Professional Documents
Culture Documents
Management Article 2015
Management Article 2015
Ostracism, or being ignored or excluded by others (Ferris, Brown, Berry, & Lian, 2008; Williams,
2001), represents a pervasive and universal social
phenomenon dating back to the beginning of recorded history (Forsdyke, 2005). Surveys have
found approximately 75% of respondents have experienced ostracism (e.g., from loved ones;
Faulkner, Williams, Sherman, & Williams, 1997),
and one survey found 66% of employees reported
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holders express
written permission. Users may print, download, or email articles for individual use only.
280
February
FIGURE 1
The Moderated Mediation Relationship between Workplace Ostracism and Job Performance
2015
281
282
February
2015
283
will not seek to verify their negative self-perceptions but will instead seek to portray themselves in
the most positive light possible (Sedikides, 1993).
We argue that a motive to self-enhance is particularly pointed in domains that self-worth is contingent upon. That is, because their self-worth is heavily invested in success and failure in contingent
domains, individuals will strive to avoid failure in
such domains, as failure in a contingent domain
arouses feelings of shame and anxiety that induce
fluctuations in state self-esteem (Crocker, Karpinski, Quinn, & Chase, 2003a; Deci & Ryan, 1995).
Given the potential negative self-relevant implications associated with failure in a contingent domain, it has been proposed that this desire to selfenhance and to demonstrate competence in a
contingent domain will override the desire to act in
a way that is consistent with an individuals selfperceptions (Ferris et al., 2009b, 2010). In other
words, rather than engaging in poor performance
consistent with self-esteem level when self-esteem
level is low (Korman, 1976), when an individuals
self-esteem is contingent on a domain, the individual should instead seek to do well in that domain
regardless of self-esteem level.
More generally, this suggests that contingent selfesteem moderates the effects of self-esteem level,
with self-esteem level not predicting an individuals behavior in a domain if the individuals selfesteem is contingent on that domain. With respect
to organizational settings, the extent to which the
individuals self-esteem is contingent on workplace
performance, or the importance of performance to
self-esteem (hereafter, IPSE), should similarly
moderate the relation of self-esteem level to job
performance (Ferris et al., 2010). That is, when
self-esteem is contingent on demonstrating good
performance at work, individuals should strive to
maintain high levels of job performance regardless
of their self-esteem level (i.e., individuals should
self-enhance, not self-verify). For those individuals
with high IPSE, poor job performance would arouse
aversive feelings of shame, and, hence, individuals
would be strongly motivated to demonstrate good
job performance at all times (Deci & Ryan, 1995).
Taken together, this theoretical framing of when
we self-enhance and when we self-verify, and its
concomitant implications for the relation of ostracism to behavioral outcomes, can be operationalized as a moderated mediation model explaining
the effects of ostracism on job performance (see
Figure 1). In particular, consistent with a sociometer perspective, we would expect ostracism to
284
February
cruitment advertisements invited employed individuals to participate in a study on workplace attitudes and behaviors, and described participation
procedures (e.g., completing two online surveys)
and remuneration ($10 and a chance to win one of
two $100 prizes). Interested individuals were directed to complete an online pre-screen questionnaire, through which demographic information
was assessed (to ensure participants were full-time
workers) as well as the frequency with which they
interacted with other people at work (to ensure
participants interacted regularly with other organizational members). We obtained 569 individuals
who completed our pre-screen questionnaire; 398
fulfilled our pre-screen requirements and were sent
emails with a unique identifier code and links to
the online surveys at two points in time. The first
survey included a measure of workplace ostracism;
participants were also asked to nominate a work
peer (by providing a name and email address) who
could be contacted by the researchers. The second
survey, sent out approximately one week after the
completion of the first survey, assessed self-esteem
level. Finally, a link to an online survey was
emailed to work peers of those participants who
had completed the second survey, assessing focal
participants IPSE, IRBs, and OCBs. In order to
maximize response rates, we sent three reminder
emails (one week apart) to individuals who had not
completed the survey (Dillman, 2000).
Sample B. Procedures were largely the same as
those for Sample A, with the exception that participants2 were recruited using broader methods (advertisements placed in commuter and public areas,
newspapers, online forums) and compensation differed slightly ($10). Of 1,200 individuals who completed our pre-screen questionnaire, 761 met our
pre-screen requirements. The survey included measures of ostracism from coworkers and self-esteem
level. A link to an online survey was emailed to
METHOD
Procedure
Sample A. Participants1 were recruited through
advertisements posted to online forums. The re1
This sample was also used as Study 1 in Lian, Ferris,
and Brown (2012a); however, that study addressed a
different research question using a different theoretical
framework than the present study. Moreover, the measures used in this study differ from those used in the
previously published study.
