Sei sulla pagina 1di 19

Journal of Applied Psychology Copyright 2008 by the American Psychological Association

2008, Vol. 93, No. 6, 1348 –1366 0021-9010/08/$12.00 DOI: 10.1037/a0012743

The Development and Validation of the Workplace Ostracism Scale

D. Lance Ferris, Douglas J. Brown, Joseph W. Berry, and Huiwen Lian


University of Waterloo

This article outlines the development of a 10-item measure of workplace ostracism. Using 6 samples
(including multisource and multiwave data), the authors developed a reliable scale with a unidimensional
factor structure that replicated across 4 separate samples. The scale possessed both convergent and
discriminant validity, and criterion-related validity was demonstrated through the scale’s relation with
basic needs, well-being, job attitudes, job performance, and withdrawal. Overall, the present study
suggests that the Workplace Ostracism Scale is a reliable and valid measure and that the workplace
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ostracism construct has important implications for both individuals and organizations.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Keywords: ostracism, deviance, measurement

Deviant behaviors, defined as behaviors that are initiated by Survey results support the notion that ostracism is a universal
employees and that contravene organizational norms, have signif- experience, with most individuals either having ostracized others
icant organizational and individual costs (Bennett & Robinson, or having been ostracized themselves. For example, one survey
2003). A significant amount of research addressing deviant behav- indicated that 67% of respondents in relationships had used the
iors has accumulated in the past decade (Berry, Ones, & Sackett, silent treatment on a loved one, whereas 75% admitted that a loved
2007); typically, researchers have attempted to discern what causes one had used the silent treatment on them (Faulkner, Williams,
individuals to engage in deviant behavior. Recently, however, Sherman, & Williams, 1997). In the organizational domain, cases
there has been a call for researchers (a) to move beyond the study of workplace ostracism are increasingly ending up before the
of “plant floor” forms of deviance (e.g., theft and sabotage) and courts (Waddell, 1999). One survey of over 5,000 workers indi-
consider more universal and subtle forms of deviant behavior, cated that 13% of respondents had been excluded at work in the
which occur across organizations and hierarchical levels, and (b) to
previous 6 months (Hitlan, Kelly, Schepman, Schneider, & Zarate,
examine deviance from the target’s perspective, which has been
2006). Another survey revealed that, over a 5-year period, 66% of
surprisingly understudied (Bennett & Robinson, 2003).
employees had been given the silent treatment; of those surveyed,
Keeping in mind these recommendations, we examine the phe-
29% reported that others had left the room when they entered and
nomenon of being ostracized in the workplace in the present
article. Ostracism, or the extent to which an individual perceives 18% reported having been moved to an isolated location (Fox &
that he or she is ignored or excluded by others (K. D. Williams, Stallworth, 2005).
2001),1 is seemingly a universal phenomenon and occurs across Yet, despite an impressive array of anecdotal, theoretical, and
gender, age, and demographic lines (K. D. Williams, 1997). Os- empirical findings from various social science literatures (see
tracism has been seen among children, who ignore and exclude K. D. Williams, 1997, 2007, for a review), the concept of ostra-
bullies from their playgroups; among such widely diverse cultural cism has had little impact on organizational psychology. The
groups as Afghan tribes and the Amish; and even in other social omission is significant, as ostracism may potentially be related to
species, such as lions and chimpanzees (K. D. Williams, 1997). a variety of organizational outcomes, such as work attitudes,
behavior, and well-being. However, a major obstacle impeding our
understanding of workplace ostracism has been the absence of a
D. Lance Ferris, Douglas J. Brown, Joseph W. Berry, and Huiwen Lian, valid, reliable measure of the construct. To address this limitation,
Department of Psychology, University of Waterloo, Waterloo, Ontario, we introduce the construct of workplace ostracism and outline the
Canada. development of a measure of the construct in the current article. In
Phase 1 of this study is based on Joseph W. Berry’s honors thesis. This so doing, we provide initial evidence regarding the reliability,
research was supported in part by scholarships and grants from the Cana-
validity, and nomological network of the construct. The findings
dian Social Sciences and the Humanities Research Council to D. Lance
Ferris and Douglas J. Brown, respectively. We thank Stephen Risavy for
come from six independent samples and represent a total of 822
his assistance with data collection and David V. Day, Lisa Keeping, Norah employees from a variety of organizations and professions. Below,
Bree Ping, Chet Robie, and James A. Tan for their helpful comments on we review the construct of ostracism and distinguish it from
earlier versions of this article. D. Lance Ferris, Douglas J. Brown, and related organizational constructs.
Joseph W. Berry contributed equally to this article, and their authorship
was determined randomly.
1
Correspondence concerning this article should be addressed to D. In existing research on this topic, either ostracism or social exclusion
Lance Ferris, who is now at Lee Kong Chian School of Business, has been used to denote the experience of being rejected, excluded,
Singapore Management University, 50 Stamford Road, Singapore ignored, or isolated. Thus, in the present article, we use the terms ostracism
178899. E-mail: dlferris@smu.edu.sg and exclusion interchangeably.

1348
OSTRACISM 1349

Ostracism sense of belongingness by threatening to remove the individual


from the group. Research has found that even minimal forms of
Despite some early studies on the concept of rejection (e.g., ostracism, such as exclusion from a computerized ball-toss game,
Jackson & Saltzstein, 1958; Schachter, 1951; Snoek, 1962), the can reduce the sense of belonging (Williams et al., 2000) and
research on ostracism could best be characterized as piecemeal and lessen the individual’s identification with a group (Hogg &
infrequent until fairly recently (K. D. Williams, 2001). This over- Abrams, 1988; Tajfel & Turner, 1986). Individuals who are ostra-
sight is unfortunate, as ostracism represents a common human cized feel that they have done something wrong or that there is
experience familiar to many. Ostracism has been part of human life something inherently unappealing about themselves, which threat-
since the beginnings of recorded history, when the ancient Greeks ens their sense of self-esteem (K. D. Williams, 2001). When
would vote to ostracize individuals from their communities as a individuals are ostracized, their sense of control is undermined, in
form of punishment (K. D. Williams, 1997). Exile and banishment that their actions prompt no reaction from others, and they have no
represent extreme and complete forms of ostracism, but the phe- way of effecting an end to the ostracism (K. D. Williams, 1997;
nomenon is also represented by less dramatic behaviors (e.g., using Zadro et al., 2004). Finally, ostracism represents a form of “social
the silent treatment or avoiding eye contact); the more the indi- death” (Sommer, Williams, Ciarocco, & Baumeister, 2001) and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

vidual perceives that he or she is being subjected to such behav- provides a preview of what life would be like if one did not exist.
iors, the more the individual feels that he or she is being ostracized. Such cues of mortality and death threaten one’s sense of a mean-
Ostracism need not be intentional or punitive; in many instances, ingful existence (Pyszczynski et al., 2004).
we may ignore coworkers because we are engrossed in our own In addition, ostracism has been shown to cause maladaptive
work (K. D. Williams, 2001). Indeed, one of the aversive aspects responses. Experimental laboratory research on social exclusion
of ostracism is that it is often ambiguous whether one is actually has found that people who are excluded are more likely to be
being purposefully ostracized (K. D. Williams, 1997). aggressive, both toward those who excluded them and towards
Regardless of the motives underlying ostracism, it is a painful uninvolved others (Twenge, Baumeister, Tice, & Stucke, 2001).
and aversive experience. In studies with functional magnetic res- They are also less likely to engage in prosocial or helpful behav-
onance imaging, the same brain structures activated during phys- iors (Twenge, Baumeister, DeWall, Ciarocco, & Bartels, 2007)
ical pain were activated after the individual experienced social and more likely to engage in self-defeating behaviors, such as
rejection; this finding suggests that ostracism causes a sense of procrastinating and making poor choices (Twenge, Catanese, &
“social pain” (N. I. Eisenberger, Lieberman, & Williams, 2003; Baumeister, 2002). It has been suggested that the effect of ostra-
MacDonald & Leary, 2005). Given the evolutionary advantages cism on our ability to self-regulate, or our ability to alter our
that groups provide, K. D. Williams and Zadro (2005) have argued behavior to comply with social standards and achieve goals, is
that the experience of pain in response to ostracism represents an responsible for these maladaptive behaviors (Baumeister, DeWall,
adaptive response that alerts us if something is wrong and prompts Ciarocco, & Twenge, 2005). Studies have shown that, when ex-
us to remedy the situation. Supporting this notion, studies have cluded, individuals enter into a deconstructed cognitive state with
found that we appear to detect ostracism indiscriminately, regard- minimal self-awareness, a focus on the present, and no concern for
less of extenuating or discounting circumstances, such as being long-term goals (Twenge, Catanese, & Baumeister, 2003). More-
told that others have been instructed to ignore us (Zadro, Williams, over, exclusion has been shown to impair logical reasoning
& Richardson, 2004). Similarly, whether one is being ostracized (Baumeister, Twenge, & Nuss, 2002). Being self-aware and being
by in-group or out-group members does not seem to moderate the able to comprehend the long-term consequences associated with
effects of ostracism (K. D. Williams, Cheung, & Choi, 2000); even our own actions are important components of self-regulation
ostracism by a hated out-group, such as the Klu Klux Klan, has a (Carver & Scheier, 1981). Thus, it is perhaps unsurprising that
negative impact on individuals (Gonsalkorale & Williams, 2007). ostracized individuals display a variety of behaviors indicating
These studies have been taken to suggest that we appear to be their inability to self-regulate (Baumeister et al., 2005).
biased to detect ostracism, even false positives, perhaps because
the potential consequences associated with actual ostracism can be Ostracism and Organizations
severe (e.g., loss of the resources, protection, and mate access that
groups provide). Thus, ostracism by a single individual or a group, Given the role of organizations as a primary source of need
be it friend or foe, is likely to be detected by and to have a negative satisfaction for individuals (Armeli, Eisenberger, Fasolo, & Lynch,
impact on the individual. 1998; Cropanzano & Mitchell, 2005; Hackman & Oldham, 1980)
Aside from its capacity to cause social pain, ostracism is aver- and the importance of self-regulation to organizational outcomes
sive in that it simultaneously threatens four fundamental human such as performance and workplace deviance (Campion & Lord,
needs: the need for self-esteem, the need to belong, the need for 1982; Kozlowski & Bell, 2006; Porath & Bateman, 2006; Van-
control, and the need for a meaningful existence (K. D. Williams, couver, 2005), the impact of ostracism on needs and self-
1997, 2001, 2007). These four needs have a rich history in psychol- regulation may have profound consequences both for employees
ogy, with different authors arguing that these needs, either separately and the organizations that employ them. It is surprising, therefore,
or in combination, underlie all aspects of social behavior (see, e.g., that the ostracism construct has been virtually ignored by organi-
Baumeister, 1994; Baumeister & Leary, 1995; Baumeister & Tice, zational psychologists (for an exception in which researchers used
1990; Branden, 2001; Greenberg, Pyszczynski, & Solomon, 1986; a scenario study format to examine language-based exclusion, see
Pyszczynski, Greenberg, & Solomon, 1999; Pyszczynski, Green- Hitlan et al., 2006).
berg, Solomon, Arndt, & Schimel, 2004; Seligman, 1975; Taylor One possible reason for this lack of attention may be that no
& Brown, 1988; Tesser, 2000). Ostracism impacts an individual’s measure of workplace ostracism has been published to date. Indeed,
1350 FERRIS, BROWN, BERRY, AND LIAN

to the extent that it has been examined in organizations, ostracism ability estimates, factor structure, and susceptibility to method-
usually has been treated as part of other, broader constructs, such ological confounds) of the 10-item scale. Phase 3 details the
as interpersonal deviance (Bennett & Robinson, 2000), workplace convergent and discriminant validity of the 10-item scale, and
bullying (Fox & Stallworth, 2005), aggression (Neuman & Baron, Phase 4 examines the scale’s criterion-related validity.
1998), or social undermining (Duffy, Ganster, & Pagon, 2002).
There is considerable debate regarding the relative merits of ex- Phase 1: Item Generation and Reduction
amining discrete dimensions (e.g., ostracism) or broader constructs
(e.g., social undermining, aggressive or bullying behaviors; Blau, Item Generation
1998; Edwards, 2000; Hanisch & Hulin, 1991; Johns, 1998). The
fact that strong evidence supports the utility of discrete dimensions We generated scale items using a deductive item-generation
(Edwards, 2000) suggests the need for a separate measure of the approach, in which sufficient theoretical grounds from which to
discrete ostracism construct. generate items are assumed to exist (Hinkin, 1998). We consulted
Beyond the broad– discrete measurement debate, there are con- the existing literature on ostracism and related constructs, such as
ceptual and theoretical reasons for separating ostracism from the exclusion, and defined ostracism as the perception that one is being
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ignored or excluded (Leary, 2005; Masters, 1984; K. D. Williams,


This document is copyrighted by the American Psychological Association or one of its allied publishers.

