THIS ARTICLE HAS BEEN CORRECTED. SEE LAST PAGE
Journal of Applied Psychology
2010, Vol. 95, No. 6, 1009 –1031
© 2010 American Psychological Association
0021-9010/10/$12.00 DOI: 10.1037/a0020540
Self-Gain or Self-Regulation Impairment? Tests of Competing Explanations
of the Supervisor Abuse and Employee Deviance Relationship Through
Perceptions of Distributive Justice
Stefan Thau
Marie S. Mitchell
London Business School
University of Georgia
Two competing explanations for deviant employee responses to supervisor abuse are tested. A self-gain
view is compared with a self-regulation impairment view. The self-gain view suggests that distributive
justice (DJ) will weaken the abusive supervision– employee deviance relationship, as perceptions of fair
rewards offset costs of abuse. Conversely, the self-regulation impairment view suggests that DJ will
strengthen the relationship, as experiencing abuse drains self-resources needed to maintain appropriate
behavior, and this effect intensifies when employees receive inconsistent information about their
organizational membership (fair outcomes). Three field studies using different samples, measures, and
designs support the self-regulation impairment view. Two studies found that the Abusive Supervision ⫻
DJ interaction was mediated by self-regulation impairment variables (ego depletion and intrusive
thoughts). Implications for theory and research are discussed.
Keywords: abusive supervision, deviance, retaliation, self-regulation impairment
to reciprocate behavior when the gains of the behavior are greater
than its costs (Cook & Emerson, 1978; Gergen, 1980; Meeker,
1971; Pruitt, 1968). The presumed gains are that retaliation is
satisfying and enjoyable (Gouldner, 1960; Knutson, 2004; Tripp,
Bies, & Aquino, 2002), balances the exchange (Gouldner, 1960;
Molm, Quist, & Wiseley, 1994), demonstrates a capability of
defending the self (Bies, 1987; Bies & Tripp, 1995, 1996), and/or
can potentially deter future harm (Bies, 1987; Bies & Tripp, 1995,
1996; Gouldner, 1960; Knutson, 2004; Skarlicki & Folger, 1997;
Tripp & Bies, 1997). Indeed, a number of studies have found that
employees harm the organization, supervisor, and other members
with deviant behaviors to pay back organizational authorities for
abuse (Detert, Treviño, Burris, & Andiappan, 2007; Duffy, Ganster, & Pagon, 2002; Mitchell & Ambrose, 2007; Tepper et al.,
2009; Tepper, Henle, Lambert, Giacalone, & Duffy, 2008; Thau,
Bennett, Mitchell, & Marrs, 2009).
The self-gain perspective is parsimonious, but is also limited
and possibly inaccurate. In the context of ongoing and hierarchical
work relationships, responding to abuse with deviance— behavior
that violates norms and is harmful to the organization and its
members (Bennett & Robinson, 2000; S. L. Robinson & Bennett,
1995)—likely causes more costs than gains for victims (Aquino,
Tripp, & Bies, 2001; Tepper et al., 2009). Retaliation violates
standards of appropriate behavior and incurs a costly risk of
counterretaliation from both the supervisor and the organization
(Aquino et al., 2001; Heider, 1958; Tepper et al., 2009). Hence,
self-gain principles do not so easily explain why employees reciprocate abuse with deviance. This begs the question of why would
victims of an abusive supervisor fail to maintain appropriate or
rational behavior.
One potential answer to this question is the idea that the victimization experience promotes self-regulation impairment
(Baumeister, 2001; Thau, Aquino, & Poortvliet, 2007). The experience of abuse challenges victims to process, interpret, and un-
An often replicated finding is that employees who are victims of
supervisor abuse are more likely to engage in behaviors that are
harmful to organizations and their members. For example, employees who are belittled, put down, or screamed at by authorities are
more likely to come in late without consent or lash out against
members of the organization. This phenomenon seems to be robust
across samples, measures, and contexts (for reviews, see Aquino &
Thau, 2009; Tepper, 2007) and has been parsimoniously explained
by various exchange theories (e.g., social exchange theory, Homans, 1961; reciprocity principles, Gouldner, 1960; equity theory,
Adams, 1965). Accordingly, employees view their relationship
with their supervisor as a form of social exchange in which each
party seeks to reciprocate the costs and benefits received and
maintain balanced contributions. Arguably, the psychological and
sometimes physical costs employees experience through abuse
exceed the benefits, creating an exchange imbalance.
Negative reciprocity principles (Gouldner, 1960) suggest that
individuals attempt to resolve exchange imbalances in a quid pro
quo fashion such that employees who are harmed by organizational authorities retaliate by harming the organization, supervisors, and other organizational members (Gouldner, 1960; Mitchell
& Ambrose, 2007; Skarlicki & Folger, 1997). A self-gain assumption is fundamental to these explanations: People rationally choose
This article was published Online First August 16, 2010.
Stefan Thau, Organizational Behavior Subject Area, London Business
School, London, England; Marie S. Mitchell, Department of Management,
Terry College of Business, University of Georgia.
