commitment, fairness, leader support, and con- to capture the OCBI versus OCBO distinction and used it in
scientiousness), relationships with OCB were not the same across another RE meta-analysis.
the set of study effect sizes. This result, of course, indicates the Table 4 presents the results of this analysis and shows that the
potential existence of moderators. relationships with OCBI were not significantly different than those
We investigated whether the OCB dimensions explained any of with OCBO for any of the predictor variables. Thus, here again,
the variance in these relationships by assessing the effects of the there does not seem to be any evidence of differential relationships
four dummy-coded variables that captured the five OCB dimen- with predictors across dimensions of OCB.
sions. The results of this analysis are shown in Table 3 and indicate
no significant effects for the dimensions for any predictor vari- 1
The same multiple comparison analyses were conducted using the
able.1 Overall, our results suggest that the relationship between the other four dimensions as the base group. Because the results of these
predictors (i.e., satisfaction, commitment, fairness, leader support, comparisons were substantively identical to those using altruism as the
and conscientiousness) and OCB does not depend on how OCB is base group (i.e., no variance was explained by the set of dummy codes),
behaviorally defined. they are not reported here.
DIMENSIONALITY OF OCB 59

Table 2
Relationship Between Organizational Citizenship Behavior and Satisfaction, Commitment,
Fairness, Leader Support, and Conscientiousness

Variable r (95% CI) rc (95% CI) Kd, Kc, N "2

Satisfaction .20 (.18–.22) .24 (.22–.26) 22, 72, 7,100 .003**


Commitment .17 (.15–.20) .20 (.17–.24) 17, 54, 5,133 .006**
Fairness .20 (.17–.22) .23 (.20–.26) 9, 40, 1,975 .003**
Leader support .25 (.22–.28) .32 (.27–.37) 10, 41, 4,349 .007**
Conscientiousness .19 (.12–.26) .23 (.32–.44) 3, 15, 848 .015**

Note. r ! uncorrected population correlation; 95% CI ! 95% confidence interval around the population
correlation; rc ! population correlation corrected for unreliability; Kd ! number of different studies; Kc !
number of correlations; N ! total sample size; " 2 ! between-studies variance.
** p % .01.

Contrasting Overall OCB With Dimensional Measures OCB and measures of OCB dimensions (dimensional measures).
Dummy coding was used to extract these differences, with the
In our literature search, we found 19 additional studies with overall measure of OCB coded as the reference group. A positive
correlations between predictors and a measure of overall OCB significant difference between the dimensional measure of OCB
(typically formed by averaging items across dimensions). The and the measure of overall OCB means that measuring OCB, as a
existence of these effect sizes allowed us to test for significant specific behavioral dimension, increases predictive validity. In
differences in predictive validity between overall measures of contrast, if there is no difference or if the difference is negative,

Table 3
Conditional Model Estimates for Differences Between the Organizational
Citizenship Behavior Dimensions

Uncorrected Corrected
Fixed effect coefficient coefficient

Satisfaction (Kd ! 22, Kc ! 72, N ! 7,100)


Intercept (altruism) .19** .23**
Civic virtue #.01 .01
Conscientiousness .00 .00
Courtesy .01 .02
Sportsmanship .02 .03
Commitment (Kd ! 17, Kc ! 54, N ! 5,133)
Intercept (altruism) .18** .21**
Civic virtue .00 .03
Conscientiousness .01 .01
Courtesy #.05 #.06
Sportsmanship #.03 #.04
Fairness (Kd ! 9, Kc ! 40, N ! 1,975)
Intercept (altruism) .19** .23**
Civic virtue #.02 #.02
Conscientiousness #.02 #.02
Courtesy .03 .02
Sportsmanship .03 .03
Leader support (Kd ! 10, Kc ! 41, N ! 4,349)
Intercept (altruism) .27** .35**
Civic virtue #.08 #.08
Conscientiousness #.02 #.03
Courtesy #.01 #.02
Sportsmanship #.01 #.02
Conscientiousness (Kd ! 3, Kc ! 15, N ! 848)
Intercept (altruism) .23* .29*
Civic virtue .02 .02
Conscientiousness #.06 #.09
Courtesy #.06 #.08
Sportsmanship #.11 #.14

Note. Altruism was the base category. All other coefficients represent the difference between each variable and
the altruism base group. Kd ! number of different studies; Kc ! number of correlations; N ! total sample size.
* p % .05. ** p % .01.
60 LEPINE, EREZ, AND JOHNSON

Table 4
Conditional Model Estimates for Differences Between the OCBI and OCBO Dimensions

Uncorrected Corrected
Fixed effect coefficient coefficient

Satisfaction (Kd ! 22, Kc ! 72, N ! 7,100)


Intercept (OCBI) .20** .24**
OCBO .00 .01
Commitment (Kd ! 17, Kc ! 54, N ! 5,133)
Intercept (OCBI) .19** .21**
OCBO #.02 #.02
Fairness (Kd ! 9, Kc ! 40, N ! 1,975)
Intercept (OCBI) .19** .22**
OCBO .02 .02
Leader support (Kd ! 10, Kc ! 41, N ! 4,349)
Intercept (OCBI) .24** .30**
OCBO .03 .03
Conscientiousness (Kd ! 3, Kc ! 15, N ! 848)
Intercept (OCBI) .18** .22**
OCBO .02 .03