broader constructs with which it is normally assessed. With respect


to interpersonal deviance measures, these scales assess the extent 1997, 2001). Using this definition as a starting point, we generated
to which one acts as a perpetrator of a wide variety of deviant items in two ways. First, we generated items that were consistent
behavior, such as swearing at, aggressing against, or ostracizing with this definition but specific to the workplace. Consistent with
others. In contrast, our ostracism scale takes the perspective of the recommendations (Hinkin, 1998), we generated items that were
target and examines the effect of being ostracized. succinct and easily comprehensible. Additionally, we generated
Although aggregate measures (e.g., bullying, aggression, and items that were behavioral in nature and that did not confound
social undermining) may take the perspective of the target, they affective responses or other consequences with ostracism behaviors.
also confound the experience of being ostracized with other forms We next supplemented the items that we created with items
of behavior that involve interacting with the targeted individual. taken from preexisting scales. Although not meant to assess ostra-
Such behaviors are antithetical to the experience of being ostra- cism per se, these scales contained items that were consistent with
cized and thus are conceptually different from ostracism. For the definition of being ignored or excluded. For example, the
example, social undermining scales contain items that assess being social undermining scale (Duffy et al., 2002) contains the item
belittled or criticized (Duffy et al., 2002), and aggression/bullying “Others gave you the ‘silent treatment’ at work,” which captures
scales contain items that assess committing verbal or physical ostracism quite nicely. Using these two approaches, we obtained
assault, delivering unfair performance appraisals, and threatening 40 items; these were subsequently screened for redundancy and
the target with job loss or demotion (Fox & Stallworth, 2005; representativeness of the ostracism construct by D. Lance Ferris,
Neuman & Baron, 1998). These types of interactions, although still Douglas J. Brown, and Joseph W. Berry, resulting in a set of 25
aversive, allow the target to feel acknowledged by the source, items. Of these items, 11 were taken from other scales such as Fox
through dialogue or even physical attack, and to know when the and Stallworth’s (2005) bullying scale and Bennett and Robinson’s
episode has ended. There is also less ambiguity about whether a (2000) deviance scale.
conflict is occurring when one is being abused. Indeed, a recurrent
theme in accounts by long-term targets of ostracism is that they Item Reduction: Item-Sort Task
would prefer verbal or even physical abuse to being ostracized
(K. D. Williams, 2001). One study comparing the effects of being We next sought to assess the substantive validity of the 25 items
ostracized and the effects of being argued with demonstrated that with an item-sort task (Anderson & Gerbing, 1991).2 Item-sort
being ostracized had a significantly larger impact on self-esteem tasks are recommended for use in the early stages of scale devel-
and feelings of belonging, control, and a meaningful existence opment (Anderson & Gerbing, 1991; Hinkin, 1998), as they pro-
(Zadro, Williams, & Richardson, 2005). vide a guide for removing items that are not conceptually consis-
Thus, although ostracism may co-occur with other behaviors, tent with the construct under investigation and predict which items
such as bullying, aggression, or social undermining, it remains a will perform best in a confirmatory factor analysis (Anderson &
theoretically distinct concept and is potentially even more impact- Gerbing, 1991).
ful than are the behaviors with which it may co-occur. Before such Participants and procedure. We recruited 20 employed par-
questions can be addressed, however, a reliable and construct-valid ticipants (6 male) from a variety of occupations (e.g., teacher, store
measure of ostracism in the workplace needs to be developed. Our clerk, laborer, computer programmer) to complete the item-sort
purpose in the present study was to develop such a measure. task. We recruited them by asking graduate students to pass on the
item-sort task to working friends and family members. The mean
age of participants was 28 years (SD ⫽ 5.6), and participants had
Present Study
To create a reliable and valid measure of workplace ostracism, 2
Substantive validity is the extent to which the items of a scale reflect
we followed steps outlined in the survey measure development the construct of interest (Anderson & Gerbing, 1991). It is similar to
literature (Hinkin, 1998). Below, we present the results of these content validity but differs in its focus, which is on items, whereas content
studies in four phases. Phase 1 details the initial generation of a validity is normally assessed at the scale level. However, substantive and
pool of potential scale items and the refinement of the items into content validity are intertwined, in that a scale cannot have content validity
a 10-item scale. Phase 2 details the psychometric properties (reli- if its items do not possess substantive validity.
OSTRACISM 1351

been working at their jobs for an average of 6.1 years (SD ⫽ 3.3). Participants and procedure. We recruited a sample of 99
Although a sample size of 20 may seem small, this is the recom- full-time employees (55 male, 39 female, 5 undeclared) from a
mended size for an item-sort task, in that it produces reliable diverse set of occupations (e.g., manager, electrician) using re-
agreement coefficients and correlations exceeding .90 across dif- cruitment posters that were placed in public places (e.g., coffee
ferent population samples (Anderson & Gerbing, 1991). shops, shopping/grocery stores). The posters contained a link for
When participants completed an item-sort task, we provided the online survey. Participants were employed in various indus-
them with a definition of ostracism, as well as definitions of other tries, including sales (18%), financial (12%), education (9%),
related constructs, and asked them to sort a list of scale items arts/entertainment/media (9%), manufacturing (9%), computer
according to which construct each scale item appeared to be (7%), government (5%), and health care (5%). Participants had a
assessing. Items that are assigned to their correct constructs dem- mean age of 31.7 years (SD ⫽ 9.4 years); they worked an average
onstrate higher levels of substantive validity than do items that are of 37.2 hr (SD ⫽ 9.2) per week and had been employed at their
assigned to incorrect constructs. In addition to the definition of organization for an average of 4.5 years (SD ⫽ 5.9). In exchange
ostracism, we provided definitions for the related constructs of for their participation, they were entered in a drawing for a $75 gift
interpersonal justice (e.g., “To what extent has your supervisor
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

certificate at a national electronics store. Participants read the 13


This document is copyrighted by the American Psychological Association or one of its allied publishers.

treated you with dignity”; Colquitt, 2001), workplace bullying ostracism items and used an open-ended scale to indicate how
(e.g., “Spread false rumors about your work performance”; Fox & often each of the items had happened to them in the past year. We
Stallworth, 2005), workplace aggression (e.g., “Attacked you with dichotomized responses according to whether or not participants
a weapon”; Neuman & Baron, 1998), and social undermining (e.g., indicated that they had experienced the behavior. We then dichot-
“How often have coworkers not given as much help as they omized these responses for each scale item to get an indication of
promised”; Duffy et al., 2002). These constructs were chosen what percentage of the sample had actually experienced the be-
because, in the case of the undermining, aggression, and bullying havior at work in the past year.
scales, each contained items that we had included in our pool of Results. Two items (“You have been included in conversations
ostracism items in the item-generation stage. Thus, we wanted to at work” and “You had to be the one to start a conversation in
ensure that these items better reflected ostracism, in comparison to order to be social at work”) stood out as having relatively high
the original scale constructs from which they were taken (for frequencies. It seemed logical to us that these items would occur to
item-sort purposes, we considered the items taken from other a majority of individuals in organizations and thus would limit the
scales to reflect ostracism, rather than the construct they were variance of our scale. Additionally, given that items with low
originally intended to measure). Interpersonal justice was included variance can negatively impact the predictive ability of a scale
as a construct, as it taps into a similar construct (interactions with (DeVellis, 1991), we examined the item with the lowest variance
others) and is commonly used by organizational psychologists. (“Others at work stopped talking to you”). As the goal of any scale
We used two indices suggested by Anderson and Gerbing should be to balance domain sampling with scale parsimony
(1991) to assess the substantive validity of the scale items: the (Cronbach & Meehl, 1955) and given the high/low variances of the
proportion of substantive agreement (PSA), which assesses the items in question, we considered whether these three items could
proportion of respondents who assign an item to its intended be removed from our scale without sacrifice of content coverage.
construct, and the coefficient of substantive validity (CSV), which As an examination of our scale reveals, each of the three items
represents the extent to which respondents assign an item to its dealt with individuals conversing with the target individual, and
intended construct more than to any other construct. The latter other items in the scale (e.g., “Others at work shut you out of the
coefficient can be tested for statistical significance to determine conversation”) assessed this subject. Thus, we came to the decision
whether the number of respondents who assign the item to its that these three items could be removed from the scale without
intended construct can be accounted for by random chance alone.3 sacrificing the content coverage of the scale and that doing so
Results. The PSA and CSV coefficients should be used in a would be beneficial in terms of increasing the variance of the scale.
comparative manner to retain a subset of items with the largest The preceding actions resulted in a 10-item scale, which we
values (Anderson & Gerbing, 1991). Thus, to balance substantive termed the Workplace Ostracism Scale (WOS). These 10 items
validity and scale economy, we retained items with PSA ⱖ.95 and (ranging in occurrence from 20% to 54%) were representative of
CSV ⱖ.90. The resulting scale contained 13 items (see Appendix). the ostracism construct, in that they tapped into the notion of being
All of the items were significantly assigned to the ostracism
construct beyond chance levels.
3
The formula for PSA is PSA ⫽ nc/N, where nc represents the number
of participants who assign an item to its posited construct and N is the total
Item Reduction: Frequency Analysis
number of participants. CSV is calculated using the formula CSV ⫽
Having established the substantive validity of our 13-item scale, (nc ⫺ no)/N, where nc and N are as previously defined and no refers to the
we sought to ensure that the items listed actually occurred in the highest number of assignments of the item to any other construct in the set
(Anderson & Gerbing, 1991). To test whether an item’s CSV is significant,
workplace. It was entirely possible that some of our items de-
one must determine the minimum value that the CSV could have if random
scribed situations that, although consistent with the definition of
chance alone still could not account for the number of assignments that an
ostracism, either never occur in the workplace or occur all the item had received to its posited construct. Any CSV higher than this value
time. Both of these issues would reduce the overall variability of would indicate that chance could not account for this number of assign-
our scale and limit its ability to predict outcomes (Hinkin, 1998). ments. A chi-square goodness-of-fit test can be used to determine the
Thus, we conducted a study in which we asked individuals whether minimum value. A more thorough description of this procedure can be
they had experienced the 13 ostracism items in the previous year. found in Anderson and Gerbing (1991).
1352 FERRIS, BROWN, BERRY, AND LIAN

ignored or excluded at work and also demonstrated acceptable places. An initial examination for outliers revealed two multivar-
substantive validity and variability. Having established our scale, iate outliers, as indexed by Mahalanobis distance (following cur-
we next assessed its psychometric properties. rent recommendations, we employed a conservative standard of
p ⬍ .001; Tabachnick & Fidell, 2007). Given that multivariate
outliers can have undue influence, creating both Type I and Type
Phase 2: Psychometric Properties of the WOS
II errors, and that the number of outliers was extremely small, we
The psychometric properties of the WOS were evaluated in deleted these cases. We also removed 5 participants with incom-
terms of its reliability, factor structure, and the extent to which it plete data, leaving a final sample of 154 individuals (66% female).
is unduly contaminated by method effects, or variables that poten- Participants were employed in various industries, including fi-
tially bias the measurement of a variable due to the method used nance (16%), education (15%), health care (11%), government
(i.e., self-report). In particular, impression management and posi- (10%), manufacturing (10%), computer (8%), and sales (6%). The
tive/negative affectivity have been implicated as potentially im- mean age of participants was 37.4 years (SD ⫽ 10.8); they worked
portant method effects that influence self-report measures (Tan & an average of 41.4 hr (SD ⫽ 6.1) per week and had been employed
at their job for an average of 7.0 years (SD ⫽ 8.7). In exchange for
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Hall, 2005; L. J. Williams & Anderson, 1994; L. J. Williams,


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Gavin, & Williams, 1996). Impression management, defined as “a their participation, they were given a 1-in-20 chance at winning a
tendency for an individual to present him or herself . . . in a way $50 cash prize. Participants completed a short online survey con-
that makes the person look positive with regard to culturally taining the WOS. The survey contained other measures for
derived norms and standards” (Ganster, Hennessey, & Luthans, criterion-related validity purposes (see Table 1 for a complete
1983, p. 322), has been argued to potentially create spurious listing of scales included).
correlations or to suppress existing correlations between variables. Sample 3. We recruited an independent sample of 140 partic-
It could be argued that people would be reluctant to admit to being ipants, employed in diverse occupations (e.g., consultant, office
ostracized out of fear that their admission would make them appear clerk, graphic designer, systems analyst, operations manager),
badly. using posters placed in commuter areas and other public places.
Similarly, positive affectivity (PA) and negative affectivity Two multivariate outliers as indexed by Mahalanobis distance
(NA), defined as individual differences in the extent to which were detected and removed. We also removed 9 participants who
individuals experience chronic positive or negative emotionality, had provided incomplete data, leaving a final sample of 129
have been argued to influence self-reports of workplace situations individuals (60% female). Participants were employed in various
(Burke, Brief, & George, 1993; Watson, Pennebaker, & Folger, industries, including financial (15%), computer (12%), manufac-
1987). To ensure that the WOS was not being adversely influenced turing (8%), engineering (7%), sales (7%), health care (6%),
by impression management, PA, or NA, we examined the extent to office/administrative support (5%), government (5%), and arts/
which the WOS was confounded with these three method effects. entertainment/media (5%). The mean age of participants was 31.0
years (SD ⫽ 8.4); they worked an average of 41.6 hr (SD ⫽ 7.8)
per week and had been employed at their current job for an average
Method of 5.1 years (SD ⫽ 5.0). Participants completed a short online
Participants and Procedure survey containing the WOS and were compensated $10 for their
time.
Reliability analyses and confirmatory factor analyses were con- Sample 4. We recruited an independent sample of 131 partic-
ducted in four separate samples, and method effects were exam- ipants, representing diverse occupations (e.g., accountant, nurse,
ined in Sample 1. bank teller, sales manager), using posters placed on online forums.
Sample 1. We recruited 295 participants (68% male) from We removed 6 participants who had provided incomplete data,
various occupations (e.g., waiter, sales associate, retail clerk, IT leaving a final sample of 125 individuals (55% male). Participants
consultant, general laborer, assistant manager) using posters, were employed in various industries, including sales (13%), finan-
placed in commuter areas (e.g., bus shelters) and public places cial (12%), health care (10%), manufacturing (9%), computers
(e.g., coffee shops), that invited them to complete an online survey. (8%), government (7%), construction (7%), and education (6%).
They were employed in various industries, including sales (15%), The mean age of participants was 34.1 years (SD ⫽ 9.9); they
computer (12%), financial (8%), education (8%), government worked an average of 40.8 hr (SD ⫽ 6.6) per week and had been
(6%), manufacturing (5%), health care (5%), and food services employed at their current job for an average of 4.8 years (SD ⫽
(5%). Participants had a mean age of 30.1 years (SD ⫽ 9.4); they 6.3). Participants were compensated $10 for their time and were
worked an average of 36.9 hr (SD ⫽ 11.4) per week and had been entered in a drawing to win one of two $200 cash prizes. Partic-
employed at their organization for an average of 4.5 years (SD ⫽ ipants completed a short online survey that contained the WOS.
10.1). In exchange for their participation, they were entered in a The survey contained other measures for criterion-related validity
drawing with a 1-in-20 chance of winning a $75 gift certificate for purposes (see Table 1 for a complete listing of the scales included).
a national electronics store. Participants completed the WOS, an
impression management scale, and the Positive and Negative Af- Measures
fect Schedule (PANAS;Watson, Clark, & Tellegen, 1988) online.
Sample 2. We recruited an independent sample of 161 partic- Unless otherwise noted, all ratings in this article were made with
ipants, employed in various occupations (e.g., sales, data analyst, a 7-point scale (1 ⫽ strongly disagree and 7 ⫽ strongly agree).
secretary, attorney, store manager, engineer), using newspaper ads WOS. We measured workplace ostracism with the 10-item
and recruitment posters placed in commuter areas and other public scale developed in Phase 1 (see Appendix).
OSTRACISM 1353

Table 1
Overview of Measures Used in Each Sample

Measure (source) Sample 1 Sample 2 Sample 3 Sample 4

WOS X X X X
Undermining (Duffy et al., 2002) X
POS (R. Eisenberger et al., 1986) X
PANAS (Watson et al., 1988) X
Impression management (Paulhus, 1991) X
Environmental policies Xa
Needs (K. D. Williams et al., 2000) Xa
Depression (Caplan et al., 1980) Xa
Anxiety (House & Rizzo, 1972) Xb
Job satisfaction (Brayfield & Rothe, 1951) Xb
Job satisfaction (Rosen et al., 2006) Xa
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Affective commitment (Meyer & Allen, 1997) X


This document is copyrighted by the American Psychological Association or one of its allied publishers.