We thank Lance Ferris, Gillian Ku, Madan Pillutla, Niro Sivanathan,
Christian Tröster, and Ryan Vogel for helpful comments on previous
versions of this article.
Correspondence concerning this article should be addressed to Stefan
Thau, Organizational Behavior Subject Area, London Business School,
Regent’s Park, London NW1 4SA, England. E-mail: sthau@london.edu
1009
1010
THAU AND MITCHELL
derstand the causes and consequences of being harmed. These
self-regulatory activities drain needed self-resources (like attention
and willpower) to consider whether responding to abuse with
deviance would violate normative expectations and provoke counterretaliation. In short, the experience of abuse promotes selfregulation impairment, which supersedes normative and rational
considerations, and the self-regulation impairment prompted by
abuse explains why employees reciprocate abuse with deviance
(Aquino & Thau, 2009; Thau et al., 2007).
Both self-gain and self-regulation impairment views provide
plausible explanations of deviant reactions to supervisor abuse.
Yet, past research has not offered a test of these two competing
explanations. This article addresses this gap in the literature. We
explore the influence of distributive justice (DJ) on the abusive
supervision– employee deviance relationship. DJ involves employees’ perceptions of fairness about the exchange of outcomes allocated by the organization compared with their contributions to the
organization (Colquitt, 2001; Cropanzano & Ambrose, 2001;
Deutsch, 1975). We posit that a comparative test is possible when
taking into account that employees differentiate among various
exchanges (Cropanzano & Mitchell, 2005; Lavelle, Rupp, &
Brockner, 2007) and that abused employees may simultaneously
experience high or low levels of DJ. Lavelle et al. (2007) argued,
“It is possible for employees to experience a high level of social
exchange with one party (e.g., coworkers) and at the same time
report relatively low levels of social exchange with another (e.g.,
the organization)” (p. 847). Indeed, research shows that different
forms of exchange jointly predict behavior (e.g., Erdogan & Enders, 2007; Sluss, Klimchak, & Holmes, 2008; Spencer & Rupp,
2009; Tangirala, Green, & Ramanujam, 2007). Empirical evidence
also shows that reactions to abusive supervision are influenced by
perceptions of other exchanges (e.g., Tepper et al., 2008; Thau et
al., 2009). We extend these ideas and suggest that DJ can provide
a strong inference test (Platt, 1964) to compare the self-gain and
self-regulation impairment views because each makes opposing
predictions for how DJ moderates the relationship between abusive
supervision and employee deviance.
The self-gain view suggests that fair distributions would weaken
the abusive supervision– employee deviance relationship. Compensatory exchange principles (Adams, 1965; Ekeh, 1974; Gergen,
1980; Homans, 1961; Walster, Walster, & Berscheid, 1978) imply
that high levels of DJ make for a beneficial aspect of the organizational membership, which should offset the costs of abuse.
Conversely, the self-regulation impairment view suggests that fair
distributions should strengthen the abusive supervision– employee
deviance relationship because DJ communicates information about
the benefits of one’s organizational membership that is inconsistent with the costs of supervisor abuse (e.g., “My boss mistreats
me” but “My organization compensates me fairly for what I do”).
Employees’ attention and attempts to deal with the inconsistent
information (Baumeister & Vohs, 2007; Richards & Gross, 1999;
J. L. Robinson & Demaree, 2007) further drain self-regulatory
abilities and make deviance even more likely.
By providing a comparative test of the self-gain versus selfregulation impairment views, our research makes four contributions to
the literature. First, we tested the traditional, dominant view of retaliation to supervisor abuse (self-gain) against the self-regulation impairment view. We recognize that previous studies provide indirect
evidence of self-regulation impairment. One study found that em-
ployees abused by their supervisors become psychologically exhausted (Tepper, 2000); psychologically exhausted employees are
drained emotionally, cognitively, and physically, which implies
that self-resources are drained. Indeed, DeJordy’s (2008) theoretical framework assumes that psychological exhaustion is a result
of drained self-resources. Another study found that employees
excluded from their work group engage in interpersonally deviant
and self-defeating behaviors (Thau et al., 2007). Another study
found that the ability to control the self was negatively related to
counterproductive behaviors (Marcus & Schuler, 2004). There are
also data showing that experimentally induced self-regulation impairment promotes aggression (e.g., Baumeister, DeWall, Ciarocco, & Twenge, 2005; Bushman & Baumeister, 1998; Twenge,
Baumeister, Tice, & Stucke, 2001) and predicts retaliatory responses in people who were insulted (DeWall, Baumeister, Stillman, & Gailliot, 2007). Overall, these studies suggest that supervisor abuse may prompt self-regulation impairment, which then
influences deviant responses. Still, none of these studies provided
a full test of the relationships of abuse, self-regulation impairment,
and deviance. More importantly, they did not provide a test of the
self-gain versus self-regulation impairment views. Our study extends this research agenda.