Note. Organizational citizenship behavior directed toward individuals (OCBI) was the base category. The
organizational citizenship behavior directed toward the organization (OCBO) coefficients represent the differ-
ences between the OCBI and OCBO groups. Kd ! number of different studies; Kc ! number of correlations;
N ! total sample size.
** p % .01.

there is no advantage or there is a disadvantage in measuring statistically significant. Whereas the uncorrected correlation be-
specific OCB dimensions. tween commitment and overall OCB was .27, the correlation for
The results of these comparisons are reported in Table 5 and the courtesy dimension was .13, and the correlation for the sports-
showed only one statistically significant difference between the manship dimension was .15. The correlations with commitment
overall measure and dimensional measures. This difference was in and the other dimensional measures were also lower than the
the relationship with commitment, and the direction of this effect correlation for the overall measure; however, these differences
was negative. We explored which specific dimensions were the were not statistically significant.2 The dimensional measure
cause of this effect and found only two differences that were dummy-coded coefficients were also negative (but not statistically
significant) when the predictor was satisfaction, fairness, and
leader support. Overall, it appears that predictive relationships
Table 5 with the broader OCB criterion were as good as, or superior to,
Conditional Model Estimates for Differences Between Studies those with the narrower dimensional criteria.
Measuring Overall Organizational Citizenship Behavior (OCB)
and Studies Measuring OCB Dimensions
Discussion
Uncorrected Corrected
Fixed effect coefficient coefficient Kd, Kc, N Although scholars suggest that OCB is composed of conceptu-
ally distinct behavioral dimensions, we have shown that these
Satisfaction
Overall OCB .22** .31** 10, 10, 9,248 dimensions have yet to be distinguished from one another in the
Dimensions OCB #.05 #.07 22, 72, 7,100 empirical literature beyond factor analysis. The results of our
Commitment analyses suggest that most of the dimensions of OCB, at least those
Overall OCB .27** .32** 9, 9, 2,228 conceptualized by Organ (1988), are highly related to one another
Dimensions OCB #.10* #.12* 17, 54, 5,133
Fairness
and that there are no apparent differences in relationships with the
Overall OCB .24** .31** 8, 8, 8,154 most popular set of predictors.
Dimensions OCB #.05 #.08 9, 40, 1,975
Leader support
Overall OCB .30** .41** 2, 2, 6,961 OCB as a Latent Construct
Dimensions OCB #.05 #.09 10, 41, 4,349
Conscientiousness Overall, our results seem to be consistent with the notion that the
Overall OCB .11** .13** 4, 4, 619 five dimensions of OCB are not much more than equivalent
Dimensions OCB .08 .10 3, 15, 848

Note. Overall OCB conveys the correlation for studies measured with an 2
overall OCB measure. Dimensions OCB conveys how much the dimen- The same type of analyses to find differences between the overall
sional measure adds to or subtracts from the correlations using overall measure of OCB and the unique dimensions were conducted for all the
OCB measures. Kd ! number of different studies; Kc ! number of other predictors (i.e., satisfaction, fairness, leader support, and conscien-
correlations; N ! total sample size. tiousness). However, because none of the differences were significant, they
* p % .05. ** p % .01. are not reported here.
DIMENSIONALITY OF OCB 61