In-role behavior (L. J. Williams & Anderson, 1991) Xc


Citizenship behavior (L. J. Williams & Anderson, 1991) Xc
Deviant behavior (Aquino et al., 1999) Xc
Deviant behavior (Bennett & Robinson, 2000) Xa X
Turnover intentions (Wayne et al., 1997) Xa
Job search behaviors (Blau, 1994) Xa
LMX (Graen et al., 1982) X
Group cohesion (Seashore, 1954) X
Interpersonal justice (Colquitt, 2001) X
Perceived coworker OCB norms X

Note. WOS ⫽ Workplace Ostracism Scale; POS ⫽ perceived organizational support; PANAS ⫽ Positive and
Negative Affect Schedule; LMX ⫽ leader–member exchange.
a
Scale was assessed in a 5-month follow-up of a subset of Sample 1 participants. b Sample 2 scale was
assessed with significant-other reports. c Sample 3 scale was assessed with work-peer reports.

Impression management. We used Paulhus’s (1991) 20-item termed the baseline model with that of a confounded measurement
Impression Management Scale to assess impression management model. For the baseline model, a given latent method effect vari-
(␣ ⫽ .79; Meston, Heiman, Trapnell, & Paulhus, 1998). This scale able (e.g., impression management) is modeled as having no
has been used extensively in organizational research (see Li & relation to the latent variable representing the WOS. The con-
Bagger, 2006); it assesses construct validity and reliability founded measurement model differs from the baseline model in
(Paulhus, 1991). Participants responded to questions such as that it adds paths from the latent method effect variable to the
“I sometimes tell lies if I have to.” indicators of the WOS latent variable (i.e., the method effect is
PANAS. We used Watson, Clark, and Tellegen’s (1988) 20- allowed to “confound” measurement of the WOS). A significant
item PANAS scale to assess trait PA (␣ ⫽ .82) and NA (␣ ⫽ .89). improvement in chi-square from the baseline model to the con-
These scales have been extensively validated (see Watson et al., founded measurement model suggests the presence of method
1988) and are the measures most commonly used for assessment of effects, in that the data are better represented by allowing the
PA and NA in organizational research (Thoresen, Kaplan, Barsky, method effect to influence the WOS at the measurement (i.e.,
Warren, & de Chermont, 2003). Participants used 5-point scales indicator) level. Should the confounded measurement model pro-
(1 ⫽ very slightly or not at all to 5 ⫽ extremely) to indicate how vide a significantly better fit than does the baseline model, one can
they generally felt (e.g., enthusiastic, proud). estimate the variance the latent method effect accounts for in the
indicators of the WOS by squaring the path estimates from the
Analytic Strategy latent method effect variable to the indicators of the WOS. (For a
more complete description of the procedures for detecting method
A confirmatory factor analysis was performed for each sample effects, the interested reader is referred to L. J. Williams and
with AMOS 7.0. Given that the WOS was created a priori to assess Anderson, 1994, or L. J. Williams et al., 1996.) We estimated
a single underlying factor, we tested a hypothesized unitary factor baseline models (and their associated confounded measurement
structure, with all WOS scale items loading on a single latent models) separately for each method effect (i.e., impression man-
factor. We used Hu and Bentler’s (1999) two-index presentation agement, PA, and NA).
strategy, which suggests assessing model fit with the standardized
root-mean-square residual (SRMR) and the comparative fit index
(CFI). According to Hu and Bentler, values approaching .95 for the Results
CFI and values no higher than .08 for the SRMR indicate good Reliability and Confirmatory Factor Analyses
model fit.
We followed established procedures for testing for method The coefficient alpha reliability estimates were .89, .93, .96, and
effects with latent variable modeling (L. J. Williams & Anderson, .94 for Samples 1 through 4, respectively. The average corrected
1994). Briefly, these tests involve comparing the fit of what is item–total correlations ranged from .65 to .86, with a mean of .75
1354 FERRIS, BROWN, BERRY, AND LIAN

across the four samples. Moreover, all interitem correlations were however, there are a number of constructs that assess negative
positively intercorrelated. Taken as a whole, the alphas, corrected interactions in the workplace, to which one would expect ostracism
item–total correlations, and interitem correlations indicated that to be related. One such construct is social undermining (Duffy et
the WOS possesses adequate internal reliability (Cortina, 1993). al., 2002). Social undermining is defined as “behavior intended to
The confirmatory factor analyses in all four samples supported hinder, over time, the ability to establish and maintain positive
the factor structure of the WOS. Meeting conventional cutoff interpersonal relationships, work-related success, and favorable
criteria (Hu & Bentler, 1999), CFI values ranged between .94 and reputation” (Duffy et al., 2002, p. 332). Ostracism can be classified
.98 (average CFI ⫽ .96) and SRMR values ranged between .03 and as a form of social undermining, as it inhibits positive interper-
.04 (average SRMR ⫽ .04) across the four samples. The average sonal relationships and is associated with an unfavorable reputa-
chi-square across the samples was 90.7 (with 35 degrees of free- tion. Indeed, the social undermining scale includes an item that
dom). All scale items significantly ( p ⬍ .001) loaded on the latent could be classified as assessing ostracism. However, ostracism is
ostracism variable. Factor loadings ranged from .50 to .93 across separate from other undermining behaviors, as it represents a
the samples (average factor loading across the samples was .78); unique form of undermining in which members of the group do not
all of these loadings surpassed the conventional cutoff value of .40
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

communicate with the individual. Moreover, although undermin-


This document is copyrighted by the American Psychological Association or one of its allied publishers.

(Hinkin, 1998). Thus, the results of the factor analyses indicated ing must be perceived as intended to hinder an individual’s ad-
that, across four different samples, a single underlying factor vancement in the organization, one can use ostracism to punish an
provided a good fit to the data. individual without necessarily intending to hinder his or her ad-
vancement. Thus, we expected that ostracism would relate to
Method Effects undermining, in that it might co-occur with other undermining
behaviors, but that ostracism and undermining would not correlate
The correlations between the WOS and impression manage-
excessively highly.
ment, PA, and NA are presented in Table 2. The WOS was uncor-
Another construct to which ostracism should relate is interper-
related with impression management and correlated significantly with
sonal justice, which represents the extent to which the individual is
positive and negative affectivity.4 Paralleling the correlational results,
treated with respect and propriety (Colquitt, 2001). When one is
there was no significant chi-square change between the baseline
ostracized, one presumably believes that such treatment is unwar-
model and the impression-management-confounded measurement
model, ⌬␹2(10, N ⫽ 264 ) ⫽ 13.6, p ⬎ .05, which indicated that ranted and disrespectful. (Indeed, ostracism is posited to influence
impression management did not influence the measurement of the the individual’s sense of a meaningful existence; it is unlikely that
WOS. However, there were significant chi-square changes between treatment that causes one to question one’s existence could be
the baseline model and the PA-confounded-measurement model, deemed respectful.) Thus, we would expect that feelings of inter-
⌬␹2(10, N ⫽ 274) ⫽ 25.8, p ⬍ .05, and the NA-confounded- personal injustice should co-occur with ostracism. However, the
measurement model, ⌬␹2(10, N ⫽ 280) ⫽ 41.6, 1, p ⬍ .05; these two constructs remain separate, in that the WOS measures a subset
results suggest that PA and NA do influence the measurement of of the behaviors that can lead to the perception that one is being
the WOS. We next examined the magnitude of the influence of treated unjustly. That is, ostracism is best conceptualized as an
PA/NA method bias by squaring the paths from the method factor antecedent of perceptions of interpersonal justice. Moreover, os-
to the WOS indicators. The results indicated that PA accounted for tracism is not the same as interpersonal justice, as ostracism is not
less than 4% and NA accounted for approximately 5% of the the sole manner in which individuals can be treated with disrespect.
systematic variance; both effects are small. These results are For example, they may be subjected to improper remarks or com-
consistent with those of other investigations of method effects ments (Colquitt, 2001); even if such remarks resulted in perceptions
(e.g., Keeping & Levy, 2000; Munz, Huelsman, Konold, &
McKinney, 1996; L. J. Williams et al., 1996), which have con- 4
As seen in Table 1, the size of the correlations between the WOS and
cluded that, although method effects may exist, the size of the
PA/NA were similar in magnitude to the size of the correlations between
effects is essentially negligible.5 Taken together with the lack of
PA/NA and other established scales that were measured at the same time
impression management method effects, these results support the (e.g., social undermining and perceived organizational support; these con-
psychometric properties of the WOS in that the scale is not unduly structs are discussed in Phase 3: Convergent and Discriminant Validity).
confounded with method effects. 5
Seeking further evidence that the effects of PA and NA are negligible,
we compared the fit of the confounded measurement model with that of a
Phase 3: Convergent and Discriminant Validity third model, which, like the confounded measurement model, had paths
freed from the method effect to the WOS scale indicators. However, in the
Part of developing the nomological network around the ostra- third model we also fixed the factor loadings from the latent WOS factor
cism construct involves examination of the relation of ostracism to its indicators to be equivalent to the values estimated from the baseline
with similar and dissimilar constructs, or convergent and discrimi- model. This approach determined whether the method effect biased the
nant validity. Convergent validity represents the extent to which a measurement of the WOS items from the baseline model values (adapted
from Williams et al., 1996). For comparisons involving both the PA and
scale relates to other measures of the construct or similar con-
NA method effects, the change in fit of the models was not significant. This
structs; discriminant validity represents the extent to which a scale result indicates that, although PA and NA may account for some variance
demonstrates low or null correlations with dissimilar measures in the WOS scale items, this variance does not significantly bias the WOS
(Campbell & Fiske, 1959; Hinkin, 1998). The fact that no other factor loadings from the baseline model (which assumes no method ef-
measure of ostracism (workplace or otherwise) currently exists fects). More details on these analyses can be obtained from D. Lance
renders that aspect of convergent validity potentially problematic; Ferris.
OSTRACISM 1355

Table 2
Descriptive Statistics, Zero Order Correlations, and Alphas for Sample 1

Variable M SD 1 2 3 4 5 6 7 8

1. Age (years) 30.14 9.39


2. Gender 0.32 0.45 0.21ⴱⴱ
3. WOS 1.61 0.62 0.06 0.07 0.89
4. Social undermining 1.55 0.65 ⫺0.09 0.02 0.56ⴱⴱ 0.93
5. Impression management 4.11 0.83 0.17ⴱⴱ ⫺0.08 ⫺0.07 ⫺0.12ⴱ 0.79
6. NA 2.12 0.73 ⫺0.19ⴱⴱ 0.07 0.27ⴱⴱ 0.27ⴱⴱ ⫺0.24ⴱⴱ 0.89
7. PA 3.26 0.57 0.01 0.08 ⫺0.23ⴱⴱ ⫺0.27ⴱⴱ 0.17ⴱⴱ ⫺0.17ⴱⴱ 0.82
8. POS 4.40 1.25 ⫺0.05 0.02 ⫺0.39ⴱⴱ ⫺0.34ⴱⴱ 0.11 ⫺0.25ⴱⴱ 0.26ⴱⴱ 0.93

Note. n ranges from 283 to 295; alphas are on the diagonal in bold. For gender, 0 ⫽ male, 1 ⫽ female. WOS ⫽ Workplace Ostracism Scale; NA ⫽
negative affectivity; PA ⫽ positive affectivity; POS ⫽ perceived organizational support.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.