A second theoretical and practical contribution of our study was
to provide evidence to answer the question: Can fair organizational
outcomes mitigate harmful responses to supervisor abuse, or does
doing so make matters worse? Answering this question is important because the dynamics of abusive supervision and resulting
employee deviance is a problem for organizations. Ex post organizational control (e.g., punishment) of line managers’ interpersonal treatment of employees can be costly because it involves
extensive monitoring (e.g., Dornbusch & Scott, 1975; Prendergast
& Topel, 1993). Instead of controlling line managers’ interpersonal behaviors, the self-gain view suggests that organizations can
lessen victims’ deviant reactions by providing them ex ante benefits (DJ) that compensate for the abuse (cf. Arrow, 1963; Barnard,
1938).
A third contribution of our research is that we tested our predictions programmatically (S. T. Fiske, 2004; Locke, 1996, 2007).
Study 1 tests the self-gain and self-regulation impairment views of
the abusive supervision– employee deviance relationship by examining the moderating effects of DJ in a cross-sectional sample of
working adults. Studies 2 and 3 constructively replicate (Lykken,
1968) these findings and extend them by examining the mediating
effects of variables capturing self-regulation impairment (Study 2:
ego depletion; Study 3: intrusive thoughts about the supervisor’s
behavior). Study 2 does so in a two-wave design with a sample of
working adults. Study 3 replicates Study 2 in a longitudinal,
multiwave, within-individual design to test whether the predicted
relationships are homologous across levels of analysis (e.g., Chen,
Bliese, & Mathieu, 2005; Dalal, Lam, Weiss, Welch, & Hulin,
2009).
Last, a review of the abusive supervision literature (Tepper,
2007) suggests that research examining processes by which abusive supervision influences outcomes is needed, as is research
exploring boundary conditions (like DJ). Therefore, the fourth
contribution of our study was examining a neglected mediator
(self-regulation impairment) and moderator (DJ). We first expound
on the theoretical foundation of the predictions for our studies.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
1011
The Self-Gain View Versus the Self-Regulation
Impairment View
A Comparative Test of Self-Gain Versus SelfRegulation Impairment Through DJ
Many different exchange theories explain behavior (for reviews,
see Cropanzano & Mitchell, 2005; Cropanzano & Rupp, 2008).
Seminal principles on which these applications are based share the
underlying assumption that people are rational actors, guided by
self-interest (e.g., Blau, 1964; Coleman, 1990; Ekeh, 1974; Emerson,
1976; Homans, 1961). Accordingly, people choose a particular
course of action because they gain something desirable from that
action and exchange with others to maximize their own benefits
(Coleman, 1990; Emerson, 1976; Homans, 1961; Kelley &
Thibaut, 1978).
The self-gain assumption is parsimonious, but like others before
us (Clark & Mills, 1979; Elster, 1989; A. P. Fiske, 1991), we claim
it is limited. Self-gain explanations are, at first glance, consistent
with the positive relationship between supervisor abuse and employee deviance. However, data and theoretical arguments suggest
that employees’ responding to supervisor abuse with deviance can
be self-defeating (Ferris, Brown, & Heller, 2009; Thau et al.,
2007). Retaliating with deviance violates normative standards and
may provoke others to punish the supervisor’s victim with further
harm (Aquino & Thau, 2009). For example, as an agent of the
organization’s interests, the supervisor may counterretaliate by
withdrawing valued outcomes (e.g., promotions, resources, helpful
feedback; Aquino et al., 2001; Tepper et al., 2009). If the risks of
counterretaliation are greater than the benefits of victims’ retaliation, then rational cost– benefit considerations do not fully account
for why victims of abuse respond with deviance.
The self-regulation impairment view offers a potential solution
to the supervisor abuse– employee deviance puzzle. Selfregulation theory (Baumeister, 1998) posits that the executive
function of the self is responsible for behavior. Self-regulatory
abilities (specifically self-resources) maintain impulses, active
choice making, persistence, and emotions that influence behavior
(consciously or unconsciously; Schmeichel & Baumeister, 2004).
Schmeichel and Baumeister (2004) argued that self-resources “inhibit, override, or alter responses that may arise as a result of
physiological processes, habit, learning, or the press of the situation” (p. 86). Some situations drain self-resources, which then
impair one’s ability to maintain normative behavior.
If we apply the self-regulation impairment view to explain the
supervisor abuse– employee deviance relationship, employees retaliate with deviance because supervisor abuse is psychologically
challenging, motivating the self to process, interpret, and understand the causes and consequences of the harm (Aquino & Thau,
2009; Thau et al., 2007). The experience and focus on the abuse
drain self-resources, which are needed to consider normative expectations and the risks of counterretaliation incurred by victims’
retaliation, making deviance more likely.
Yet, both self-gain and self-impairment explanations can account for the main effect of supervisor abuse on employee deviance. A first step toward a comparative test of the two explanations
is to identify a boundary condition that would lead to opposing
predictions for either of the two explanations. We argue that DJ
can provide a fair comparative test of the two views. We explain
our reasoning below.