indicators of OCB. From a theoretical standpoint, these results different dimensions. For example, peers might be best suited to
suggest that scholars should begin to explicitly think of Organ’s rate those behaviors that fit in the domain of interpersonal citizen-
(1988) OCB as a latent construct. Given that the five dimensions ship performance (or OCBI), whereas supervisors might be best
seem to be behavioral manifestations of positive cooperativeness suited to rate those behaviors that fit in the domains of organiza-
at work, perhaps this latent OCB construct should be redefined as tional citizenship performance (or OCBO) and job–task citizenship
a general tendency to be cooperative and helpful in organizational performance.
settings (Motowidlo, 2000; Organ, 1997). We note that this con- Inadequate domain of predictors or OCB criteria. The focus
ceptualization is somewhat consistent with the behavioral elements of almost all OCB research to date has been either to distinguish
included in the affiliative–promotive dimension of extrarole be- OCB from task performance or to identify predictors of OCB.
havior as proposed by Van Dyne et al. (1995). Because the focus has not been on differentiating the behavioral
From a research method standpoint, a latent definition of OCB dimensions from one another, researchers have not invested effort
means that Organ’s (1988) five dimensions should be thought of as in identifying predictors with differential relationships. Thus, it is
somewhat imperfect indicators of the same underlying construct. possible that there are certain unidentified (or not often studied)
Consistent with this idea, when OCB is the focal construct of variables that differentially predict OCB dimensions. Researchers
interest, scholars should avoid focusing on the specific dimensions should use theory to generate hypotheses that specifically attempt
of OCB when conducting research and interpreting results. This to identify antecedents that have differential relationships across
would be like interpreting relationships with individual items from dimensions.
a multi-item measure of a unidimensional construct. Interpreting A related plausible explanation for our findings is that perhaps
differential relationships among indicators is problematic because the domain of OCB dimensions is incomplete (Coleman & Bor-
observed differences are likely nothing more than sampling error man, 2000). There may be dimensions of OCB that have yet to be
or a reflection of the relative imperfection inherent in individual identified but, nevertheless, would have smaller relationships with
indicators. Measurement and analysis of OCB, as a latent con- other dimensions and would be predicted by a different set of
struct, should be commensurate with its status as such (Nunnally & antecedents. It might be worthwhile to conduct studies aimed at
Bernstein, 1994). A measure of OCB, as a latent construct, should identifying a broader set of behavioral dimensions that fit the
include items that fully capture the elements implied within the definition of OCB than the five dimensions proposed by Organ
boundaries of the definition and should not include items that (1988). Some of these dimensions might include adaptive perfor-
capture elements outside the boundaries of the definition. How- mance (Hesketh & Neal, 1999; LePine, 2001; LePine, Colquitt, &
ever, there are several possible alternative explanations for our Erez, 2000), functional participation (e.g., Van Dyne et al., 1994),
results, and researchers should examine these explanations before personal initiative (Frese, Kring, Soose, & Zempel, 1996), voice
our findings are used to guide the development of OCB theory and behavior (LePine & Van Dyne, 1998, 2001b; Van Dyne & LePine,
empirical research. 1998), and the elements of job dedication related more directly to
Correlated method variance. One possible explanation for the the task (e.g., Coleman & Borman, 2000; Van Scotter & Motow-
pattern of meta-analytic results is that researchers have measured idlo, 1996).
the five dimensions with the same instrument and rater (normally Low statistical power. Another alternative explanation for our
the supervisor) at the same point in time. Correlated method findings relates to statistical power. Specifically, we may have
variance inflates observed relationships between dimensions and failed to find differences in relationships across dimensions be-
makes it more difficult to uncover differential relationships be- cause the number of effect sizes in some of the analyses was too
tween specific predictors across dimensions. As an example, Mo- low to detect the differences. This problem is most relevant to the
towidlo and Van Scotter (1994) found a correlation of .17 between analyses with trait conscientiousness (see Table 3) because there
task and contextual performance using different supervisor raters were only 15 effect sizes to meta-analyze. Although the ratio of the
to measure these two constructs. Conway (1996), in contrast, number of effect sizes to the number of variables (the four dimen-
found a correlation of .55 between the two constructs when the sion dummy-coded variables) was greater than 10:1 for the other
same raters were used to measure the constructs. In a similar predictors, statistical power is an issue that readers always need to
manner, Van Dyne and LePine (1998) found a correlation of .36 bear in mind when interpreting results such as ours. Our hope is
between peer-rated helping behavior and supervisor-rated task that the research we report here triggers the type of additional work
performance. However, the correlation between helping behavior necessary to address this limitation.
and task performance was .65 when both variables were measured
by the same source (either peers or supervisors). To eliminate
correlated method variance as an explanation for our results, OCB as an Aggregate Construct
research needs to be conducted in which the OCB dimensions are
rated using different sources and/or methods. As one reviewer In light of the preceding discussion, what can we conclude about
suggested, the best estimate of the between-dimension relation- the appropriate way to represent OCB? On the one hand, there
ships might be to measure the dimensions using different methods seems to be enough evidence to at least consider conceptualizing
but to correct the correlations for unreliability of each method. OCB as a latent construct. On the other hand, there are limitations
Research using this approach may be most important in situations with the existing empirical research, and thus, the evidence is not
in which scholars are interested in distinguishing between dimen- such that it can be used as a basis for categorically rejecting other
sions with respect to relationships with other variables in the models as viable options.
nomological network. Scholars need to keep in mind that theory In fact, as we described earlier, the definition of OCB has
and logic should guide the choice of methods for measuring the evolved over time and has become more similar to the definition of
62 LEPINE, EREZ, AND JOHNSON