p ⬍ .05. ⴱⴱ p ⬍ .01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

of interpersonal injustice, the experience would be antithetical to that (Tajfel & Turner, 1986). Similarly, ostracized individuals are less
of ostracism. likely to perceive their group members as being friendly, overall.
When creating a new measure that assesses negative interper- However, ostracism is not synonymous with group cohesion, in
sonal behaviors, one needs to establish that the new scale does not that members of groups that are not cohesive need not be ostra-
simply reflect a lack of other, positive interpersonal behaviors or cizing one another; instead, they may dislike or feel apathetic
relationships (Duffy et al., 2002). In this sense, it is important to toward each other or they may display low commitment and pride
differentiate ostracism from perceived organizational support in their group.
(POS), which represents employees’ beliefs regarding the extent to Given that ostracism can occur across the life span (K. D.
which the organization cares about the employees’ well-being (R. Williams, 1997), we expected that a valid measure of ostracism
Eisenberger, Huntington, Hutchison, & Sowa, 1986; Rhoades & would be unrelated to an individual’s age.6 Finally, we also ex-
Eisenberger, 2002), and leader–member exchange (LMX), or the pected that ostracism would be unrelated to one’s perception of the
quality of the exchange relationship between a leader and his or government’s environmental policies. Given that perceptions of
her follower (Gerstner & Day, 1997; Graen & Uhl-Bien; 1995). government environmental policies represent a variable that is
Ostracism is undoubtedly associated with a lack of support in the conceptually and theoretically unrelated to ostracism, if a null
workplace or poor relations, but it is not synonymous with these relation between the two constructs were supported, the case for
constructs. For example, it is possible that one could receive low the discriminant validity of the WOS would be further supported.
levels of support from coworkers but at the same time not be
ostracized (e.g., if the coworkers are too busy to help out). Simi-
Method
larly, poor relationships can result from a host of factors (e.g., lack
of trust) other than ostracism. That is, you can have a bad rela- Participants and Procedure
tionship with your leader without the leader ostracizing you. Thus,
although we expected that ostracism would relate negatively to Participants were the individuals from Samples 1 through 4 and
perceptions of support and LMX, we expected that they would are discussed in Phase 2. We calculated the correlations between
remain differentiable constructs. age and ostracism for all four samples. For Sample 1, participants
Finally, ostracism should also relate to perceptions of group- completed measures of POS and social undermining in the same
level phenomenon, such as norms for how individuals treat each survey as the WOS. Additionally, a subset of individuals from
other and perceptions of the cohesiveness of their work group. Sample 1 (described more fully in Phase 4) completed a measure
Group organizational citizenship behavior (OCB) norms reflect assessing their views on their government’s environmental policies
guidelines for the performance of behaviors that enhance the social 5 months after they had completed the WOS (see Table 1 for a
environment in which employees work (e.g., “Members of my listing of other scales completed). Finally, approximately 2 days
group advocate the importance of helping fellow coworkers”; after they had completed the WOS, participants in Sample 4 were
Ehrhart & Naumann, 2004, p. 962). Groups that help out fellow sent a link to an online survey that contained measures of inter-
employees should be less likely to engage in ostracizing behavior; personal justice, LMX, group cohesion, and OCB norms. Of the
however, although a group’s norms may not proscribe individuals
helping one another out, this does not imply that individuals will
6
necessarily ostracize each other instead. Rather, there simply may One may also expect that ostracism would be unrelated to gender.
not be a widespread belief or practice (i.e., a norm) of assisting However, individuals who are demographically different from the majority
in their workplace may be excluded, or may perceive they are being
individuals.
excluded, by demographically different others. As we did not assess this
Group cohesion is characterized by interpersonal attraction, possibility directly, we chose not to use gender as a variable for discrimi-
commitment to group tasks, and pride in the group (Mullen & nant validity. In our own data, ostracism was not significantly related to
Copper, 1994). Those who are ostracized are likely to feel low gender in Sample 1 (r ⫽ ⫺. 06, p ⬎. 05), Sample 3 (r ⫽ ⫺. 05, p ⬎. 05),
levels of group cohesion, as ostracism can influence sense of or Sample 4 (r ⫽ ⫺. 04, p ⬎. 05) but was significantly related to gender
belonging and, thus, identification with and pride in a group in Sample 2 (r ⫽ ⫺. 18, p ⬍.05).
1356 FERRIS, BROWN, BERRY, AND LIAN

original 131 participants, 111 completed the follow-up survey. significantly related to ostracism was subjected to confirmatory
Those who completed the follow-up survey did not differ from the factor analyses. Here, we assessed whether a single-factor model
original 131 participants in terms of ostracism levels, t(131) ⫽ (for which the covariance between the latent ostracism and other
⫺.51, p ⬎ .05. construct factor was set to 1.0) provided a better fit to the data than
did a two-factor model for which the latent ostracism and other
Measures construct factor were allowed to covary freely. If the chi-square
were significantly worse for the single-factor model than for the
WOS. We measured workplace ostracism with the scale de- two-factor model, this would suggest that the proper way to model
scribed in Phase 1. the scale items would be as loading on two separate latent factors;
POS. We used the 9-item version of N. I. Eisenberger et al.’s in turn, this result would suggest that ostracism is distinguishable
(1986) Perceived Organizational Support Scale (␣ ⫽ .93). Partic- from the other construct (Anderson & Gerbing, 1988). Finally,
ipants responded to items such as “Help is available from the drawing on the results of the confirmatory factor analyses, we used
organization when I have a problem.” Fornell and Larcker’s (1981) test to see if two constructs are
Social undermining. We used Duffy et al.’s (2002) 12-item separate. This test states that two constructs are considered sepa-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Co-worker Undermining Scale (␣ ⫽ .93).7 Participants were asked


This document is copyrighted by the American Psychological Association or one of its allied publishers.

rate when the average squared factor loading of each indicator on


how often they had experienced a variety of behaviors from their its respective latent construct is higher than is the shared variance
coworkers (e.g., “Spread rumors about you”) on a 7-point scale (i.e., squared correlation) between the two latent constructs.
(1 ⫽ never to 7 ⫽ everyday).
Interpersonal justice. We used Colquitt’s (2001) four-item
measure of interpersonal justice. Participants were asked whether Results and Discussion
they were treated with dignity, respect, and politeness and if
coworkers refrained from improper remarks or comments (see Tables 2, 3, 4, and 5 present the correlations between ostracism
Colquitt, 2001). They replied on a 5-point scale (1 ⫽ To a small and age for Samples 1, 2, 3, and 4, respectively. Table 2 also
extent to 5 ⫽ To a large extent; ␣ ⫽ .94). contains the correlations between ostracism, undermining, and
Leader–member exchange. We used the seven-item LMX-7 POS; Table 6 contains the correlation between ostracism and
scale (Graen, Novak, & Sommerkamp, 1982). Participants used a perceptions of government environmental policies; and Table 5
4-point Likert scale (1 ⫽ not at all to 4 ⫽ completely) to respond contains the correlations between ostracism and LMX, interper-
to questions such as “How well do you feel that your immediate sonal justice, group cohesion, and perceived coworker OCB norms.
supervisor understands your problems and needs?” (␣ ⫽ .91). As can be seen, ostracism was significantly related to undermining
Group cohesion. We used Seashore’s (1954) four-item mea- (r ⫽ .56, p ⬍ .01), POS (r ⫽ ⫺.39, p ⬍ .01), interpersonal justice
sure of group cohesion. Participants used a 4-point Likert scale (r ⫽ ⫺.62, p ⬍ .01), LMX (r ⫽ ⫺.43, p ⬍ .01), group cohesion (r ⫽
(1 ⫽ Not very good to 4 ⫽ Great, couldn’t be better) to evaluate ⫺.53, p ⬍ .01), and perceived OCB norms (r ⫽ ⫺.63, p ⬍ .01) but
their work group in terms of, for example, “the way employees get was unrelated to perceptions of government policies (r ⫽ .10, p ⬎
along with each other” and “the way employees stick together” .10) or age (rs ⫽ ⫺.06, ⫺.11, ⫺.01, and ⫺.08 for Samples 1 through
(␣ ⫽ .91). 4, respectively, all ps ⬎ .10).
Perceived coworker OCB norms. We adapted L. J. Williams We next conducted confirmatory factor analyses to ensure the
and Anderson’s (1991) OCB measure to assess participants’ per- WOS was separable from the six constructs with which it did
ceptions of descriptive work group norms. Participants were asked correlate significantly (undermining, POS, interpersonal justice,
to indicate, regardless of how they themselves were treated, LMX, group cohesion, and perceived OCB norms). For all six
whether their coworkers engaged in five OCBs (e.g., “Members of constructs, a two-factor model provided a significantly better fit
my team help fellow co-workers when needed,” “Members of my than did a model for which the variance between the two latent
team take the time to listen to each other’s problems and worries”; factors was set to 1.0: undermining, ⌬␹2(1, N ⫽ 251) ⫽ 132.6;
␣ ⫽ .86). POS, ⌬␹2(1, N ⫽ 251) ⫽ 221.0; interpersonal justice, ⌬␹2(1, N ⫽
Perception of government environmental policies. A four-item 101) ⫽139.6; LMX, ⌬␹2(1, N ⫽ 102) ⫽ 182.2; group cohesion,
semantic differential scale assessed perceptions of the govern- ⌬␹2(1, N ⫽ 102) ⫽ 259.3; perceived OCB norms, (1, N ⫽ 104) ⫽
ment’s environmental policies along four dimensions (good– bad, 131.4; all ps ⬍ .05. This result suggests that the WOS is a separate
wise–foolish, beneficial– harmful, useful– useless; ␣ ⫽ .92). construct from undermining, POS, interpersonal justice, LMX,
Scores on each dimension ranged from 1 to 5; higher scores group cohesion, and perceived OCB norms. Similarly, the results
indicated more negative perceptions toward government environ- of the Fornell and Larcker (1981) test suggested that the average
mental policies. squared factor loadings of the scale items on their respective
constructs were higher than was the shared variance between the
constructs (0.52 vs. 0.35 for the WOS and undermining, 0.57 vs.
Analytic Strategy 0.16 for POS, 0.71 vs. 0.42 for interpersonal justice, 0.61 vs. 0.21
for LMX, 0.68 vs. 0.34 for group cohesion, and 0.60 vs. 0.48 for
Although it is important to demonstrate convergent validity via
high relations between two constructs, it is also important to
demonstrate that constructs do not converge to the point of being 7
For all analyses involving undermining in the present study, the item
redundant. Thus, we assessed convergent and discriminant validity “Gave you the silent treatment” from the original undermining scale was
in three ways. First, the significance of the zero-order correlations removed, as this item was deemed to reflect ostracism (see Phase 1, Item
between the various scales was examined. Any construct that was Generation).
OSTRACISM 1357

Table 3
Descriptive Statistics, Zero Order Correlations, and Alphas for Sample 2

Variable M SD 1 2 3 4 5 6 7 8 9 10

1. Age (years) 37.44 10.83 —


2. Gender 0.66 0.48 0.03 —
3. WOS 1.51 0.77 ⫺0.11 ⫺0.18ⴱ 0.93
4. Affective commitment 4.46 1.39 0.25ⴱⴱ 0.10 ⫺0.24ⴱⴱ 0.86
5. Job satisfaction 3.74 0.85 0.25ⴱⴱ 0.16ⴱ ⫺0.34ⴱⴱ 0.61ⴱⴱ 0.93
6. Job satisfaction (significant other) 3.53 0.97 0.25ⴱⴱ 0.16 ⫺0.33ⴱⴱ 0.46ⴱⴱ 0.69ⴱⴱ 0.95
7. Anxiety 3.18 0.82 ⫺0.06 ⫺0.11 0.26ⴱⴱ ⫺0.12 ⫺0.31ⴱⴱ ⫺0.27ⴱⴱ 0.84
8. Anxiety (significant other) 2.84 0.80 0.06 ⫺0.07 0.24ⴱⴱ ⫺0.13 ⫺0.27ⴱⴱ ⫺0.34ⴱⴱ 0.34ⴱⴱ 0.85
9. Organizational deviant behaviors 1.99 0.94 ⫺0.23ⴱⴱ ⫺0.29ⴱⴱ 0.61ⴱⴱ ⫺0.25ⴱⴱ ⫺0.43ⴱⴱ ⫺0.51ⴱⴱ 0.25ⴱⴱ 0.20ⴱ 0.88
10. Interpersonal deviant behaviors 1.74 0.96 ⫺0.22ⴱⴱ ⫺0.32ⴱⴱ 0.62ⴱⴱ ⫺0.18ⴱ ⫺0.28ⴱⴱ ⫺0.28ⴱⴱ 0.26ⴱⴱ 0.11 0.73ⴱⴱ 0.87
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. n ranges from 129 to 154; alphas are in boldface. For gender, 0 ⫽ male, 1 ⫽ female. WOS ⫽ Workplace Ostracism Scale.
This document is copyrighted by the American Psychological Association or one of its allied publishers.


p ⬍ .05. ⴱⴱ p ⬍ .01.

perceived OCB norms). As do the results of the confirmatory experimental studies that have manipulated levels of ostracism
factor analyses, this suggests that the constructs are separate. In (e.g., K. D. Williams et al., 2002; Zadro et al., 2004), but evidence
summary, the results presented in this phase support the conver- of the long-term impact of ostracism is scant. However, it has been
gent and discriminant validity of the WOS insomuch as the WOS proposed that exposure to ostracism will result in internalization of
was shown to be highly related to, yet discriminant from, under- the threats to these needs in the form of lower trait levels of
mining, LMX, interpersonal justice, POS, group cohesion, and self-esteem, belongingness, control, and a meaningful existence
perceived coworker OCB norms and was unrelated to age and (K. D. Williams, 1997). Thus, to the extent that the WOS possesses
perceptions of government environmental policies. criterion-related validity, we hypothesized the following:

Phase 4: Criterion-Related Validity Hypothesis 1: Ostracism will be negatively related to levels of


self-esteem, belongingness, control, and a meaningful existence.
Criterion-related validity, or the extent to which a construct is
related to variables derived from theory, is an important aspect of Well-Being
construct validity (Cronbach & Meehl, 1955; Hinkin, 1998). Thus,
we sought to establish the beginnings of a nomological network to Well-being is the extent to which one experiences optimal
further validate our measure of workplace ostracism. In particular, psychological functioning, broadly defined as subjective feel-
we focused on five classes of variables with which ostracism ings of happiness and lack of stress in one’s life (Diener,
should relate: basic needs, well-being, attitudes, performance, and Sapyta, & Suh, 1998; Ryan & Deci, 2001). Ostracism should be
withdrawal. related to lower levels of well-being for at least three reasons.
First, as mentioned earlier, being ostracized is a painful and
negative event (N. I. Eisenberger et al., 2003). Given their
Basic Needs
relative rarity, negative events have been hypothesized to pro-
Ostracism is thought to be an aversive behavior, in that it duce especially intense emotional and physical reactions, such
threatens the four fundamental needs of belongingness, self- as strain (Taylor, 1991). Second, individuals who are ostracized
esteem, control, and a meaningful existence (K. D. Williams, by definition experience lower levels of social support than do
1997, 2001). Evidence for this assertion has been collected in individuals who are included. Social support has been shown to

Table 4
Descriptive Statistics, Zero Order Correlations, and Alphas for Sample 3

Variable M SD 1 2 3 4 5 6 7 8

1. Age (years) 31.12 8.37


2. Gender 0.60 0.49 0.14
3. WOS 1.48 0.74 ⫺0.01 ⫺0.05 0.96
4. Interpersonal deviant behaviors 1.34 0.63 ⫺0.20 ⫺0.14 0.50ⴱⴱ 0.93
5. Organizational deviant behaviors 1.41 0.64 0.00 ⫺0.07 0.28ⴱⴱ 0.66ⴱⴱ 0.93
6. In-role behavior 4.13 0.56 0.05 ⫺0.04 ⫺0.26ⴱ ⫺0.39ⴱⴱ ⫺0.50ⴱⴱ 0.83
7. OCBI 3.92 0.60 0.12 0.03 ⫺0.15 ⫺0.17 ⫺0.23ⴱ 0.61ⴱⴱ 0.87
8. OCBO 3.93 0.56 0.11 0.06 ⫺0.30ⴱⴱ ⫺0.45ⴱⴱ ⫺0.59ⴱⴱ 0.74ⴱⴱ 0.48ⴱⴱ 0.73