Besides the socioemotional outcomes employees experience as
a result of their supervisors’ behaviors, the organization allocates
economic outcomes such as pay and rewards (e.g., Cropanzano &
Ambrose, 2001). Whether employees derive utility from these
outcomes depends to a large degree on whether they were fairly
allocated (e.g., McFarlin & Sweeney, 1992; Skitka, Winquist, &
Hutchinson, 2003), referred to as DJ (Homans, 1961). Equity
theory (Adams, 1965) defines DJ as employees’ assessment of
whether received rewards are matched proportionately to their
contributions to the organization (Greenberg, 1988). Justice scholars contend that DJ is the most instrumental form of justice
because it is rational assessments of gains and losses (Cropanzano,
Rupp, Mohler, & Schminke, 2001). Data support these arguments
and show that DJ is the best predictor of personal outcomes (for
meta-analytic evidence, see Cohen-Charash & Spector, 2001;
Colquitt, Conlon, Wesson, Porter, & Ng, 2001). Because DJ involves overall economic assessments of employees’ organizational
membership, we believe it can provide for a fair test of the
self-gain versus self-regulation impairment perspectives.
Self-gain explanations of the supervisor abuse– deviance relationship assume that the joint considerations of socioemotional and economic outcomes are central to determining employees’ reactions to
abusive supervision. Although abuse by a legitimized organizational
agent makes for a costly exchange, fair organizational outcomes (DJ)
are perceived as a beneficial exchange. Traditional exchange reasoning (Adams, 1965; Ekeh, 1974; Emerson, 1976; Gergen, 1980;
Homans, 1961; Walster et al., 1978) presumes that employees
consider a cost– benefit ratio derived from their organizational
membership and harm the organization when costs exceed benefits. Thus, the compensatory logic of cost– benefit considerations
implies that DJ should offset the costs of supervisor abuse, making
deviance less likely.
In contrast to the self-gain account of DJ, the self-regulation
impairment view predicts that DJ should strengthen (rather than
weaken) the abusive supervision– employee deviance relationship.
Accordingly, not only is abuse a costly aspect of employees’
organizational membership, but it also challenges the self to understand, process, and interpret the causes and consequences of
being harmed. The self-regulatory strength model (Baumeister,
Bratslavsky, Muraven, & Tice, 1998) suggests that harm drains
self-resources that are needed to maintain appropriate behavior
(e.g., Stucke & Baumeister, 2006). Fair rewards (DJ) are generally
perceived as a beneficial aspect of organizational membership; but
in combination with supervisor abuse, high DJ provides inconsistent information about employees’ utility to their organization. The
subsequent focus on that inconsistent information further drains
self-resources that maintain appropriate behavior.
Theory and data on social cognition support these ideas. Information within a knowledge domain (e.g., the utility I derive of my
membership with the organization) is organized into more specific
information (e.g., the costs I experience through the interpersonal
treatment of an organizational authority; the benefits I experience
through the rewards I get from the organization; for a review, see
S. T. Fiske & Taylor, 2008). People can hold consistent or inconsistent cognitions within the same knowledge domain. Dissonance
theory (Festinger, 1957) predicts that inconsistent cognitions cre-
1012
THAU AND MITCHELL
ate an aversive state, alerting attention to the inconsistency and
motivating attempts to resolve it (Aronson, 1968; Festinger, 1957;
Greenwald & Ronis, 1978).
We applied these principles to our study. Employees who are
harmed by a legitimized organizational agent who also receive fair
organizational rewards hold inconsistent cognitions about their
utility to their organization and should, therefore, experience dissonance. As a result, they will be motivated to pay attention to the
inconsistent information and deal with the dissonance. Selfregulatory strength research argues that both paying attention to
and dealing with dissonance drain self-regulatory resources
(Baumeister et al., 1998; Baumeister & Vohs, 2007; Richards &
Gross, 1999; J. L. Robinson & Demaree, 2007)—the same resources that are also used to understand, process, and interpret the
experience of abuse. The further draining of self-resources explains why DJ strengthens the supervisor abuse– deviance relationship.
Research provides direct evidence for the argument that dealing
with inconsistent information and dissonance impairs selfregulatory abilities, and when that occurs, behavior that requires
self-regulation is affected. For example, a study by Schmader and
Johns (2003) found that inconsistent information received about
individuals’ overall capabilities reduced their working memory
capacity and subsequent performance. Another study (J. L. Robinson & Demaree, 2007) investigated the influence of emotional
information (i.e., experiencing sadness) that is inconsistent with
people’s behavioral expressions (i.e., smiling) on their behavior
(i.e., performance). The results demonstrated that the dissonance
experience was cognitively taxing and worsened performance on
subsequent memory tasks that required self-regulation.