contextual performance, an aggregate construct. If, indeed, this is Other Issues


the future direction of the OCB construct, then OCB scholars need
In conducting this study, it became clear to us that there are
to ensure that measurement is consistent with an aggregate per-
several characteristics of OCB research that hinder the systematic
formance definition. This means that OCB scholars should prob-
accumulation of knowledge. Perhaps most important, scholars
ably avoid measuring OCB by asking respondents how many
have introduced many new OCB (or OCB-like) constructs and
instances of a specific behavior from a standard pool of items are
measures without adequate construct validity support. In the pool
typically enacted (e.g., counting the number of instances of helping
of 133 studies, we found more than 40 different measures of
or obtaining ratings of the frequency with which an employee behavior that scholars have referred to as OCB or something very
helps other employees). Instead, researchers should measure OCB similar (e.g., contextual performance, extrarole behavior). Almost
in a manner that is consistent with how behavioral performance without exception, however, the authors of these studies did not
constructs are measured. That is, they should identify activities empirically evaluate the extent to which the OCB constructs they
that contribute positively and negatively to the organization (albeit, introduced were different than or similar to others that already
the organization’s social and psychological context) and then existed. This shortcoming, together with the fact that full scales
obtain ratings of how likely it is that an employee would engage in were rarely reported, made it impossible to use relationships with
those behaviors (e.g., Motowidlo & Van Scotter, 1994; Van Scot- these newer dimensions in our meta-analysis.
ter & Motowidlo, 1996). The score for OCB would then be some We also note that in the pool of 133 studies, researchers con-
aggregate of the scores on those activities. One benefit of this sidered more than 200 different correlates from a wide variety of
approach is that it helps prevent situations in which a score on different theoretical perspectives. To some extent, this type of
OCB is incongruent with its definition—the value to the organi- breadth is encouraging because it suggests that OCB researchers
zation of those activities that contribute to the organization’s social are not overly constrained by a dominant paradigm that hinders the
or psychological context. For example, scores for employees on development of new and useful ways of thinking about OCB. The
organizational citizenship as measured (at least in part) by a count downside of this breadth, however, is that there has been less effort
of “helping behaviors” are likely to be low when individuals in a focused on replicating and conducting studies that systematically
particular setting work independently, when individuals are uni- extend previous empirical research. The majority of studies of
formly competent in their jobs, or when task demands are predict- OCB include correlates that are unique to that particular study.
able. However, these “low” scores on OCB will be quite mislead- This is why, in our meta-analysis, we could consider only 5 OCB
ing if the employees routinely engage in other activities that do correlates. One avenue for future research might be to develop a
positively affect the psychological and social context of the orga- meta-theory of OCB that explicitly acknowledges the different
nization (e.g., not complaining about working conditions, suggest- types of correlates. This way, empirical research on OCB might
ing administrative improvements). accumulate in a more systematic fashion.
A second and related reason that might weigh against an
exclusively latent definition of OCB is that the behavioral Summary and Conclusion
dimensions of OCB seem to be conceptually distinct. Assuming On the basis of the meta-analyses reported in this article, it
that there may be different organizationally relevant conse- appears that the relationships among the dimensions of OCB are
quences of these behaviors, it may be appropriate to consider generally as high as reliability estimates and that there are no
them individually. For example, it is not difficult to imagine meaningful differences in relationships with predictors across di-
someone who tends to be very helpful and cooperative toward mensions. Accordingly, the dimensions of OCB as currently op-
others (high altruism) yet at the same time tends to be hesitant erationalized by the majority of scholars perhaps may be best
about getting involved in the decisions that affect the organi- thought of as imperfect indicators of OCB. Consistent with this
zation (low civic virtue). The former behavior is likely to be idea, it might be worthwhile to begin explicitly defining OCB as a
more important to organizations in which members work in latent construct. To date, such a conceptualization has not been
small groups or teams. The latter behavior is likely to be more explicitly stated in the literature. We offered potential alternative
important in organizations in which members’ active participa- explanations for our results. Correlated method variance and in-
tion is critical to the development of effective responses to complete sets of predictors or criteria stand as plausible explana-
competitive threats. Indeed, although there are only a few tions. Because the definition of OCB has evolved to be more
studies that have considered consequences of OCB, relation- consistent with contextual performance (an aggregate construct),
ships may differ across dimensions, and some dimensions may and because there may be practical reasons to separately consider
have independent effects on the same outcome (Podsakoff et al., the behavioral dimensions, we urge scholars to assess these alter-
2000). native perspectives as well. Overall, there needs to be an increase
In sum, although we found some evidence to support the in the amount of effort focused on developing theory that can
latent definition of OCB, there are reasons to consider the guide OCB measurement and analysis. Such theory should explic-
itly define OCB with respect to its dimensions. We hope that this
aggregate model as a viable alternative. Future research should
article serves as an impetus for such effort.
focus on teasing out the relative advantages and disadvantages
of these two approaches. At the very least, scholars should
explicitly articulate their definition of OCB with respect to the References
dimensions and ensure that measurement is consistent with their References marked with an asterisk indicate studies included in one of
definition. the meta-analyses.
DIMENSIONALITY OF OCB 63