2015
work peers of those participants who had completed our survey, assessing focal participants
IPSE and deviant behaviors.
Participants
Sample A. Out of the 398 invites emailed to
potential participants, 297 individuals responded
and completed the first survey (75% response rate),
and 274 completed the second survey (92% retention rate). We next emailed a link to an online
survey to the work peers of those participants who
had completed the second survey; 158 responded
(58% response rate). Overall, we had 158 matched
data for our analyses.3 The mean age of focal participants was 32.28 years (SD 8.60) and the average hours worked per week was 39.61 (SD
4.51). Participants reported being employed in
their current organization an average of 3.86 years
(SD 4.15) and having worked in their present
position for 2.90 years (SD 3.72). Participants
(45% male) came from a diverse set of occupations
(e.g., clerk, technician, manager, accountant, consultant) and were employed in a variety of industries, including computers and mathematics (18%),
business and finance (14%), education (10%), sales
(8%), and government (7%).
The mean age of work peers (43% male) was
34.73 years (SD 9.05) and the average hours
worked per week was 39.94 (SD 5.69). To ensure
nominated work peers were qualified to report focal participants workplace behaviors, focal participants were asked to provide the contact of a work
peer with whom they worked closely and who
knew them well. Work peers were also asked how
well they knew focal participants; results indicated that work peers knew focal participants
fairly well (M 5.31 on a seven-point Likert scale
where 1 not at all and 7 extremely well;
SD 1.19). The work peers name (to whom the
survey link and the payment check was sent), email
address (where the survey link was sent), and residential address (where the check for payment was
sent) were checked to ensure focal participants
did not complete peer surveys.
3
We conducted t tests to examine whether focal participants whose work peers responded to our surveys
differed from those whose work peers did not respond.
No significant differences between the two groups were
found for age (t 1.58, p .10), gender (t .62, p
.10), tenure (t .52, p .10), workplace ostracism (t
.06, p .10), and self-esteem level (t .81, p .10).
285
286
original wording, which did not differentiate between ostracism from supervisors or coworkers
(e.g., Others ignored you at work), while Sample
B used a modified version that assessed ostracism
from coworkers specifically (e.g., Coworkers ignored you at work). Participants responded using
a seven-point Likert scale (1 never and 7
always).
Self-esteem level. Rosenbergs (1965) 10-item
self-esteem scale was used, with participants indicating their agreement with statements on a ninepoint Likert scale (1 very strongly agree and
9 very strongly disagree).
IPSE. Crocker et al.s (2003b) four-item competence-contingent self-esteem scale was adapted to
assess the importance of performance to self-esteem (see Ferris et al., 2010, for items).5 Work peers
rated the extent to which participants self-esteem
was contingent on being competent in the workplace (e.g., Doing well at work gives him/her a
sense of self-respect). Responses were made on a
seven-point scale (1 strongly disagree and
7 strongly agree).
IRBs and OCBs. Williams and Andersons (1991)
21-item scale was used to assess IRBs and OCBs in
Sample A. Of the 21 items, 7 measured interpersonally directed OCBs (OCBIs; e.g., Helps others
who have been absent), 7 measured organizationally directed OCBs (OCBOs; e.g., Conserves and
protects organizational property), and the other 7
measured IRBs (e.g., Meets formal performance
requirements of the job). Work peers indicated
their agreement regarding the extent to which their
coworkers engaged in each behavior on a five-point
Likert scale (1 strongly disagree and 5
strongly agree).
Workplace deviance. We assessed deviant behaviors with Bennett and Robinsons (2000) deviance scale adapted to measure workplace deviant
behaviors from a work peers perspective in Sample
B. Of the 19 items, 12 measured organizational
deviance (e.g., My work peer took property from
work without permission), and 7 measured inter5
February
personal deviance (e.g., My work peer said something hurtful to someone at work). Work peers
indicated the frequency with which focal participants engaged in each behavior over the past year
on a seven-point Likert scale (1 never and
7 daily).
Data Analysis
Following Edwards and Lamberts (2007) framework, our moderated mediation hypothesis was
tested by showing that moderation occurs between
the mediating variable and the dependent variables
(i.e., between self-esteem level and IRB, OCB, and
workplace deviance) and that mediating effects
vary according to the level of the moderator (i.e.,
IPSE; Edwards & Lambert, 2007). We used two multiple regression models to test our moderated mediation hypothesis. The first model tested whether
workplace ostracism predicted self-esteem level
(Edwards & Lambert, 2007: Equation 3). The second
model included workplace ostracism, self-esteem
level, IPSE, and the interaction between self-esteem
level and IPSE in the regression equation with job
performance (i.e., IRB, OCB, and workplace deviance) as the dependent variables (Edwards & Lambert, 2007, Equation 10). Self-esteem level and IPSE
were centered prior to computing the interaction
term to reduce multicollinearity.