There are also theoretical arguments and data in the organizational literature supporting the notion that inconsistent information
about one’s organizational membership will strengthen the impact
of supervisor abuse on employee deviance. This research posits
that inconsistent value information about the work relationship
(abuse from the supervisor but fair rewards from the organization
in our study) communicates to employees that the cause for the
abuse lies within themselves rather than in the situation (i.e., the
abusive supervisor; Duffy, Ganster, Shaw, Johnson, & Pagon,
2006; Saul, Simon, & Colquitt, 2008). Accordingly, high levels of
DJ mean that the supervisor’s abuse is a personal attack and not a
reflection of an overall poor system of allocating unfair socioemotional and economic outcomes (Kelley, 1968). Researchers argue
that employees who experience attacks on the self have poor
self-regulatory abilities and are more likely to engage in deviant
work behaviors (Aquino & Douglas, 2003; Thau et al., 2007).
In sum, we present competing hypotheses (self-gain vs. selfregulation impairment) that explain the abusive supervision–
employee deviance relationship and test which view provides for a
better theory. If the self-gain view is the better explanation, then
DJ should weaken the relationship between abusive supervision
and employee deviance, as compensatory logic suggests that perceived gains (DJ) would offset the costs of abuse. Conversely, if
the self-regulation impairment view better explains deviant reactions to supervisor abuse, then DJ should strengthen the relationship between abusive supervision and employee deviance, as the
inconsistent information about employees’ organizational utility
would further drain self-resources needed to maintain appropriate
behavior. Therefore, we predict the following:
Hypothesis 1a: According to the self-gain view, the positive
relationship between abusive supervision and employee deviance will be weaker when DJ is high rather than low.
Hypothesis 1b: According to the self-regulation impairment
view, the positive relationship between abusive supervision
and employee deviance will be stronger when DJ is high
rather than low.
Study 1: An Initial Comparative Test of Self-Gain
Versus Self-Regulation Impairment Explanations for
Deviant Reactions to Supervisor Abuse
Method
Sample and procedure. We distributed surveys to individuals who were randomly selected and called for jury duty by a
county circuit court in the southeastern United States. We addressed the potential jurors at the beginning of the day, while they
waited to learn whether they would be required to serve. We
explained that the survey had nothing to do with the jury or court
system; rather, they were told that we wished to understand more
about sensitive issues that affect employees at work. Interested
participants picked up surveys and returned them to us on completion. Over the course of 4 weeks, 229 individuals agreed to
participate in the study (31.9% response rate). Complete data were
available for 216 cases. The average age of these participants was
43.60 years (SD ⫽ 11.73), and their tenure with the supervisor was
4.34 years (SD ⫽ 4.80); 57.41% were women.
Measures. Abusive supervision was assessed with Mitchell
and Ambrose’s (2007) five-item short version of Tepper’s (2000)
measure. Respondents rated perceptions of supervisor abuse over
the past year on a 7-point scale (1 ⫽ strongly disagree, 7 ⫽
strongly agree; e.g., My boss ridicules me).
DJ was assessed with Colquitt’s (2001) four-item measure (e.g.,
Do your outcomes reflect the effort you have put into your work?).
Respondents rated their agreement on a 5-point scale (1 ⫽ to a
small extent, 5 ⫽ to a great extent).
Employee deviance was assessed with measures of organizational
and supervisor-directed workplace deviance. We used Bennett and
Robinson’s (2000) 12-item organizational deviance measure (e.g.,
Taken property from work without permission). Because we also
assessed supervisor-directed deviance, we excluded one item that
clearly overlapped with this construct (Neglected to follow my
boss’s instructions). We assessed supervisor-directed deviance
with Mitchell and Ambrose’s (2007) 10-item measure (e.g., Acted
rudely toward my supervisor). For both measures, respondents
indicated the extent to which they intentionally engaged in the
respective behaviors over the past year on a 7-point scale (1 ⫽
never, 7 ⫽ daily).
Control variables were also considered. We controlled for employee age, tenure with the supervisor, and gender because these
demographic characteristics have been related to deviant behaviors
(Aquino & Douglas, 2003), victimization (Bowling & Beehr,
2006; Tepper, Duffy, Henle, & Lambert, 2006), and DJ (CohenCharash & Spector, 2001). We also controlled for procedural
justice (PJ; using Colquitt’s, 2001, seven-item measure) because
justice research suggests that PJ may account for the predicted
interaction (for a review, see Colquitt & Greenberg, 2003).
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
Results
Descriptive statistics. Table 1 shows variables’ means, standard deviations, zero-order correlations, and reliability coefficients. The measures displayed acceptable reliabilities. The results
show that abusive supervision was significantly positively related
to supervisor-directed deviance, but not to organizational deviance, and was significantly negatively related to DJ.
Hypothesis tests results. We used multiple regression analyses to test our hypotheses; the results are reported in Table 2.1 All
predictor variables were mean centered to minimize multicollinearity among them (Cohen, Cohen, West, & Aiken, 2003). We
first entered control variables. Next, the main effects of abusive
supervision and DJ were added, and finally, we added their interaction. The results show that over and above the effects of the main
effects and the control variables, there was a significant Abusive
Supervision ⫻ DJ interaction on supervisor-directed deviance
(b ⫽ .07, p ⬍ .01) but not on organizational deviance (b ⫽ .04, ns).