*Allen, T. D., & Rush, M. C. (1998). The effects of organizational ratings: The role of subordinate–supervisor conscientiousness similarity.
citizenship behavior on performance judgments: A field study and a Group and Organization Management, 23, 189 –216.
laboratory experiment. Journal of Applied Psychology, 83, 247–260. Eastman, K. (1994). In the eyes of the beholder: An attributional approach
Barnard, C. (1938). The functions of the executive. Cambridge, MA: to ingratiation and organizational citizenship behavior. Academy of
Harvard University Press. Management Journal, 37, 1379 –1391.
*Bateman, T. S., & Organ, D. W. (1983). Job satisfaction and the good Erez, A., Bloom, M. C., & Wells, M. T. (1996). Using random rather than
soldier: The relationship between affect and employee “citizenship.” fixed effects models in meta-analysis: Implications for situational spec-
Academy of Management Journal, 26, 587–595. ificity and validity generalization. Personnel Psychology, 49, 275–306.
*Becker, T. E., & Billings, R. S. (1993). Profiles of commitment: An *Fahr, J., Earley, P. C., & Lin, S. (1997). Impetus for action: A cultural
empirical test. Journal of Organizational Behavior, 14, 177–190. analysis of justice and organizational citizenship behavior in Chinese
Becker, T. E., & Randall, D. M. (1994). Validation of a measure of society. Administrative Science Quarterly, 42, 421– 444.
organizational citizenship behavior against an objective behavioral cri- *Findley, H. M. (1995). Fairness in social exchange: The relationship of
terion. Educational and Psychological Measurement, 54, 160 –167. leader–member exchange and performance appraisal fairness to orga-
Bickel, P. J., & Doksum, K. A. (1977). Mathematical statistics. San nizational citizenship behavior. Unpublished doctoral dissertation, Lou-
Francisco: Holden-Day. isiana State University, Baton Rouge.
*Bies, R. J., Martin, C. L., & Brockner, J. (1993). Just laid off, but still a Frese, M., Kring, W., Soose, A., & Zempel, J. (1996). Personal initiative at
“good citizen?” Only if the process is fair. Employee Responsibilities work: Differences between East and West Germany. Academy of Man-
and Rights Journal, 6, 227–238. agement Journal, 39, 37– 63.
*Blakely, G. L., Fuller, J., & Smith, D. (1996, November). Are chameleons George, J. M. (1990). Personality, affect, and behavior in groups. Journal
good citizens? The relationship between self-monitoring and organiza- of Applied Psychology, 75, 107–116.
tional citizenship behavior. Paper presented at the annual meeting of the George, J. (1991). State or trait: Effects of positive mood on prosocial
Southern Management Association, New Orleans, LA. behaviors at work. Journal of Applied Psychology, 76, 299 –307.
Bobko, P. (1995). Correlation and regression: Principles and applications George, J. M., & Bettenhausen, K. (1990). Understanding prosocial be-
for industrial psychology and management. New York: McGraw-Hill. havior, sales performance, and turnover: A group-level analysis in a
service context. Journal of Applied Psychology, 75, 698 –709.
Borman, W. C., & Motowidlo, S. J. (1993). Expanding the criterion
George, J. M., & Brief, A. P. (1992). Feeling good– doing good: A
domain to include elements of contextual performance. In N. Schmitt &
conceptual analysis of the mood at work– organizational spontaneity
W. C. Borman (Eds.), Personnel selection in organizations (pp. 71–98).
relationship. Psychological Bulletin, 112, 310 –329.
San Francisco: Jossey-Bass.
George, J. M., & Jones, G. R. (1997). Organizational spontaneity in
Borman, W. C., & Motowidlo, S. J. (1997). Task performance and con-
context. Human Performance, 10, 153–170.
textual performance: The meaning for personnel selection research.
*Graham, J. W., & Van Dyne, L. (1998). Civic virtue: Contrasting two
Human Performance, 10, 99 –109.