Integrating these two models (Edwards & Lambert, 2007: Equation 12), we used simple effects
analyses to calculate the strength of the direct effects of self-esteem level on job performance, and
the strength of the mediating (i.e., indirect) effects
of self-esteem level at both high and low levels of
the moderator (i.e., IPSE). Significance tests of
these direct effects, indirect effects, and their differences between high and low levels of the moderator were conducted with a bootstrap approach,
which produced 1,000 bootstrapping samples and
allowed us to construct bias-corrected confidence
intervals for each significance test (Edwards & Lambert, 2007; Mooney & Duval, 1993). The bootstrap
approach is more advantageous than the more commonly used Sobel test approach as it overcomes the
high Type I error rate due to the violation of normal
distribution assumptions (Shrout & Bolger, 2002).
RESULTS
Table 1 presents the means, standard deviations,
alphas, and correlations of the measured variables
for Sample A; Table 2 presents the same for Sample
2015
TABLE 2
Descriptive Statistics, Zero-Order Correlations, and
Alphas (Sample B)a
TABLE 1
Descriptive Statistics, Zero-Order Correlations, and
Alphas (Sample A)a
Mean SD
1.
2.
3.
4.
5.
6.
WOS
IPSE
SE
OCBI
OCBO
IRB
1.90
5.09
3.71
4.02
3.89
4.08
1.13
1.07
.65
.57
.64
.57
.96
.18*
.33**
.33**
.38**
.43**
.85
.02
.38**
.28**
.35**
Mean SD
.87
.24** .85
.22** .59** .82
.26** .68** .77** .82
n 158. Female 0, male 1. WOS workplace ostracism; IPSE importance of performance to self-esteem; SE
self-esteem; OCBI individually directed organizational citizenship behaviors; OCBO organizationally directed organizational citizenship behaviors; IRB in-role behaviors. The numbers in bold on the diagonal are the alphas.
* p .05
** p .01
a
287
1.
2.
3.
4.
5.
n 273. Female 0, male 1. WOS workplace ostracism; IPSE importance of performance to self-esteem; SE
self-esteem; OD organizational deviance; ID interpersonal
deviance. The numbers in bold on the diagonal are the alphas.
** p .01
a
Sample A
Sample B
Intercept
WOS
R2
3.71** (.05)
.19** (.04)
.11**
3.87** (.04)
.29** (.03)
.21**
a
n 158 and 273 for Sample A and Sample B, respectively.
Female 0, male 1. WOS workplace ostracism. Workplace
ostracism refers to coworkers in Sample B. Values are unstandardized regression coefficients (standard error estimates are
listed in parentheses).
** p .01
288
February
TABLE 4
Self-Esteem Level by Importance of Performance to Self-Esteem Predicting OCBs, IRBs, and Deviancea
Step 1
Step 2
Variable
OCBI
OCBO
IRB
OD
ID
Intercept
WOS
SE
IPSE
R2
Intercept
WOS
SE
IPSE
SE IPSE
R2
4.02** (.04)
.11** (.04)
.14* (.08)
.18** (.04)
.24**
4.02** (.05)
.10** (.04)
.17* (.08)
.18** (.04)
.12* (.07)
.02*
.26
3.89** (.05)
.17** (.04)
.12 (.08)
.13** (.04)
.20**
3.89** (.05)
.16** (.04)
.15 (.08)
.13** (.04)
.13* (.07)
.02*
.22
4.08** (.04)
.16** (.04)
.13* (.06)
.15** (.04)
.28**
4.09** (.04)
.15** (.04)
.18** (.06)
.16** (.04)
.18** (.05)
.05**
.33
1.67** (.05)
.47** (.05)
.10 (.08)
.16** (.05)
.38**
1.65** (.05)
.44** (.05)
.14 (.08)
.16** (.05)
.26** (.07)
.03**
.41
1.61** (.05)
.48** (.05)
.06 (.08)
.11* (.05)
.36**
1.60** (.05)
.45** (.05)
.10 (.08)
.11* (.05)
.20** (.07)
.02**
.38
Overall R2
a
n 158 and 273 for Sample A and Sample B, respectively. Female 0, male 1. WOS workplace ostracism; SE self-esteem;
IPSE importance of performance to self-esteem; OCBI individually directed organizational citizenship behaviors; OCBO organizationally directed organizational citizenship behaviors; IRB in-role behaviors; OD organizational deviance; ID interpersonal
deviance. Values are unstandardized regression coefficients (standard error estimates are listed in parentheses). All lower-order terms used
in interactions were standardized prior to analysis.