We probed the form of the interaction on supervisor-directed
deviance using procedures recommended by Cohen et al. (2003)
and calculated the simple slopes of supervisor-directed deviance
on abusive supervision one standard deviation above and one
standard deviation below the mean of DJ. The results support
Hypothesis 1b, showing that the abusive supervision–supervisordirected deviance relationship was more positive for high DJ (b ⫽
.48, p ⬍ .001) than low DJ (b ⫽ .31, p ⬍ .001; see Figure 1).2
Discussion
Study 1 provides support for the self-regulation impairment
view and shows that employees who experience supervisor abuse
are more likely to engage in deviance directed at their supervisor
when organizational outcomes are fair (Hypothesis 1b). These
results cannot be explained through the self-gain view. If employees had based their decision to engage in deviant behaviors on
cost– benefit considerations, then the costs associated with the
abuse should have been offset by the benefits of fair organizational
outcomes, making retaliation less likely. More specifically, the
relationship between abusive supervision and employee deviance
should have been weakened when DJ was high rather than low. In
contrast, our results show the opposite: The relationship was
strengthened when DJ was high rather than low.
This interaction pattern is consistent with the argument that the
abusive supervision experience impairs self-regulatory abilities
and influences employee deviance. According to the selfregulation impairment view, employee deviance occurs because
employees’ self-resources were depleted due to the victimization
experience (Aquino & Thau, 2009; Thau et al., 2007). Perceptions
of fair outcomes (DJ) provide a further drain on self-resources, as
people typically need these resources to pay attention to and deal
with the dissonant information.
Although there are strengths to Study 1’s design (e.g., its diverse, randomly selected sample of working adults), there are
limitations to consider. Unlike some previous abusive supervision–
employee deviance research, we did not make specific predictions
about the target of the deviance (supervisor vs. organization; e.g.,
Mitchell & Ambrose, 2007; Tepper et al., 2009; Thau et al., 2009)
because the abuse experience should affect victims’ self-regulatory
capabilities broadly. We had no a priori reason to believe the
1013
attention to norms is reduced only in the domain of actions toward
the supervisor versus the organization. Yet, the interaction was
significant only on supervisor-directed deviance. Exploring the
interaction on organizational deviance through simple slope analyses revealed a marginally significant positive slope for high DJ
and a nonsignificant slope for low DJ. It is possible that the failure
to find the interaction on organizational deviance is due to statistical power issues (Cohen et al., 2003). Interactions in field studies
are difficult to find (McClelland & Judd, 1993). Constructive
replication is needed to determine whether they hold up across
samples and methods (Lykken, 1968). Therefore, we conducted a
second study to see whether Study 1’s findings generalize across
samples.
Apart from constructive replication, we also wanted to address
other potential limitations of Study 1’s design with the second
study. First, Study 1’s variables were all measured at one point in
time, which limits causal inferences and can lead to percept–
percept inflations that challenge statistical conclusion validity
(Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Second, though
Study 1 supports the pattern of the interaction predicted by Hypothesis 1b, we do not know whether the assumed self-regulation
impairment accounts for the results found, as we did not assess
self-regulation impairment directly. Measurement of selfregulation impairment is needed to answer this question. Therefore, Study 2 examines the mediation effect of self-regulation
impairment with a measure of ego depletion on the abusive
supervision– employee deviance relationship and seeks to replicate
the moderating effects of DJ.
Study 2: Ego Depletion as a Form of Self-Regulation
Impairment and Mediator of the Relationship Between
Abusive Supervision and Employee Deviance
An assumption in the self-regulation impairment view is that abusive supervision directly drains self-resources needed for selfregulation, which is why victims respond with deviance. Thus, the
1
For all hypothesis tests presented in Studies 1, 2, and 3, the same set
of analyses was conducted without control variables as a supplemental test.
In all analyses for each of the studies, the results without the control
variables lead to the same substantive conclusions as those presented
within the results sections of Studies 1, 2, and 3.
2
With Study 1’s data, we also collected employees’ perceptions of
organizational support (POS; with Eisenberger, Cummings, Armeli, &
Lynch’s, 1997, eight-item measure). POS is typically considered an “exchange” variable (see Cropanzano & Mitchell, 2005). We reran the analyses using POS as a potential mediator; the results provide no support for
the mediation effect of POS. The Abuse ⫻ DJ interaction did not predict
POS (b ⫽ .07, 95% CI [⫺.06, .12], ns). Including POS to Model 3 of Study
1’s results did not substantially affect the interaction on supervisor-directed
deviance (b ⫽ .07, 95% CI [.02, .11], p ⬍ .01) or organizational deviance
(b ⫽ .04, 95% CI [⫺.01, .09], ns). POS predicted neither supervisordirected (b ⫽ ⫺.02, 95 % CI [⫺.31, .31], ns) nor organizational deviance
(b ⫽ ⫺.05, 95% CI [⫺.12, .02], ns) when the Abuse ⫻ DJ interaction was
controlled for. Finally, there were no significant indirect effects of abuse on
deviance via POS at different levels of DJ (bias-corrected 95% confidence
intervals for the indirect effects included zero). Therefore, we believe these
results suggest that social exchange relations, specifically POS, did not
present as a mediator of the abusive supervision– employee deviance
relationship.