forms of civic virtue based on political philosophy. Unpublished
Borman, W. C., White, L. A., & Dorsey, D. W. (1995). Effects of ratee task
manuscript.
performance and interpersonal factors on supervisor and peer perfor-
*Hatcher, L., Ross, T. L., & Collins, D. (1989). Prosocial behavior, job
mance ratings. Journal of Applied Psychology, 80, 168 –177.
complexity, and suggestion contribution under gainsharing plans. Jour-
Brief, A. P., & Motowidlo, S. J. (1986). Prosocial organizational behaviors.
nal of Applied Behavioral Science, 25, 231–248.
Academy of Management Review, 11, 710 –725.
*Hattrup, K., O’Connell, M. S., & Wingate, P. H. (1998). Prediction of
Byrk, A. S., Raudenbush, S. W., & Congdon, R. T., Jr. (1996). HLM:
multidimensional criteria: Distinguishing task and contextual perfor-
Hierarchical linear and nonlinear modeling with the HLM/2L and
mance. Human Performance, 11, 305–319.
HLM/3L programs. Chicago: Scientific Software International. Hesketh, B., & Neal, A. (1999). Technology and performance. In D. R.
*Carson, K. D., & Carson, P. P. (1998). Career commitment, competen- Ilgen & E. D. Pulakos (Eds.), The changing nature of performance:
cies, and citizenship. Journal of Career Assessment, 6, 195–208. Implications for staffing, motivation, and development (pp. 21–55). San
*Chen, X., Hui, C., & Sego, D. J. (1998). The role of organizational Francisco: Jossey-Bass.
citizenship behavior in turnover: Conceptualization and preliminary tests Howard, A. (1995). The changing nature of work. San Francisco: Jossey-
of key hypotheses. Journal of Applied Psychology, 83, 922–931. Bass.
Cochran, W. G. (1937). Problems arising in the analysis of a series of Hui, C., Law, K. S., & Chen, Z. X. (1999). A structural equation model of
similar experiments. Journal of the Royal Statistical Society, 4(Suppl.), the effects of negative affectivity, leader–member exchange, and per-
102–118. ceived job mobility on in-role and extra-role performance: A Chinese
Coleman, V. I., & Borman, W. C. (2000). Investigating the underlying case. Organizational Behavior and Human Decision Processes, 77,
structure of the citizenship performance domain. Human Resource Man- 3–21.
agement Review, 10, 25– 44. Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis. Newbury
Conway, J. M. (1996). Additional construct validity evidence for the Park, CA: Sage.
task/contextual performance distinction. Human Performance, 9, 309 – Ilgen, D. R., & Pulakos, E. D. (1999). The changing nature of perfor-
329. mance: Implications for staffing, motivation, and development. San
Crant, M. J. (1995). The Proactive Personality Scale and objective job Francisco: Jossey-Bass.
performance among real estate agents. Journal of Applied Psychol- Katz, D. (1964). The motivational basis of organizational behavior. Be-
ogy, 80, 532–537. havioral Science, 9, 131–146.
*Deckop, J. R., Mangel, R., & Cirka, C. C. (1999). Getting more than you *Kemery, E. R., Bedeian, A. G., & Zacur, S. R. (1996). Expectancy-based
pay for: Organizational citizenship behavior and pay-for-performance job cognitions and job affect as predictors of organizational citizenship
plans. Academy of Management Journal, 42, 420 – 428. behaviors. Journal of Applied Social Psychology, 26, 635– 651.
*Deluga, R. J. (1995). The relationship between attributional charismatic *Kidwell, R. E., Jr., Mossholder, K. W., & Bennett, N. (1997). Cohesive-
leadership and organizational citizenship behavior. Journal of Applied ness and organizational citizenship behavior: A multilevel analysis using
Social Psychology, 28, 1652–1669. work groups and individuals. Journal of Management, 23, 775–793.
*Deluga, R. J. (1998). Leader–member exchange quality and effectiveness *Konovsky, M. A., Elliott, J., & Pugh, S. D. (1995, August). The dispo-
64 LEPINE, EREZ, AND JOHNSON