* p .05
** p .01
TABLE 5
Analysis of Simple Effectsa
Dependent Variables
IPSE
Direct Effects
(PYX)
PMX
PYM
Indirect Effects
(PYMPMX)
Total Effects
(PYX PYMPMX)
OCBI
Low
High
Differences
Low
High
Differences
Low
High
Differences
Low
High
Differences
Low
High
Differences
.10**
.10**
.00
.16**
.16**
.00
.15**
.15**
.00
.44**
.44**
.00
.44**
.44**
.00
.19**
.19**
.00
.19**
.19**
.00
.19**
.19**
.00
.29**
.29**
.00
.29**
.29**
.00
.30**
.04
.26*
.29*
.01
.28*
.37**
.01
.39**
.40**
.12
.52**
.30**
.10
.40**
.06**
.01
.05*
.06*
.00
.06*
.08**
.00
.08**
.12**
.03
.15**
.09*
.03
.12**
.16**
.11**
.05*
.22**
.16**
.06*
.23**
.15**
.08**
.56**
.41**
.15**
.54**
.42**
.12**
OCBO
IRB
OD
ID
a
PYX path from workplace ostracism to dependent variables; PMX path from workplace ostracism to self-esteem; PYM path from
self-esteem to dependent variables.
* p .05
** p .01
2015
289
FIGURE 2
Interaction between Self-Esteem and IPSE
on OCBI
FIGURE 4
Interaction between Self-Esteem and IPSE on IRB
FIGURE 3
Interaction between Self-Esteem and IPSE
on OCBO
DISCUSSION
In the present paper, we argue that whether or
not individuals self-verify or self-enhance depends
on contingencies of self-esteem; integrating this
prediction with ostracism models suggesting that
self-esteem level acts as a mediator of ostracisms
effects, we proposed and tested a novel model to
better understand how workplace ostracism influences job performance. Across two field samples
using multi-source, multi-wave data, our results
help to uncover both mediating and moderating
mechanisms responsible for ostracisms effect on
employee job performance. In so doing, our study
makes contributions to the ostracism, self-esteem,
and motivation literatures.
FIGURE 5
Interaction between Self-Esteem and IPSE on
Peer-Rated Organizational Deviance
290
FIGURE 6
Interaction between Self-Esteem and IPSE on
Peer-Rated Interpersonal Deviance
February
FIGURE 8
The Indirect (Mediated) Effect of Workplace
Ostracism on OCBO at High and Low Levels
of IPSE
FIGURE 7
The Indirect (Mediated) Effect of Workplace
Ostracism on OCBI at High and Low Levels
of IPSE
FIGURE 9
The Indirect (Mediated) Effect of Workplace
Ostracism on IRB at High and Low Levels
of IPSE
2015
FIGURE 10
The Indirect (Mediated) Effect of Workplace
Ostracism on Peer-Rated Organizational
Deviance at High and Low Levels of IPSE
see Hiller & Hambrick, 2005, for potential organizational problems associated with overly positive
self-views). Second, by outlining a moderator of
ostracisms effects, a better understanding is gained
regarding for whom ostracism will relate to job
performance. In particular, only those whose selfesteem was not contingent upon workplace performance evinced performance decrements in the
face of workplace ostracism. This suggests that
organizations may wish to implement programs
designed to select or develop employees with high
IPSE. However, caution should be exercised prior
to undertaking such steps, as previous studies
have also shown that contingent self-esteem can
lead to negative consequences for individuals
(e.g., feeling depressed following negative feedback in the contingent domain; Crocker et al.,
2003b). Given workplace ostracism and contingent
self-esteem research is still relatively new, more
research is necessary prior to making firm suggestions to managers.
Implications for Self-Esteem and
Motivation Literatures
Aside from contributions to the ostracism and
negative interpersonal behavior literatures, our reFIGURE 11
The Indirect (Mediated) Effect of Workplace
Ostracism on Peer-Rated Interpersonal Deviance
at High and Low Levels of IPSE
291
292
February
2015
293
294
February
Crocker, J., & Wolfe, C. T. 2001. Contingencies of selfworth. Psychological Review, 108: 593 623.
2015
295
Leary, M. R., Gallagher, B., Fors, E., Buttermore, N., Baldwin, E., Kennedy, K., & Mills, A. 2003. The invalid-
296
February
2015
297