THAU AND MITCHELL
1014
Table 1
Study 1: Summary Statistics and Zero-Order Correlations
Variable
1.
2.
3.
4.
5.
6.
7.
8.
Age
Tenure with supervisor
Gendera
Procedural justice
Abusive supervision
Distributive justice
Supervisor-directed deviance
Organizational deviance
M
SD
1
2
3
4
5
6
7
8
43.60
4.34
0.57
3.24
1.59
3.36
1.46
1.60
11.72
4.80
0.50
1.00
1.02
1.31
0.58
0.54
—
.37
⫺.05
.01
.05
⫺.03
.37
⫺.13
—
⫺.06
.14
⫺.00
.04
⫺.02
⫺.03
—
⫺.08
.04
⫺.11
⫺.09
⫺.18
.91
⫺.28
.59
⫺.13
⫺.05
.91
⫺.24
.57
.07
.94
⫺.11
.01
.80
.37
.79
Note. N ⫽ 216. Cronbach’s alpha are shown in italics along the diagonal. Correlations greater than 兩.14兩 are significant at p ⬍ .05, and those greater than
兩.18兩 are significant at p ⬍ .01 (two-tailed).
a
0 ⫽ male, 1 ⫽ female.
strength of self-resources is a critical indicator of self-regulation
impairment (for a review, see Baumeister & Heatherton, 1996).
Baumeister and colleagues (e.g., Baumeister et al., 1998; Bushman &
Baumeister, 1998; Schmeichel & Baumeister, 2004) argued that harm
to the self (i.e., abusive supervision) is particularly relevant because it
effectively depletes one’s “ego” (called ego depletion; Baumeister et
al., 1998; Bushman & Baumeister, 1998). Ego depletion is defined as
a “reduction in the self’s capacity or willingness to engage in volitional action (including controlling the environment, controlling the
self, making choices, and initiating action) caused by prior exercise of
volition” (Baumeister et al., 1998, p. 1253). In short, ego depletion
captures the availability, or rather expenditure, of self-regulatory
resources (Gailliot, Schmeichel, & Baumeister, 2006; Twenge,
Muraven, & Tice, 2004).
Table 2
Study 1: Regression for Effect of Abusive Supervision ⫻
Distributive Justice Interaction on Supervisor-Directed Deviance
and Organizational Deviance
Supervisor-directed
deviance
Variable
Age
Tenure with supervisor
Gendera
Procedural justice
⌬R2
⌬F
Abusive supervision
Distributive justice
⌬R2
⌬F
Abusive Supervision ⫻
Distributive Justice
⌬R2
⌬F
R2
Adjusted R2
F
b
95% CI
Organizational
deviance
b
ⴱ
⫺.01
.00
⫺.12
.04
[⫺.01, ⫺.02] ⫺.01
[⫺.01, .02]
.00
[⫺.25, .01]
⫺.19ⴱ
[⫺.04, .12]
⫺.01
.04
2.12
.39ⴱⴱⴱ [.32, .47]
.08
⫺.01
[⫺.07, .05]
.01
ⴱⴱⴱ
.31
50.89
.07ⴱⴱ
[.02, .11]
.02ⴱⴱ
7.87
.38ⴱⴱⴱ
.36
17.98
.04
95% CI
[⫺.01, .00]
[⫺.01, .02]
[⫺.33, ⫺.05]
[⫺.11, .08]
.05ⴱ
2.80
[⫺.01, .16]
[⫺.06, .08]
.01
0.63
[⫺.01, .10]
.01
2.40
.07ⴱ
.04
2.13
Note. N ⫽ 217. Effects are unstandardized regression coefficients. CI ⫽
confidence interval.
a
0 ⫽ male, 1 ⫽ female.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001, two-tailed.
On the basis of this reasoning, we argue that the abusive supervision experience is too taxing and will drain needed self-resources,
specifically ego depletion (e.g., Finkel & Campbell, 2001; Gailliot et
al., 2006; Sassenberg, Moskowitz, Jacoby, & Hansen, 2007; Twenge
et al., 2004). Moreover, and extending Study 1, we expect that
self-regulation impairment via ego depletion will be stronger when
victimized employees also perceive organizational outcomes to be fair
(DJ) because the self will use additional resources to process the
inconsistent information about the employee’s organizational utility.
Therefore, we predict the following:
Hypothesis 2: The positive relationship between abusive supervision and self-regulation impairment is stronger when DJ
is high rather than low.