sitional and contextual predictors of citizenship behavior in Mexico. behavior: The importance of the employee’s perspective. Academy of
Paper presented at the annual meetings of the Academy of Management, Management Journal, 37, 1543–1567.
Vancouver, British Columbia, Canada. *Motowidlo, S. J. (1984). Does job satisfaction lead to consideration and
*Konovsky, M. A., & Organ, D. W. (1996). Dispositional and contextual personal sensitivity? Academy of Management Journal, 27, 910 –915.
determinants of organizational citizenship behavior. Journal of Organi- Motowidlo, S. J. (2000). Some basic issues related to contextual perfor-
zational Behavior, 17, 253–266. mance and organizational citizenship behavior in human resource man-
*Konovsky, M. A., & Pugh, S. D. (1994). Citizenship behavior and social agement. Human Resource Management Review, 10, 115–126.
exchange. Academy of Management Journal, 37, 656 – 669. Motowidlo, S. J., Borman, W. C., & Schmit, M. J. (1997). A theory of
Law, K. S., & Wong, C. (1999). Multidimensional constructs in structural individual differences and contextual performance. Human Perfor-
equation analysis: An illustration using the job perception and job mance, 10, 71– 83.
satisfaction constructs. Journal of Management, 25, 143–160. Motowidlo, S. J., & Schmit, M. J. (1999). Performance assessment in
Law, K. S., Wong, C. S., & Mobley, W. H. (1998). Toward a taxonomy of unique jobs. In D. R. Ilgen & E. D. Pulakos (Eds.), The changing nature
multidimensional constructs. Academy of Management Review, 23, 741– of performance: Implications for staffing, motivation, and development
755. (pp. 56 – 87). San Francisco: Jossey-Bass.
LePine, J. A. (2001). Team adaptation and postchange performance: Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that task perfor-
Effects of team composition in terms of members’ cognitive ability and mance should be distinguished from contextual performance. Journal of
personality. Manuscript submitted for publication. Applied Psychology, 79, 475– 480.
LePine, J. A., Colquitt, J. A., & Erez, A. (2000). Adaptability to changing *Munene, J. C. (1995). “Not-on-seat”: An investigation of some correlates
task contexts: Effects of general cognitive ability, conscientiousness, and of organisational citizenship behavior in Nigeria. Applied Psychology:
openness to experience. Personnel Psychology, 53, 563–593. An International Review, 44, 111–122.
LePine, J. A., Hanson, M., Borman, W., & Motowidlo, S. J. (2000). *Netemeyer, R. G., Boles, J. S., McKee, D. O., & McMurrian, R. (1997).
Contextual performance and teamwork: Implications for staffing. In An investigation into the antecedents of organizational citizenship be-
G. R. Ferris & K. M. Rowland (Eds.), Research in personnel and human haviors in a personal selling context. Journal of Marketing, 61, 85–98.
resources management (Vol. 19, pp. 53–90). Stamford, CT: JAI Press. *Neuman, G. A., & Kickul, J. R. (1998). Organizational citizenship be-
*LePine, J. A., & Van Dyne, L. (1998). Predicting voice behavior in work haviors: Achievement orientation and personality. Journal of Business
groups. Journal of Applied Psychology, 83, 853– 868.
and Psychology, 13, 263–279.
LePine, J. A., & Van Dyne, L. (2001a). An attributional model of helping
Niehoff, B. P., & Moorman, R. H. (1993). Justice as a mediator of the
in the context of work groups. Academy of Management Review, 26,
relationship between methods of monitoring and organizational citizen-
67– 84.
ship behavior. Academy of Management Journal, 36, 527–556.
*LePine, J. A., & Van Dyne, L. (2001b). Voice and cooperative behavior
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York:
as contrasting forms of contextual performance: Evidence of differential
McGraw-Hill.
relationships with Big Five personality characteristics and cognitive
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.).
ability. Journal of Applied Psychology, 86, 326 –336.
New York: McGraw-Hill.
*MacKenzie, S. B., Podsakoff, P. M., & Fetter, R. (1991). Organizational
O’Reilly, C., III, & Chatman, J. (1986). Organizational commitment and
citizenship behavior and objective productivity as determinants of man-
psychological attachment: The effects of compliance, identification, and
agerial evaluations of salespersons’ performance. Organizational Behav-
internalization on prosocial behavior. Journal of Applied Psychology,
ior and Human Decision Processes, 50, 123–150.
71, 492– 499.
*MacKenzie, S. B., Podsakoff, P. M., & Rich, G. A. (1998). Transforma-
Organ, D. W. (1977). A reappraisal and reinterpretation of the satisfaction
tional and transactional leadership and salesperson performance. Un-
published manuscript, Indiana University Bloomington. causes performance hypothesis. Academy of Management Review, 2,
*McManus, M. A., & Kelly, M. L. (1999). Personality measures and 46 –53.
biodata: Evidence regarding their incremental predictive value in the life Organ, D. W. (1988). Organizational citizenship behavior: The good
insurance industry. Personnel Psychology, 52, 137–148. soldier syndrome. Lexington, MA: Lexington Books.
*McNamara, M. L. (1993). The effect of welfare corporatist structures on Organ, D. W. (1994). Organizational citizenship behavior and the good
organizational citizenship behavior. Unpublished doctoral dissertation, soldier. In M. G. Rumsey, C. B. Wallace, & J. H. Harris (Eds.),
Brigham Young University, Provo, UT. Personnel selection and classification (pp. 53– 67). Hillsdale, NJ:
*Moorman, R. H. (1991). Relationship between organizational justice and Erlbaum.
organizational citizenship behaviors: Do fairness perceptions influence Organ, D. W. (1997). Organizational citizenship behavior: It’s construct
employee citizenship? Journal of Applied Psychology, 76, 845– 855. clean-up time. Human Performance, 10, 85–97.
Moorman, R. H. (1993). The influence of cognitive and affective based job Organ, D. W., & Ryan, K. (1995). A meta-analytic review of attitudinal
satisfaction measures on the relationship between satisfaction and orga- and dispositional predictors of organizational citizenship behavior. Per-
nizational citizenship behavior. Human Relations, 46, 759 –776. sonnel Psychology, 48, 776 – 801.
*Moorman, R. H., & Blakely, G. L. (1995). Individualism– collectivism as *Podsakoff, P. M., Ahearne, M., & MacKenzie, S. B. (1997). Organiza-
an individual difference predictor of organizational citizenship behavior. tional citizenship behavior and the quantity and quality of work group
Journal of Organizational Behavior, 16, 127–142. performance. Journal of Applied Psychology, 82, 262–270.
*Moorman, R. H., Blakely, G. L., & Niehoff, B. P. (1998). Does perceived *Podsakoff, P. M., & MacKenzie, S. B. (1994). Organizational citizenship
organizational support mediate the relationship between procedural jus- behaviors and sales unit effectiveness. Journal of Marketing Re-
tice and organizational citizenship behavior? Academy of Management search, 31, 351–363.
Journal, 41, 351–357. Podsakoff, P. M., MacKenzie, S. B., & Bommer, W. H. (1996a). Meta-
*Moorman, R. H., Niehoff, B. P., & Organ, D. W. (1993). Treating analysis of the relationships between Kerr and Jermier’s substitutes for
employees fairly and organizational citizenship behavior: Sorting the leadership and employee job attitudes, role perceptions, and perfor-
effects of job satisfaction, organizational commitment, and procedural mance. Journal of Applied Psychology, 81, 380 –399.
justice. Employee Responsibilities and Rights Journal, 6, 209 –225. *Podsakoff, P. M., MacKenzie, S. B., & Bommer, W. H. (1996b). Trans-
*Morrison, E. W. (1994). Role definitions and organizational citizenship formational leadership behaviors and substitutes for leadership as deter-
DIMENSIONALITY OF OCB 65