Note. n ⫽ 95; alphas are on the diagonal in boldface. Other than the WOS (Workplace Ostracism Scale) and age, all scales were completed by a work
peer. For gender, 0 ⫽ male, 1 ⫽ female. OCBI ⫽ organizational citizenship behavior, interpersonal; OCBO ⫽ organizational citizenship behavior,
organizational.

p ⬍ .05. ⴱⴱ p ⬍ .01.
1358 FERRIS, BROWN, BERRY, AND LIAN

Table 5
Descriptive Statistics, Zero Order Correlations, and Alphas for Sample 4

Variable M SD 1 2 3 4 5 6 7

1. Age (years) 34.10 9.86


2. Gender 0.45 0.50 0.26ⴱⴱ
3. WOS 2.30 1.06 ⫺0.08 ⫺0.04 0.94
4. OCB norms 4.37 1.28 ⫺0.06 0.01 ⫺0.63ⴱⴱ 0.86
5. Interpersonal justice 3.38 1.05 ⫺0.04 ⫺0.02 ⫺0.62ⴱⴱ 0.66ⴱⴱ 0.94
6. LMX 2.49 0.71 0.03 0.15 ⫺0.43ⴱⴱ 0.47ⴱⴱ 0.51ⴱⴱ 0.91
7. Cohesion 2.49 0.73 0.03 ⫺0.07 ⫺0.53ⴱⴱ 0.69ⴱⴱ 0.69ⴱⴱ 0.53ⴱⴱ 0.91

Note. n ranges from 125 to 110; alphas are on the diagonal in boldface. For gender, 0 ⫽ male, 1 ⫽ female. WOS ⫽ Workplace Ostracism Scale; OCB ⫽
organizational citizenship behavior; LMX ⫽ leader-member exchange.
ⴱⴱ
p ⬍ .01.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

be a critical resource for those combating stressful circum- cognitions that ostracism engenders, we expected that ostracism
stances (Hobfoll, 1989), so we expected that individuals who would be associated with negative workplace attitudes.
were ostracized would show lower levels of well-being. Finally, To test this idea, we examined two of the most studied attitude
being ostracized at work removes not only social support but constructs in organizational behavior: job satisfaction and affective
more tangible forms of support, as existing stressful circum- commitment. Job satisfaction has been defined as “a pleasurable or
stances can be exacerbated by coworkers who refuse to help out positive emotional state resulting from the appraisal of one’s job or
or provide advice. We therefore hypothesized that ostracism job experiences” (Locke, 1976, p. 1300). The experience of ostra-
should be related to poorer well-being, as indexed by an indi- cism is essentially antithetical to the experience of job satisfaction,
vidual’s depression and anxiety levels. in that ostracism is neither pleasant nor positive. As such, ostra-
cism should be negatively related to job satisfaction. With respect
Hypothesis 2: Ostracism will be positively related to anxiety
to affective commitment, affectively committed individuals iden-
and depression
tify with, are involved in, and share the values of an organization
(Meyer & Herscovitch, 2001). When individuals are ostracized by
Attitudes others in the organization, however, their sense of belonging to and
Attitudes develop through the affect and cognitions associ- identification with the organization is undermined. Ostracized
ated with an object (Zanna & Rempel, 1988). Given the social individuals do not have the opportunity to create interpersonal ties,
pain that individuals experience when they are ostracized (N. I. which promote involvement with the organization, and they pre-
Eisenberger et al., 2003), ostracism likely results in negative sumably perceive being ostracized as going against their general
affect (K. D. Williams et al., 2002). Similarly, exclusion has values. As a result, being ostracized should be related to lower
been shown to result in more negative cognitive evaluations of levels of affective commitment. Taking these ideas together, we
individuals (Twenge et al., 2001). Given the negative affect and hypothesized as follows:

Table 6
Descriptive Statistics, Zero Order Correlations, and Alphas for Sample 1, 5-Month Follow-Up

Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. Age (years) 34.16 10.00 0.14 ⫺0.03


2. Gender 0.44 0.50 0.06
3. WOS 1.62 0.62
4. Belongingness 3.61 0.91 ⫺0.01 ⫺0.03 ⫺0.31ⴱ 0.88
5. Meaning 3.61 0.95 ⫺0.01 0.12 ⫺0.31ⴱⴱ 0.88ⴱⴱ 0.91
6. Control 2.98 0.72 ⫺0.05 0.16 ⫺0.27ⴱ 0.68ⴱⴱ 0.74ⴱⴱ 0.71
7. OBSE 3.69 0.71 0.06 0.16 ⫺0.41ⴱⴱ 0.51ⴱⴱ 0.63ⴱⴱ 0.55ⴱⴱ 0.89
8. Depression 3.17 3.17 0.09 0.17 0.37ⴱⴱ ⫺0.57ⴱⴱ ⫺0.59ⴱⴱ ⫺0.44ⴱⴱ ⫺0.54ⴱⴱ 0.88
9. Job satisfaction 3.54 0.97 0.02 0.08 ⫺0.35ⴱⴱ 0.69ⴱⴱ 0.66ⴱⴱ 0.55ⴱⴱ 0.64ⴱⴱ ⫺0.54ⴱⴱ 0.95
10. JSB 2.07 0.82 ⫺0.18 ⫺0.11 0.47ⴱⴱ ⫺0.45ⴱⴱ ⫺0.43ⴱⴱ ⫺0.21 0.35ⴱⴱ 0.29ⴱ ⫺0.46ⴱⴱ 0.89
11. Turnover intentions 3.52 1.56 ⫺0.18 ⫺0.20 0.44ⴱⴱ ⫺0.50ⴱⴱ ⫺0.51ⴱⴱ ⫺0.45ⴱⴱ 0.59ⴱⴱ 0.58ⴱⴱ ⫺0.77ⴱⴱ 0.64ⴱⴱ 0.89
12. OD 1.91 0.71 ⫺0.12 0.10 0.23 ⫺0.23 ⫺0.15 ⫺0.05 ⫺0.17 0.08 ⫺0.25ⴱ 0.28ⴱ 0.29ⴱ 0.79
13. ID 1.53 0.75 ⫺0.01 ⫺0.05 0.23 ⫺0.02 ⫺0.03 0.12 ⫺0.10 0.01 ⫺0.15 0.25ⴱ 0.09 0.57ⴱⴱ 0.86
14. Environmental policies 2.96 0.76 0.07 0.10 0.10 ⫺0.04 ⫺0.09 0.06 ⫺0.38ⴱⴱ 0.13 ⫺0.23 0.08 0.20 0.10 0.19 0.92

Note. N ⫽ 68, alphas are on the diagonal in bold. WOS ⫽ Workplace Ostracism Scale; OBSE ⫽ organization-based self-esteem; JSB ⫽ job search
behaviors; OD ⫽ organizational deviance; ID ⫽ interpersonal deviance.

p ⬍ .05. ⴱⴱ p ⬍ .01.
OSTRACISM 1359

Hypothesis 3: Ostracism will be negatively related to job Method


satisfaction and affective commitment.
Participants and Procedure
Performance The participants were the individuals (or subset of individuals)
from Samples 1, 2, and 3 from Phase 2, with the differences
Modern conceptualizations of job performance suggest that outlined below. Table 1 illustrates which samples included which
overall job performance is best captured by examining three types measures for criterion-related validity purposes.
of behavior: in-role behavior, OCB, and deviant behavior (Dalal, Sample 1. The initial ostracism measure was completed by
2005; Rotundo & Sackett, 2002; Viswesvaran & Ones, 2000). 295 participants. Of these, we invited 144 individuals who had
Given the negative impact of ostracism on self-regulatory ability agreed to complete a follow-up study to participate in a second
(Baumeister et al., 2005), we expected that ostracism would be survey 5 months later. Participants were entered in a drawing with
related to each of these aspects of job performance. a 1-in-20 chance of winning a $75 gift certificate at a national
In-role behavior refers to work-related duties and responsibili- electronics store. We had an initial response rate of 67% (96
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ties that are usually contained within a formal job description participants), although after we screened individuals who had
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(Murphy, 1989; Rotundo & Sackett, 2002). OCB differs from changed jobs in the intervening 5 months, we were left with 68
in-role performance in that OCB is behavior that is not part of a participants. Our final sample and the original 295 participants did
formal job description (Katz, 1964) but that contributes positively not differ in terms of ostracism levels.
to the social and psychological environment of the organization Sample 2. Participants from Sample 2 completed other scales
(Organ, 1988; Rotundo & Sackett, 2002; Van Dyne, Cummings, & for criterion-related purposes (see Table 1) at the same time that
Parks, 1995). Although in-role behavior and OCB differ in terms they completed the ostracism measure. We also asked the original
of their role definitions, both forms of behavior require individuals 142 participants to nominate a significant other (someone who
to regulate themselves in order to maintain the persistence and knew them well) to complete a separate survey. Significant others
effort needed to perform tasks and/or maintain the image of good were contacted via e-mail with a link to a separate online survey;
organizational citizens (Bolino, 1999; Latham & Pinder, 2005). reminder e-mails were sent after 1 and 2 weeks (Dillman, 2000).
In contrast, deviant behavior, defined as behavior that contra- Of the significant others contacted, 123 responded and provided
venes organizational norms (Bennett & Robinson, 2000), can be matched data for our analyses. Significant others who participated
interpreted as a lack of self-control or self-regulation (Baumeister were entered in a drawing for one of two $50 cash prizes.
& Heatherton, 1996; Marcus & Schuler, 2004). Deviant behavior, Sample 3. To get objective reports of the work behavior of
like self-regulation failure, reflects a focus on the short-term ben- focal participants, we asked our 140 participants in Sample 3 to
efits of one’s actions without consideration of the long-term con- provide us with the contact information for a work peer (someone
sequences (Baumeister & Scher, 1988). Thus, given that ostracism who worked under the same supervisor as themselves). Work peers
impairs the individual’s ability to self-regulate (Baumeister et al., were contacted via e-mail with a link to a separate online survey;
2005), we predicted that this self-regulation failure in the work- reminder e-mails were sent after 1 and 2 weeks (Dillman, 2000). In
place would manifest in lower levels of in-role performance and total, 95 work peers completed the survey; each was given $10 in
OCB and increased levels of deviant behavior. More formally, we compensation.
hypothesized as follows:

Hypothesis 4: Ostracism will be negatively related to in-role Measures


performance and OCB and will be positively related to devi-
WOS. We measured workplace ostracism with the 10-item
ant behavior.
scale developed in Phase 1.
Needs. We adapted van Beest and Williams’s (2006) 15-item
Withdrawal needs measure (5 items per need) to assess sense of belonging
(e.g., “I feel like I belong to my work group”; ␣ ⫽ .88), control
Finally, another reaction to being ostracized would be to avoid (e.g., “I feel I have the ability to significantly alter events at work”;
the source of the ostracism. For the individual, withdrawing from ␣ ⫽ .71), and meaningful existence (e.g., “I feel non-existent at
the situation can represent an adaptive response to protect psycho- work”; ␣ ⫽ .91) on a scale of 1 (not at all) to 5 (extremely).
logical needs and avoid the pain associated with ostracism. For Self-esteem was measured with a 5-item organization-based self-
members of organizations, withdrawal can be manifested in esteem measure (e.g., “I make a difference around here”; Van
thoughts and behaviors associated with planning to leave the Dyne & Pierce, 2004; ␣ ⫽ .89) rated on a scale of 1 (strongly
organization. Thinking about leaving is reflected in turnover in- disagree) to 5 (strongly agree).
tentions, or the extent to which the individual is mentally consid- Depression. Depression was assessed with Caplan, Cobb,
ering leaving the organization (Hanisch & Hulin, 1990). Planning French, Van Harrison, and Pinneau’s (1980) five-item depression
to leave the organization is reflected in job search behaviors, such scale (␣ ⫽ .88). Participants indicated the extent to which they felt
as preparing résumés, going on interviews, and scanning the clas- sad, unhappy, good, depressed, blue, and cheerful along a 4-point
sified ads (Blau, 1994). We hypothesized as follows: scale (1 ⫽ Never or a little of the time to 4 ⫽ Most of the time).
Anxiety. We used House and Rizzo’s (1972) seven-item Job-
Hypothesis 5: Ostracism will be positively related to turnover Induced Anxiety Scale (␣ ⫽ .84). Participants responded to ques-
intentions and job search behaviors. tions such as “I have felt nervous before attending meetings.”
1360 FERRIS, BROWN, BERRY, AND LIAN

Job satisfaction. Job satisfaction was assessed with the five- supported: Ostracism was significantly negatively related to self-
item Brayfield and Rothe (1951) measure (in Sample 2; ␣ ⫽ .93) rated job satisfaction measured 5 months later (r ⫽ ⫺.35, p ⬍ .01;
and with three items from the Michigan Organizational Assess- see Table 6) and cross sectionally (r ⫽ ⫺.34, p ⬍ .01; see Table
ment Questionnaire (in Sample 1 follow-up; Rosen, Levy, & Hall, 3) and was significantly negatively related to job satisfaction
2006; ␣ ⫽ .95). For each item, participants were asked to indicate ratings provided by a significant other (r ⫽ ⫺.33, p ⬍ .01; see
their agreement along a 5-point scale (1 ⫽ strongly disagree to Table 3). Similarly, ostracism was significantly negatively related
5 ⫽ strongly agree) with items such as “All in all, I am satisfied to affective commitment (r ⫽ ⫺.24, p ⬍ .01; see Table 3).
with my job” (Rosen, Levy, & Hall, 2006) and “I find real Hypothesis 4 predicted that ostracism would be negatively re-
enjoyment in my work” (Brayfield & Rothe, 1951). lated to in-role performance and OCB and positively related to
Affective commitment. We used Meyer and Allen’s (1997) deviant behavior. This hypothesis was largely supported: Ostra-
six-item affective commitment measure (e.g., “My organization cism was significantly negatively related to work peer ratings of
has a great deal of personal meaning for me”; ␣ ⫽ .86; Brown, the focal participants’ in-role performance (r ⫽ ⫺.26, p ⬍ .05; see
Ferris, Heller, & Keeping, 2007). Table 4). Ostracism was unrelated to deviant behavior assessed
In-role behavior and OCB. We used L. J. Williams and after a 5-month interval (organizational and interpersonal devi-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ance, rs ⫽ .23, ps ⬍ .10; see Table 6) but was significantly