minants of employee satisfaction, commitment, trust, and organizational *Tansky, J. W. (1993). Justice and organizational citizenship behavior:
citizenship behaviors. Journal of Management, 22, 259 –298. What is the relationship? Employee Responsibilities and Rights Jour-
*Podsakoff, P. M., MacKenzie, S. B., & Fetter, R. (1993). Substitutes for nal, 6, 195–207.
leadership and the management of professionals. Leadership Quar- *Tompson, H. B., & Werner, J. M. (1997). The impact of role conflict/
terly, 4, 1– 44. facilitation on core and discretionary behaviors: Testing a mediated
*Podsakoff, P. M., MacKenzie, S. B., Moorman, R. H., & Fetter, R. (1990). model. Journal of Management, 23, 583– 601.
Transformational leader behaviors and their effects on followers’ trust in *Vandewalle, D., Van Dyne, L., & Kostova, T. (1995). Psychological
leader, satisfaction, and organizational citizenship behaviors. Leadership ownership: An empirical examination of its consequences. Group and
Quarterly, 1, 107–142. Organization Management, 20, 210 –226.
Podsakoff, P. M., MacKenzie, S. B., Paine, J. B., & Bachrach, D. G. *Van Dyne, L., & Ang, S. (1998). Organizational citizenship behavior of
(2000). Organizational citizenship behaviors: A critical review of the contingent workers in Singapore. Academy of Management Journal, 41,
theoretical and empirical literature and suggestions for future research. 692–703.
Journal of Management, 26, 513–563. Van Dyne, L., Cummings, L. L., & McLean Parks, J. M. (1995). Extra-role
*Podsakoff, P. M., Niehoff, B. P., MacKenzie, S. B., & Williams, M. L. behaviors: In pursuit of construct and definitional clarity (A bridge over
(1993). Do substitutes for leadership really substitute for leadership? An muddied waters). In L. L. Cummings & B. M. Staw (Eds.), Research in
empirical examination of Kerr and Jermier’s situational leadership organizational behavior (Vol. 17, pp. 215–285). Greenwich, CT: JAI
model. Organizational Behavior and Human Decision Processes, 54, Press.
1– 44. Van Dyne, L., Graham, J. G., & Dienesch, R. M. (1994). Organizational
*Pond, S. B., Nacoste, R. W., Mohr, M. F., & Rodriguez, C. M. (1997). citizenship behavior: Construct redefinition, operationalization, and val-
The measurement of organizational citizenship behavior: Are we assum- idation. Academy of Management Journal, 37, 765– 802.
ing too much? Journal of Applied Social Psychology, 27, 1527–1544. Van Dyne, L., & LePine, J. A. (1998). Helping and voice extra-role
*Puffer, S. M. (1987). Prosocial behavior, noncompliant behavior, and
behavior: Evidence of construct and predictive validity. Academy of
work performance among commission salespeople. Journal of Applied
Management Journal, 41, 108 –119.
Psychology, 72, 615– 621.
Van Scotter, J. R., & Motowidlo, S. J. (1996). Evidence for two factors of
Raudenbush, S. W. (1994). Random effects models. In H. Cooper & L. V.
contextual performance: Job dedication and interpersonal facilitation.
Hedges (Eds.), The handbook of research synthesis (pp. 301–321). New
Journal of Applied Psychology, 81, 525–531.
York: Russell Sage Foundation.
*VanYperen, N. W., van den Berh, A. E., & Willering, M. C. (1999).
*Schappe, S. P. (1998). The influence of job satisfaction, organizational
Towards a better understanding of the link between participation in
commitment, and fairness perceptions on organizational citizenship be-
decision-making and organizational citizenship behaviour: A multilevel
havior. Journal of Psychology, 132, 277–290.
analysis. Journal of Occupational and Organizational Psychology, 72,
*Schaubroeck, J., & Fink, L. S. (1998). Facilitating and inhibiting effects
377–392.
of job control and social support on stress outcomes and role behavior:
A contingency model. Journal of Organizational Behavior, 19, 167– Wagner, J. A., III, & LePine, J. A. (1999). Effects of participation on
195. performance and satisfaction: Additional meta-analytic evidence. Psy-
*Schnake, M., Cochran, D., & Dumler, M. P. (1995). Encouraging orga- chological Reports, 84, 719 –725.
nizational citizenship: The effects of job satisfaction, perceived equity *Walz, S. M., & Niehoff, B. P. (1996). Organizational citizenship behav-
and leadership. Journal of Managerial Issues, 7, 209 –221. iors and their effect on organizational effectiveness in limited-menu
*Schnake, M., Dumler, M. P., & Cochran, D. S. (1993). The relationship restaurants. Academy of Management Best Paper Proceedings, 56th
between “traditional” leadership, “super” leadership, and organizational Annual Meeting, Cincinnati, OH, 307–311.
citizenship behavior. Group and Organization Management, 18, 352– Williams, L. J., & Anderson, S. E. (1991). Job satisfaction and organiza-
365. tional commitment as predictors of organizational citizenship and in-role
*Scholl, R. W., Cooper, E. A., & McKenna, J. F. (1987). Referent selection behaviors. Journal of Management, 17, 601– 617.
in determining equity perceptions: Differential effects on behavioral and *Witt, L. A., & Silver, N. C. (1994). The effects of social responsibility and
attitudinal outcomes. Personnel Psychology, 40, 113–124. satisfaction on extrarole behaviors. Basic and Applied Social Psychol-
Schwab, D. P. (1980). Construct validity in organizational behavior. In ogy, 15, 329 –338.
B. M. Staw & L. L. Cummings (Eds.), Research in organizational
behavior (Vol. 2, pp. 3– 43). Greenwich, CT: JAI Press.
*Smith, C. A., Organ, D. W., & Near, J. P. (1983). Organizational citi- Received June 22, 2000
zenship behavior: Its nature and antecedents. Journal of Applied Psy- Revision received March 15, 2001
chology, 68, 653– 663. Accepted March 15, 2001 !

View publication stats

You might also like