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Anderson’s (1991) scales for in-role behavior and OCB: a seven-


item OCB scale with individuals as the target (e.g., “Help others positively related to self-rated deviant behavior (organizational
who have been absent”; ␣ ⫽ .87); a seven-item OCB scale with the deviance, r ⫽ .61, p ⬍ .01; interpersonal deviance, r ⫽ .62, p ⬍
organization as the target (e.g., “Conserves and protects organiza- .01; see Table 3) and work-peer-rated deviant behavior (organiza-
tional property”; ␣ ⫽ .73); and a seven-item scale of in-role tional deviance, r ⫽ .28, p ⬍ .01; interpersonal deviance, r ⫽ .50,
behavior (e.g., “Fulfills responsibilities specified in job descrip- p ⬍ .01; see Table 4). Although ostracism was significantly
tion”; ␣ ⫽ .83). negatively related to organization-directed citizenship behavior
Deviant behaviors. We used Bennett and Robinson’s (2000) (r ⫽ ⫺.30, p ⬍ .01; see Table 4), it was not related to interpersonal
Workplace Deviance Scale to measure organizational deviance (16 citizenship behavior (r ⫽ ⫺.15, p ⬎ .10; see Table 4).
items, e.g., “Intentionally work slower than you could have Finally, Hypothesis 5 predicted that ostracism would be posi-
worked”; ␣ ⫽ .93) and interpersonal deviance (8 items, e.g., tively related to withdrawal. This hypothesis was supported, in that
“Made fun of someone at work”; ␣ ⫽ .93). Participants indicated ostracism was significantly positively related to turnover inten-
how often they engaged in these behaviors with a 7-point scale tions (r ⫽ .44, p ⬍ .01) and job search behaviors (r ⫽ .47, p ⬍
(1 ⫽ never to 7 ⫽ daily). .01), both of which were measured 5 months later (see Table 6).
Turnover intentions. Turnover intentions were assessed with a Finally, given that we had measured undermining in Sample 1,
five-item scale (Wayne, Shore, & Liden, 1997; ␣ ⫽ .89). Partic- we were able to marshal additional supporting evidence that os-
ipants responded to questions such as “I am seriously thinking tracism is separable from undermining by examining the variables
about quitting my job.” each of them had been theorized to impact. Namely, ostracism was
Job search behaviors. We used a seven-item version of Blau’s theorized to impact belonging, self-esteem, control, and a mean-
(1994) Job Search Behavior Scale (␣ ⫽ .89). For each item, ingful existence, and undermining was not. Consistent with this
participants indicated the frequency with which they had per- theory, undermining was not significantly related to belonging,
formed behaviors (e.g., “Prepared or revised your résumé”) on a control, or a meaningful existence, and most relations between
5-point scale ranging from 1 ⫽ never (0 times) to 5 ⫽ very ostracism and belonging, self-esteem, control, and a meaningful
frequently (at least 10 times). existence remained significant after we had controlled for under-
mining (control, p ⫽ .07; all other ps ⬍ .05). These results indicate
Results that though undermining and ostracism may be highly related, they
have separate nomological networks, as one would expect for
Tables 3, 4, and 6 present the correlations of our criterion- unique constructs. We also examined whether controlling for un-
related validity variables. Hypothesis 1 predicted that ostracism dermining impacted the other significant results between ostracism
would be negatively related to belongingness, self-esteem, control, and other constructs in this sample; in all cases, ostracism still
and meaningful existence. This hypothesis was supported (see demonstrated incremental variance over and above undermining with
Table 6), with ostracism being significantly negatively related to variables such as job search behaviors (r ⫽ .25, p ⬍ .05), turnover
belongingness (r ⫽ ⫺.31, p ⬍ .05), self-esteem (r ⫽ ⫺.41, p ⬍ intentions (r ⫽ .28, p ⬍ .05), and tension (r ⫽ .39, p ⬍ .05).
.01), control (r ⫽ ⫺.27, p ⬍ .05), and meaningful existence (r ⫽
⫺.31, p ⬍ .01). General Discussion
Hypothesis 2 predicted that ostracism would be negatively re-
lated to well-being. Cross-sectional, multiwave, and multisource Despite widespread familiarity with ostracism in everyday life
data provided support for this prediction. Ostracism was signifi- and increasing evidence that ostracism has a multitude of negative
cantly positively related to anxiety (see Table 3), both when effects (K. D. Williams, 2007), the concept of ostracism has had
anxiety was self-rated (r ⫽ .26, p ⬍ .01) and when it was rated by little impact on organizational psychology. Our purpose in the
a significant other (r ⫽ .24, p ⬍ .01). Ostracism was also signif- present study was to create a reliable and valid measure to assess
icantly positively related to depression (see Table 6) measured 5 ostracism in the workplace and assist researchers who are seeking
months later (r ⫽ .37, p ⬍ .01). to examine this important and underresearched area. Our study
Hypothesis 3 predicted that ostracism would be negatively re- included six separate samples, 822 participants, and multiwave and
lated to job satisfaction and commitment. This hypothesis was multisource data. We believe the results presented herein support
OSTRACISM 1361

the psychometric properties of the WOS. Our results are consistent preponderance of the data from experimental studies suggests that
with the view that ostracism may have important implications for individuals act in maladaptive, antisocial ways following ostracism
both organizations and employees (K. D. Williams, 2001). (e.g., Twenge et al., 2001, 2007), but some studies have shown that
Across four separate samples, the psychometric properties of the individuals react in a prosocial manner following ostracism (e.g.,
WOS were supported, as demonstrated by its good factor structure K. D. Williams & Sommer, 1997). Thus, it seems possible that
and internal consistency and by the minimal impact of method moderators exist, such that some individuals react to ostracism in
effects on responses to the WOS. Convergent and discriminant either prosocial or antisocial ways. For example, a recent study
validity were demonstrated through the relation of the WOS with found that individuals with low fear of negative evaluation were
social undermining, interpersonal justice, LMX, POS, group co- more likely to act in a prosocial manner following ostracism,
hesion, perceived coworker OCB norms, age (across four sam- perhaps because they were not afraid of potentially being rejected
ples), perceptions of government environmental policies, and im- by others (Maner, DeWall, Baumeister, & Schaller, 2007). It is
pression management. However, the WOS was separate from these possible that individuals who do not fear negative evaluation may
variables, as indicated by confirmatory factor analyses and Fornell be more likely to engage in helping behaviors, whereas those who
and Larcker’s (1981) test of construct distinctiveness. Moreover, do fear negative evaluations may interpret helping behavior as
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

an item-sort task demonstrated that the ostracism scale items were another opportunity for individuals to reject them and thus subse-
conceptually distinct from workplace aggression, workplace bul- quently refrain from engaging in such behaviors. Researchers
lying, undermining, and interpersonal justice. should measure rejection sensitivity and fear of negative evalua-
Consistent with the notion that ostracism threatens fundamental tions to examine such possibilities.
human needs (self-esteem, belonging, control, and a meaningful
existence), which are posited to be necessary conditions for human Limitations and Future Directions
health, happiness, and productivity (K. D. Williams, 1997, 2001),
ostracism was related to lowered feelings of self-esteem, belong- As with any investigation, there are a number of limitations that
ing, control, and a meaningful existence, as well as to decreased should be acknowledged. First, our intention from the outset was
well-being, more negative work attitudes, and decreased produc- to generate a measure of perceived workplace ostracism from the
tivity. These results indicate that ostracism has negative implica- target’s perspective. In all of our studies, therefore, we relied
tions for individuals, but such outcomes are significant for orga- exclusively upon single-source employee self-reports of perceived
nizations as well. Organizational deviance is estimated to cost workplace ostracism. Self-report data have been criticized because
billions of dollars annually (Bennett & Robinson, 2000), and stress they may be biased by a number of methodological artifacts
in the workplace is related to increased costs for the organization (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Although we
in terms of sick days and insurance costs (Hart & Cooper, 2001; cannot be certain that our workplace ostracism measure is free of
Lofland, Pizzi, & Frick, 2004). Finally, turnover can be associated all methodological bias, we did take precautionary steps to assess
with the loss of intellectual capital and organizational knowledge, the most common biases. We found that impression management
in addition to the costs associated with replacing employees (Ram- concerns and chronic differences in PA/NA did not appear to bias
lall, 2003; Tziner & Birati, 1996). Taken together, these results responses to the ostracism items unduly (L. J. Williams & Ander-
speak to the practical importance of studying ostracism, both for son, 1994; L. J. Williams et al., 1996).
organizations and for individuals. Although research on ostracism Relatedly, in establishing the criterion-related validity of our
is still at a nascent stage, findings would suggest that allowing ostracism scale, we included self-reported outcome data; this inc-
individuals to be ostracized is detrimental to all involved and lulsion raises the possibility that common-method variance may
ultimately should be discouraged by organizations and their rep- have inflated these relationships (Podsakoff et al., 2003). Although
resentatives. In this respect, researchers may wish to focus on it is true that common-method bias may have had some influence
antecedents of ostracism in future studies to identify efficacious on our self-report data, we did take several precautions to mitigate
interventions that could reduce the incidence of ostracism in the the possibility that our data were driven solely by biases. In two of
workplace. our samples we collected outcome data from other sources (i.e.,
The five categories studied (needs, well-being, attitudes, perfor- coworker and significant other), and in a subset of another sample
mance, and withdrawal) represent only a sampling of possible we incorporated a lengthy lag between the assessment of ostracism
outcomes to which ostracism should theoretically relate (and the and outcomes. When it was assessed under these more stringent
specific measures representing each category in turn represent only conditions, the criterion-related validity of our ostracism scale was
a subset of possible measures of, for example, well-being), but our upheld. This finding suggests that ostracism’s association with
purpose in this study was not to list all the outcomes with which relevant organizational outcomes is not simply a consequence of
ostracism should be related. Rather, we intended to use a subset of methodological artifacts.
categories and measures to provide support for the criterion-related Given that all of the relationships reported in this article are cor-
validity of the WOS. However, we encourage researchers to ex- relational, readers should recognize that causal inferences cannot be
amine other outcomes or even other constructs within these cate- reliably drawn from our findings. Although it is theoretically reason-
gories. For example, though we measured global job satisfaction, able to posit that ostracism should influence the behavioral, well-
studies that measured specific facets of job satisfaction (e.g., being, and attitudinal outcomes examined in our research (e.g.,
satisfaction with coworkers, satisfaction with leaders) would illu- Baumeister et al., 2005; K. D. Williams, 2007), if causality and
minate how ostracism influences workplace satisfaction. directionality were to be inferred, potential threats to internal validity
Although most of our hypotheses were supported, ostracism was would first need to be eliminated via alternative research designs
not related to citizenship behavior directed toward individuals. The (Cook, Campbell, & Peracchio, 1990). To circumvent this limitation,
1362 FERRIS, BROWN, BERRY, AND LIAN

we strongly encourage researchers interested in investigating work- ing processes are specified, research can assist in the development
place ostracism to employ longitudinal or cross-lagged panel designs. of targeted workplace interventions to minimize the impact of
However, our purpose in the present study was to design a valid ostracism in the workplace. The four needs of self-esteem, control,
measure of workplace ostracism. In this respect, causality is less of an belongingness, and meaningful existence seem likely candidates
issue, as criterion-related validity is demonstrated by showing a rela- for mediating mechanisms, in that they are posited to be immedi-
tion between ostracism and constructs to which it should theoretically ately affected by ostracism and a substantial literature attests to the
relate. Regardless of whether one conceptualizes ostracism as an ability of these needs to motivate behavior and impact attitudes
antecedent or consequence of variables such as well-being, criterion- (e.g., Baumeister, 1994; Baumeister & Leary, 1995; Branden,
related validity is established simply by demonstrating a relation. 2001; Greenberg et al., 1986; Seligman, 1975). However, there is
Now that a measure of workplace ostracism exists, it can be used to substantial debate over which human needs are “fundamental”
address questions concerning causality. (Sheldon, Elliot, Kim, & Kasser, 2001), and ostracism seems likely
A further potential limitation of our scale is that the WOS was to influence other human needs as well (e.g., competence). Given
not designed to differentiate between different ostracism sources. these numerous contrary statements about our fundamental human
One might argue that there may be different implications of being needs, there is no shortage of potential mediators. In future, re-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ostracized by different work groups. For example, being ostracized searchers should seek to delineate exactly which needs actually
by one’s coworker may result in interpersonal deviance, and being mediate the effects of ostracism.
ostracized by one’s supervisor may result in organizational devi- In addition to research on the ostracism construct itself, we
ance, as leaders can be identified as organizational representatives believe, an important future direction for researchers is to integrate
(Rhoades & Eisenberger, 2002). By separating out the source of ostracism within other theoretical frameworks. For example, there
the ostracism, one can also examine potential cross-domain buff- has recently been increased interest in the effects of demographic
ering of support from coworkers on the effects of supervisor dissimilarity, or the extent to which an individual differs from his
ostracism (or vice versa; Duffy et al., 2002). It may be beneficial or her coworkers on the basis of demographic variables, such as
in future to differentiate the foci of the ostracism (e.g., peers, age, gender, or race (Joshi, 2006; Liao, Joshi, & Chuang, 2004).
supervisors, subordinates). Although we believe that the effect of This research has found that demographic dissimilarity can at
various foci should be considered in future work, our conceptualiza- times result in behaviors such as organizational deviance (Liao et
tion of the construct was driven by theoretical and empirical consid- al., 2004). We suggest that ostracism might be a potential medi-
erations. In this regard, the available empirical literature suggests that ating mechanism of such relationships, in that individuals who
social exclusion and ostracism are equally painful, regardless of who differ from their coworkers may feel ignored or excluded. More
instigates them (Zadro et al., 2004). Similarly, theoretical arguments broadly, placing ostracism within other interpersonal models, such
suggest that, due to evolutionary advantages, humans are hard- as the group value model (Tyler, Degoey, & Smith, 1996) might
wired to respond and react to social exclusion, regardless of its suggest other consequences of ostracism, such as interpersonal
source (K. D. Williams & Zadro, 2001). On the basis of these justice.
theoretical and empirical considerations, we opted to develop a Finally, consistent with calls to examine deviance from the
general workplace ostracism measure. target’s perspective (Bennett & Robinson, 2003), our work has
Although ostracism is a universally painful and aversive expe- examined the effects of ostracism on individuals. However, an
rience, the fact that it may be more so for certain individuals alternative approach is to examine the effects that engaging in
suggests that moderator constructs may interact with ostracism. ostracism have on ostracizers. Preliminary evidence suggests that
For example, highly neurotic or narcissistic individuals have a the act of ostracizing can increase one’s sense of control but
fragile and easily threatened sense of self (Bolger & Schilling, decrease one’s sense of belongingness (K. D. Williams, Wheeler,
1991; Bolger & Zuckerman, 1995; Twenge & Campbell, 2003), & Harvey, 2001); ostracizing coworkers is taxing and may impact
which may make ostracism particularly devastating. Similarly, one’s own performance (Ciarocco, Sommer, & Baumeister, 2001).
individuals who define themselves in terms of their relations with These findings raise the ironic possibility that employees who
others (i.e., an interdependent self-construal; Singelis, 1994) may intentionally ostracize peers, presumably to punish them, may end
find ostracism to be relatively more threatening than do people up punishing themselves in terms of impairing their own perfor-
with an independent self-construal. Although these moderator vari- mance. They also raise the question of why individuals would
ables suggest that ostracism may be more devastating for certain ostracize others, given the negative impact on the ostracizer. More
individuals, we believe that certain variables may moderate the work is needed on what prompts individuals to choose ostracism
effects of ostracism by reducing the likelihood of negative behav- over other negative interpersonal behaviors.
iors. For example, continuance commitment, or the extent to which
people feel compelled to remain at their job because of what they Summary
stand to lose (Meyer & Herscovitch, 2001), may act as a constraint
against poor performance. That is, if employees cannot afford to We believe that the WOS is a reliable and valid measure of
lose their jobs, they may be less likely to engage in deviant workplace ostracism. It is an important tool that can be used to
behavior or to reduce in- or extrarole performance because they examine, among other things, the effects of deviant behavior on
fear losing their jobs. Examination of these and other moderators targets. Ostracism represents a behavior that is qualitatively dif-
will establish possible boundary or exacerbating conditions of ferent from those behaviors examined by existing scales (i.e., it
ostracism and will further our understanding of the construct. does not involve interacting with an individual). It influences our
Another future direction for research is explicating the process fundamental human needs, is aversive, and potentially influences
through which ostracism influences outcomes. When the underly- numerous outcomes important to both employers and employees.
OSTRACISM 1363

Our studies suggest that ostracism is separable from numerous Bolger, N., & Zuckerman, A. (1995). A framework for studying person-
similar constructs (e.g., LMX, POS, interpersonal justice, under- ality in the stress process. Journal of Personality and Social Psychology,
mining, group cohesion, and perceived coworker OCB norms) and 69, 890 –902.
that it can account for incremental validity in outcomes. Taken Bolino, M. C. (1999). Citizenship and impression management: Good
together, our results suggest that ostracism is an important concept soldiers or good actors? Academy of Management Review, 24, 82–98.
Branden, N. (2001). The psychology of self-esteem: A revolutionary ap-
for research, although until now it has been ignored due to the lack
proach to self-understanding that launched a new era in modern psy-
of a measure of the construct. It is our hope that the development chology. San Francisco: Jossey-Bass.
of the WOS will encourage other researchers to examine this Brayfield, A. H., & Rothe, H. F. (1951). An index of job satisfaction.
construct and consider what role ostracism might play in their own Journal of Applied Psychology, 35, 307–311.
research areas. Brown, D. J., Ferris, L. D., Heller, D., & Keeping, L. M. (2007). Ante-
cedents and consequences of the frequency of upward and downward
References social comparisons at work. Organizational Behavior and Human De-
cision Processes, 102, 59 –75.
Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in Burke, M., Brief, A., & George, J. (1993). The role of negative affectivity
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

practice: A review and recommended two-step approach. Psychological in understanding relations between self-reports of stressors and strains:
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Bulletin, 103, 411– 423. A comment on the applied psychology literature. Journal of Applied
Anderson, J. C., & Gerbing, D. W. (1991). Predicting the performance of Psychology, 78, 402– 412.
measures in a confirmatory factor analysis with a pretest assessment of Campbell, D., & Fiske, D. (1959). Convergent and discriminant validation by
their substantive validities. Journal of Applied Psychology, 76, 732–740.
the multitrait–multimethod matrix. Psychological Bulletin, 56, 81–105.
Aquino, K., Lewis, M. U., & Bradfield, M. (1999). Justice constructs,
Campion, M. A., & Lord, R. G. (1982). A control systems conceptualiza-
negative affectivity, and employee deviance: A proposed model and
tion of the goal-setting and changing process. Organizational Behavior
empirical test. Journal of Organizational Behavior, 20, 1073–1091.
& Human Performance, 30, 265–287.
Armeli, S., Eisenberger, R., Fasolo, P., & Lynch, P. (1998). Perceived orga-
Caplan, R. D., Cobb, S., French, J. R. P., Van Harrison, R., & Pinneau,
nizational support and police performance: The moderating influence of
S. R. (1980). Job demands and worker health (ISR Rep. No. 9011). Ann
socioemotional needs. Journal of Applied Psychology, 83, 288 –297.
Arbor: University of Michigan, Institute for Social Research.
Bass, B. M., Cascio, W. F., & O’Connor, E. J. (1974). Magnitude estima-
Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control
tions of expressions of frequency and amount. Journal of Applied
theory approach to human behavior. New York: Springer-Verlag.
Psychology, 59, 313–320.
Ciarocco, N. J., Sommer, K. L., & Baumeister, R. F. (2001). Ostracism and
Baumeister, R. F. (1994). Self-esteem. In V. S. Ramachandram (Ed.),
ego depletion: The strains of silence. Personality and Social Psychology
Encyclopedia of human behavior (pp. 83– 87). San Diego, CA: Aca-
Bulletin, 27, 1156 –1163.
demic Press.
Colquitt, J. A. (2001). On the dimensionality of organizational justice: A
Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., & Twenge, J. M. (2005).
construct validation of a measure. Journal of Applied Psychology, 86,
Social exclusion impairs self-regulation. Journal of Personality and
386 – 400.
Social Psychology, 88, 589 – 604.
Baumeister, R. F., & Heatherton, T. F. (1996). Self-regulation failure: An Cook, T. D., Campbell, D. T., & Peracchio, L. (1990). Quasi experimen-
overview. Psychological Inquiry, 7, 1–15. tation. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for and organizational psychology (Vol. 1, pp. 491–576). Palo Alto, CA:
interpersonal attachments as a fundamental human motivation. Psycho- Consulting Psychologists Press.
logical Bulletin, 117, 497–529. Cortina, J. M. (1993). What is coefficient alpha? An examination of theory
Baumeister, R. F., & Scher, S. J. (1988). Self-defeating behavior patterns and applications. Journal of Applied Psychology, 78, 98 –104.
among normal individuals: Review and analysis of common self- Cronbach, L. J., & Meehl, P. C. (1955). Construct validity in psychological
destructive tendencies. Psychological Bulletin, 104, 3–22. tests. Psychological Bulletin, 52, 281–302.
Baumeister, R. F., & Tice, D. M. (1990). Anxiety and social exclusion. Cropanzano, R., & Mitchell, M. S. (2005). Social exchange theory: An
Journal of Social and Clinical Psychology, 9, 165–195. interdisciplinary review. Journal of Management, 31, 874 –900.
Baumeister, R. F., Twenge, J. M., & Nuss, C. K. (2002). Effects of social Dalal, R. S. (2005). A meta-analysis of the relationship between organi-
exclusion on cognitive processes: Anticipated aloneness reduces intelligent zational citizenship behavior and counterproductive work behavior.
thought. Journal of Personality and Social Psychology, 83, 817– 827. Journal of Applied Psychology, 90, 1241–1255.
Bennett, R. J., & Robinson, S. L. (2000). Development of a measure of DeVellis, R. F. (1991). Scale development: Theory and applications.
workplace deviance. Journal of Applied Psychology, 85, 349 –360. Newbury Park, CA: Sage.
Bennett, R. J., & Robinson, S. L. (2003). The past, present, and future of Diener, E., Sapyta, J. J., & Suh, E. (1998). Subjective well-being is
workplace deviance research. In J. Greenberg (Ed.), Organizational essential to well-being. Psychological Inquiry, 9, 33–37.
behavior: The state of the science (2nd ed., pp. 247–281). Mahwah, NJ: Dillman, D. A. (2000). Mail and Internet surveys: The tailored design
Erlbaum. method (2nd ed.). New York: Wiley.
Berry, C. M., Ones, D. S., & Sackett, P. R. (2007). Interpersonal deviance, Duffy, M. K., Ganster, D. C., & Pagon, M. (2002). Social undermining in
organizational deviance, and their common correlates: A review and the workplace. Academy of Management Journal, 45, 331–351.
meta-analysis. Journal of Applied Psychology, 92, 410 – 424. Edwards, J. R. (2000). Multidimensional constructs in organizational be-
Blau, G. (1994). Testing a two-dimensional measure of job search behavior. havior research: An integrative analytical framework. Organizational
Organizational Behavior and Human Decision Processes, 59, 288 –312. Research Methods, 4, 144 –192.
Blau, G. (1998). On the aggregation of individual withdrawal behaviors Ehrhart, M. G., & Naumann, S. E. (2004). Organizational citizenship
into larger multi-item constructs. Journal of Organizational Behavior, behavior in work groups: A group norms approach. Journal of Applied
19, 437– 451. Psychology, 89, 960 –974.
Bolger, N., & Schilling, E. A. (1991). Personality and problems of every- Eisenberger, N. I., Lieberman, M. D., & Williams, K. D. (2003, October
day life: The role of neuroticism in exposure and reactivity to daily 10). Does rejection hurt? An fMRI study of social exclusion. Science,
stressors. Journal of Personality, 59, 355–386. 302, 290 –292.
1364 FERRIS, BROWN, BERRY, AND LIAN

Eisenberger, R., Huntington, R., Hutchison, S., & Sowa, D. (1986). Perceived relationships on conformity processes. Journal of Abnormal and Social
organizational support. Journal of Applied Psychology, 71, 500 –507. Psychology, 57, 17–24.
Faulkner, S. J., Williams, K. D., Sherman, B., & Williams, E. (1997, May). Johns, G. (1998). Aggregation or aggravation? The relative merits of a
The “silent treatment”: Its incidence and impact. Paper presented at broad withdrawal construct. Journal of Organizational Behavior, 19,
69th Annual Meeting of the Midwestern Psychological Association, 453– 462.
Chicago. Joshi, A. (2006). The influence of organizational demography on the
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models external networking behavior of teams. Academy of Management Re-
with unobservable variables and measurement error. Journal of Market- view, 31, 583–595.
ing Research, 18, 39 –50. Katz, D. (1964). The motivational basis of organizational behavior. Be-
Fox, S., & Stallworth, L. E. (2005). Racial/ethnic bullying: Exploring links havioral Science, 9, 131–146.
between bullying and racism in the U.S. workplace. Journal of Voca- Keeping, L. M., & Levy, P. E. (2000). Performance appraisal reactions:
tional Behavior, 66, 438 – 456. Measurement, modeling, and method bias. Journal of Applied Psychol-
Ganster, D., Hennessey, H., & Luthans, F. (1983). Social desirability ogy, 85, 708 –723.
response effects: Three alternative models. Academy of Management Kozlowski, S. W. J., & Bell, B. S. (2006). Disentangling achievement
Journal, 26, 321–331.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

orientation and goal setting: Effects on self-regulatory processes. Jour-


Gerstner, C. R., & Day, D. V. (1997). Meta-analytic review of leader–
This document is copyrighted by the American Psychological Association or one of its allied publishers.

nal of Applied Psychology, 91, 900 –916.


member exchange theory: Correlates and construct issues. Journal of Latham, G. P., & Pinder, C. C. (2005). Work motivation theory and
Applied Psychology, 82, 827– 844. research at the dawn of the twenty-first century. Annual Review of
Gonsalkorale, K., & Williams, K. D. (2007). The KKK won’t let me play: Psychology, 56, 485–516.
Ostracism even by a despised outgroup hurts. European Journal of Leary, M. R. (2005). Varieties of interpersonal rejection. In K. D. Wil-
Social Psychology, 37, 1176 –1186. liams, J. P. Forgas, & W. V. Hippel (Eds.), The social outcast (pp.
Graen, G., Novak, M., & Sommerkamp, P. (1982). The effects of leader– 35–51). New York: Psychology Press.
member exchange and job design on productivity and job satisfaction: Li, A., & Bagger, J. (2006). Using the BIDR to distinguish the effects of
Testing a dual attachment model. Organizational Behavior and Human impression management and self-deception on the criterion validity of
Performance, 30, 109 –131. personality measures: A meta-analysis. International Journal of Selec-
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to
tion and Assessment, 14, 131–141.
leadership: Development of leader–member exchange (LMX) theory of
Liao, H., Joshi, A., & Chuang, A. (2004). Sticking out like a sore thumb:
leadership over 25 years: Applying a multi-level, multi-domain perspec-
Employee dissimilarity and deviance at work. Personnel Psychology, 57,
tive. Leadership Quarterly, 6, 219 –247.
969 –1000.
Greenberg, J., Pyszczynski, T., & Solomon, S. (1986). The causes and
Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D.
consequences of the need for self-esteem: A terror management theory.
Dunnette (Ed.), Handbook of industrial and organizational psychology
In R. F. Baumeister (Ed.), Public self and private self (pp. 189 –212).
(pp. 1297–1343). Chicago: Rand McNally.
New York: Springer-Verlag.
Lofland, J. H., Pizzi, L., & Frick, K. D. (2004). A review of health-
Hackman, J. R., & Oldham, G. R. (1980). Work redesign. Reading, MA:
related workplace productivity loss instruments. Pharmacoeconom-
Addison-Wesley.
ics, 22, 165–184.
Hanisch, K. A., & Hulin, C. L. (1990). Job attitudes and organizational
MacDonald, G., & Leary, M. R. (2005). Why does social exclusion hurt?
withdrawal: An examination of retirement and other voluntary with-
The relationship between social and physical pain. Psychological Bul-
drawal behaviors. Journal of Vocational Behavior, 37, 60 –78.
letin, 131, 202–223.
Hanisch, K. A., & Hulin, C. L. (1991). General attitudes and organizational
withdrawal: An evaluation of a causal model. Journal of Vocational Maner, J. K., DeWall, C. N., Baumeister, R. F., & Schaller, M. (2007).
Behavior, 39, 110 –128. Does social exclusion motivate interpersonal reconnection? Resolving
Hart, P. M., & Cooper, C. L. (2001). Occupational stress: Toward a more the “porcupine problem.” Journal of Personality and Social Psychology,
integrated framework. In N. Anderson, D. S. Ones, H. K. Sinangil, & C. 92, 42–55.
Viswesvaran (Eds.), Handbook of industrial, work, and organizational Marcus, B., & Schuler, H. (2004). Antecedents of counterproductive be-
psychology: Vol. 2. Organizational psychology (pp. 93–114). Thousand havior at work: A general perspective. Journal of Applied Psychology,
Oaks, CA: Sage. 89, 647– 660.
Hinkin, T. R. (1998). A brief tutorial on the development of measures for Masters, R. D. (1984). Ostracism, voice, and exit: The biology of social
use in survey questionnaires. Organizational Research Methods, 1, participation. Social Science Information, 23, 877– 893.
104 –121. Meston, C. M., Heiman, J. R., Trapnell, P. D., & Paulhus D. L. (1998).
Hitlan, R. T., Kelly, K. M., Schepman, S., Schneider, K. T., & Zarate, Socially desirable responding and sexuality self-reports. Journal of Sex
M. A. (2006). Language exclusion and the consequences of perceived Research, 35, 148 –157.
ostracism in the workplace. Group Dynamics: Theory, Research, and Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace: Theory,
Practice, 10, 56 –70. research, and application. Newbury Park, CA: Sage.
Hobfoll, S. E. (1989). Conservation of resources: A new attempt at con- Meyer, J. P., & Herscovitch, L. (2001). Commitment in the workplace:
ceptualizing stress. American Psychologist, 44, 513–524. Toward a general model. Human Resource Management Review, 11,
Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology 299 –326.
of intergroup relations and group processes. London: Routledge. Mullen, B., & Copper, C. (1994). The relation between group cohesiveness
House, R. J., & Rizzo, J. R. (1972). Role conflict and ambiguity as critical and performance: An integration. Psychological Bulletin, 115, 210 –227.
variables in a model of organizational behavior. Organizational Behav- Munz, D. C., Helsman, T. J., Konold, T. R., & McKinney, J. J. (1996). Are
ior and Human Performance, 7, 467–505. there methodological and substantive roles for affectivity in Job Diagnostic
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance Survey relationships? Journal of Applied Psychology, 81, 795– 805.
structure analysis: Conventional criteria versus new alternatives. Struc- Murphy, K. R. (1989). Dimensions of job performance. In R. F. Dillon &
tural Equation Modeling, 6, 1–55. J. W. Pellegrino (Eds.), Testing: Theoretical and applied perspectives
Jackson, J. M., & Saltzstein, H. D. (1958). The effect of person– group (pp. 218 –247). New York: Praeger.
OSTRACISM 1365

Neuman, J. H., & Baron, R. A. (1998). Workplace violence and workplace The mobilization–minimization hypothesis. Psychological Bulletin, 110,
aggression: Evidence concerning specific forms, potential causes, and 67– 85.
preferred targets. Journal of Management, 24, 391– 419. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social
Organ, D. W. (1988). Organizational citizenship behavior: The good psychological perspective on mental health. Psychological Bulletin, 103,
soldier syndrome. Lexington, England: Lexington Books. 193–210.
Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Tesser, A. (2000). On the confluence of self-esteem maintenance mecha-
Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of social nisms. Personality and Social Psychology Review, 4, 290 –299.
psychological attitudes: Vol. 1. Measures of personality and social Thoresen, C. J., Kaplan, S. A., Barsky, A. P., Warren, C. R., & de
psychological attitudes (pp. 17–59). San Diego, CA: Academic Press. Chermont, K. (2003). The affective underpinnings of job perceptions
Podsakoff, P. M., MacKenzie, S. B., Lee, J., & Podsakoff, N. P. (2003). and attitudes: A meta-analytic review and integration. Psychological
Common method biases in behavioral research: A critical review of the Bulletin, 129, 914 –945.
literature and recommended remedies. Journal of Applied Psychology, Twenge, J. M., Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., &
88, 879 –903. Bartels, J. M. (2007). Social exclusion decreases prosocial behavior.
Porath, C. L., & Bateman, T. S. (2006). Self-regulation: From goal orien- Journal of Personality and Social Psychology, 92, 56 – 66.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tation to job performance. Journal of Applied Psychology, 91, 185–192. Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001). If
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Pyszczynski, T., Greenberg, J., & Solomon, S. (1999). A dual-process you can’t join them, beat them: Effects of social exclusion on aggressive
model of defense against conscious and unconscious death-related behavior. Journal of Personality and Social Psychology, 81, 1058 –1069.
thoughts: An extension of terror management theory. Psychological Twenge, J. M., & Campbell, W. K. (2003). “Isn’t it fun to get the respect
Review, 106, 835– 845. that we’re going to deserve?” Narcissism, social rejection, and aggres-
Pyszczynski, T., Greenberg, J., Solomon, S., Arndt, J., & Schimel, J. sion. Personality and Social Psychology Bulletin, 29, 261–272.
(2004). Why do people need self-esteem? A theoretical and empirical Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2002). Social
review. Psychological Bulletin, 130, 435– 468. exclusion causes self-defeating behavior. Journal of Personality and
Ramlall, S. (2003). Managing employee retention as a strategy for increasing Social Psychology, 83, 606 – 615.
organizational competitiveness. Applied H. R. M. Research, 8, 63–72. Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2003). Social
Rhoades, L., & Eisenberger, R. (2002). Perceived organizational support: exclusion and the deconstructed state: Time perception, meaningless-
A review of the literature. Journal of Applied Psychology, 87, 698 –714. ness, lethargy, lack of emotion, and self-awareness. Journal of Person-
Rosen, C. C., Levy, P. E., & Hall, R. J. (2006). Placing perceptions of ality and Social Psychology, 85, 409 – 423.
politics in the context of the feedback environment, employee attitudes, Tyler, T., Degoey, P., & Smith, H. (1996). Understanding why the justice
and job performance. Journal of Applied Psychology, 91, 211–220. of group procedures matters: A test of the psychological dynamics of the
Rotundo, M., & Sackett, P. R. (2002). The relative importance of task, group-value model. Journal of Personality and Social Psychology, 70,
citizenship, and counterproductive performance to global ratings of job 913–930.
performance: A policy-capturing approach. Journal of Applied Psychol- Tziner, A., & Birati, A. (1996). Assessing employee turnover costs: A
ogy, 87, 66 – 80. revised approach. Human Resource Management Review, 6, 113–122.
Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A van Beest, I., & Williams, K. D. (2006). When inclusion costs and
review of research on hedonic and eudaimonic well-being. Annual ostracism pays, ostracism still hurts. Journal of Personality and Social
Review of Psychology, 52, 141–166. Psychology, 91, 918 –928.
Schacter, S. (1951). Deviation, rejection, and communication. Journal of Vancouver, J. B. (2005). The depth of history and explanation as benefit
Abnormal Psychology, 46, 190 –207. and bane for psychological control theories. Journal of Applied Psychol-
Seashore, S. E. (1954). Group cohesiveness in the industrial work group. ogy, 90, 38 –52.
Ann Arbor: University of Michigan Press. Van Dyne, L., Cummings, L. L., & Parks, J. M. (1995). Extra-role
Seligman, M. E. P. (1975). Helplessness: On depression, development, and behaviors: In pursuit of construct and definitional clarity. Research in
death. San Francisco: Freeman. Organizational Behavior, 17, 215–285.
Sheldon, K. M., Elliot, A. J., Kim, Y., & Kasser, T. (2001). What is Van Dyne, L., & Pierce, J. L. (2004). Psychological ownership and feelings
satisfying about satisfying events? Testing 10 candidate psychological of possession: Three field studies predicting employees attitudes and
needs. Journal of Personality and Social Psychology, 80, 325–339. organizational citizenship behavior. Journal of Organizational Behav-
Singelis, T. M. (1994). The measurement of independent and interdepen- ior, 25, 439 – 459.
dent self-construals. Personality and Social Psychology Bulletin, 20, Viswesvaran, C., & Ones, D. S. (2000). Perspectives on models of job
580 –591. performance. International Journal of Selection and Assessment, 8,
Snoek, J. D. (1962). Some effects of rejection upon attraction to a group. 216 –226.
Journal of Abnormal and Social Psychology, 64, 175–182. Waddell, J. (1999, July 15). Government whistleblowers move closer to
Sommer, K. L., Williams, K. D., Ciarocco, N. J., & Baumeister, R. F. trial. The Ottawa Citizen, p. D8.
(2001). When silence speaks louder than words: Explorations into the Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
intrapsychic and interpersonal consequences of social ostracism. Basic dation of brief measures of positive and negative affect: The PANAS
and Applied Social Psychology, 23, 225–243. scales. Journal of Personality and Social Psychology, 54, 1063–1070.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics. Watson, D., Pennebaker, J., & Folger, R. (1987). Beyond negative affec-
New York: Harper Collins. tivity: Measuring stress and satisfaction in the workplace. In J. M.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup Ivancevich & D. C. Ganster (Eds.), Job stress: From theory to sugges-
behavior. In S. Worchel & W. G. Austin (Eds.), The psychology of tion (pp. 141–157). New York: Haworth Press.
intergroup relations (pp. 7–24). Chicago: Nelson Hall. Wayne, S. J., Shore, L. M., & Liden, R. C. (1997). Perceived organiza-
Tan, J. A., & Hall, R. J. (2005). The effects of social desirability bias on tional support and leader–member exchange: A social exchange per-
applied measures of goal orientation. Personality and Individual Differ- spective. Academy of Management Journal, 37, 765– 802.
ences, 38, 1891–1902. Williams, K. D. (1997). Social ostracism. In R. M. Kowalski (Ed.),
Taylor, S. E. (1991). Asymmetrical effects of positive and negative events: Aversive interpersonal behaviors, (pp. 133–170). New York: Plenum.
1366 FERRIS, BROWN, BERRY, AND LIAN

Williams, K. D. (2001). Ostracism: The power of silence. New York: detection system. In K. D. Williams, J. P. Forgas, & W. von Hippel
Guilford Press. (Eds.), The social outcast: Ostracism, social exclusion, rejection, and
Williams, K. D. (2007). Ostracism. Annual Review of Psychology, 58, bullying (pp. 19 –34). New York: Psychology Press.
425– 452. Williams, L. J., & Anderson, S. E. (1991). Job satisfaction and organiza-
Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: tional commitment as predictors of organizational citizenship and in-role
Effects of being ignored over the Internet. Journal of Personality and behaviors. Journal of Management, 17, 601– 617.
Social Psychology, 79, 748 –762. Williams, L. J., & Anderson, S. E. (1994). An alternative approach to
Williams, K. D., Govan, C. L., Croker, V., Tynan, D., Cruickshank, M., & method effects by using latent-variable models: Applications in organi-
Lam, A. (2002). Investigations into differences between social- and zational behavior research. Journal of Applied Psychology, 79, 323–331.
cyberostracism. Group Dynamics: Theory, Research, and Practice, 6,
Williams, L. J., Gavin, M. B., & Williams, M. L. (1996). Measurement and
65–77.
nonmeasurement processes with negative affectivity and employee atti-
Williams, K. D., & Sommer, K. L. (1997). Social ostracism by coworkers:
tudes. Journal of Applied Psychology, 81, 88 –101.
Does rejection lead to loafing or compensation? Personality and Social
Zadro, L., Williams, K. D., & Richardson, R. (2004). How low can you go?
Psychology Bulletin, 23, 693–706.
Williams, K. D., Wheeler, L., & Harvey, J. A. R. (2001). Inside the social Ostracism by a computer is sufficient to lower self-reported levels of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

mind of the ostracizer. In J. P. Forgas, K. D. Williams, & L. Wheeler belonging, control, self-esteem, and meaningful existence. Journal of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(Eds.), The social mind: Cognitive and motivational aspects of interper- Experimental Social Psychology, 40, 560 –567.
sonal behavior (pp. 294 –320). New York: Cambridge University Press. Zadro, L., Williams, K. D., & Richardson, R. (2005). Riding the “O” train:
Williams, K. D., & Zadro, L. (2001). Ostracism: On being ignored, Comparing the effects of ostracism and verbal dispute on targets and
excluded, and rejected. In M. R. Leary (Ed.), Interpersonal rejection (pp. sources. Group Processes & Intergroup Relations, 8, 125–143.
21–53). New York: Oxford University Press. Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old
Williams, K. D., & Zadro, L. (2005). Ostracism: The indiscriminate early concept. New York: Cambridge University Press.

Appendix

Substantively Valid Ostracism Items


1. Others ignored you at work.
2. Others left the area when you entered.
3. Your greetings have gone unanswered at work.
4. You involuntarily sat alone in a crowded lunchroom at work.
5. Others avoided you at work.
6. You noticed others would not look at you at work.
7. Others at work shut you out of the conversation.
8. Others refused to talk to you at work.
9. Others at work treated you as if you weren’t there.
10. Others at work did not invite you or ask you if you wanted anything when they went out for a coffee break.
11. You have been included in conversations at work (reverse coded).ⴱ
12. Others at work stopped talking to you.ⴱ
13. You had to be the one to start a conversation in order to be social at work.ⴱ

Note. On the basis of Bass, Cascio, and O’Connor (1974), we used a seven-point Likert-type response scale (1 ⫽ Never,
2 ⫽ Once in a while, 3 ⫽ Sometimes, 4 ⫽ Fairly often, 5 ⫽ Often, 6 ⫽ Constantly, 7 ⫽ Always). An asterisk indicates that
an item was not included in the final scale.

Received November 29, 2007


Revision received April 21, 2008
Accepted April 28, 2008 䡲

Potrebbero piacerti anche