Study 1 supports the self-regulation impairment view, showing
that DJ strengthened the effects of abusive supervision on employee deviance. The underlying reason for this effect is that
abusive supervision promotes self-regulation impairment, particularly at high levels of DJ. Self-regulation impairment (via ego
depletion) that follows from the combination of supervisor abuse
and DJ makes it less likely that people will consider the potential
costs of their actions, thereby making deviant reactions more
likely. Our theoretical arguments are consistent with a mediatedmoderation framework (Edwards & Lambert, 2007): The Abusive
Supervision ⫻ DJ interaction on employee deviance is mediated
through self-regulation impairment (ego depletion). More specifically, self-regulation impairment theory (Baumeister et al., 1998)
implies that supervisor abuse indirectly affects employee deviance
through self-regulation impairment (ego depletion) and the first
stage of the indirect effect (i.e., supervisor abuse on self-regulation
impairment) is moderated by DJ. Formally, we predict the following:
Hypothesis 3: The interactive effect between abusive supervision and DJ on deviant behaviors is mediated by selfregulation impairment.
Method
Sample and procedure. Participants were recruited with the
assistance of the StudyResponse project (http://studyresponse.syr
.edu), an online data collection service that allows researchers to
advertise their studies to adult participants. The service advertised
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
1
Supervisor−Directed Deviance
1.2
1.4
1.6
1.8
2
1015
−1
−.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Figure 1.
deviance.
Low (1 SD below mean)
Study 1: Interaction between abusive supervision and distributive justice on supervisor-directed
Ego depletion was assessed with a 25-item measure assessing
the availability of self-regulatory resources (Gailliot et al., 2006;
Twenge et al., 2004). Respondents indicated the extent to which
the information in each statement was true in general on a 7-point
scale (1 ⫽ not true, 7 ⫽ very true; e.g., I feel like my willpower
is gone, I feel drained).
Employee deviance was assessed with the same supervisordirected and organizational deviance measures used in Study 1. As
with abusive supervision in Study 2, respondents were asked to
rate their engaged behavior in general.
Control variables were also considered. The same set of control
variables (age, tenure with supervisor, gender, PJ) in Study 1 was
used in Study 2.
our study to 1,000 employed individuals who were willing participants at two points in time (4 weeks apart). In exchange for their
participation in our study, participants were entered into a lottery
drawing for the chance to win 30 $50 gift certificates to Amazon
.com.
Five hundred seventy-seven individuals responded to the first
survey (a 57.7% response rate); 452 individuals responded to the
second survey (a 78.3% response rate). Complete data for the
analyses were available for 375 (supervisor-directed deviance) and
371 (organizational deviance) participants, respectively. The average age of the participants was 41.80 years (SD ⫽ 11.30), and the
average tenure with the supervisor was 4.44 years (SD ⫽ 4.53);
68% were women. Abusive supervision and the control variables
were collected at Time 1; ego depletion, supervisor-directed deviance, and organizational deviance were collected at Time 2.
Measures. Abusive supervision and DJ were assessed with
the same measures as in Study 1; however, respondents were asked
in Study 2 to rate perceived abusive supervision in general.
Results
Descriptive statistics. Table 3 shows variables’ means, standard deviations, zero-order correlations, and reliability coeffi-
Table 3
Study 2: Summary Statistics and Zero-Order Correlations
Variable
1.
2.
3.
4.
5.
6.
7.
8.
9.
Age
Tenure with supervisor
Gendera
Procedural justice
Abusive supervision
Distributive justice
Ego depletion
Supervisor-directed deviance
Organizational deviance
M
SD
1
2
3
4
5
6
7
8
9
41.80
4.44
0.68
3.54
1.49
3.59
2.51
1.34
1.55
11.30
4.53
0.47
0.87
0.81
1.01
0.73
0.58
0.57
—
.22
⫺.11
.04
⫺.12
.09
⫺.25
⫺.15
⫺.25
—
.02
.17
⫺.01
.13
⫺.09
.06
⫺.01
—
⫺.04
⫺.11
⫺.11
.09
⫺.18
⫺.12
.92
⫺.34
.67
⫺.35
⫺.18
⫺.13
.95
⫺.23
.31
.54
.41
.94
⫺.31
⫺.13
⫺.13
.92
.26
.36
.93
.75
.89
Note. N ⫽ 365. Cronbach’s alpha are shown in italics along the diagonal. Correlations greater than 兩.11兩 are significant at p ⬍ .05, and those greater than
兩.17兩 are significant at p ⬍ .01 (two-tailed).
a
0 ⫽ male, 1 ⫽ female.
1016
THAU AND MITCHELL
cients. The measures displayed acceptable reliabilities. Abusive
supervision was significantly positively related to ego depletion,
supervisor-directed deviance, and organizational deviance and
negatively related to DJ. Ego depletion was significantly and
negatively related to DJ and positively related to supervisordirected and organizational deviance.
Hypothesis tests results. Hypotheses 1, 2, and 3 were tested
in the following multiple regression models. Model 1 tested
whether DJ moderated the relationship between abuse and
supervisor-directed and organizational deviance, as it did in Study
1 (Hypothesis 1b). Model 2 tested whether DJ moderated the
relationship between abuse and ego depletion (the mediator; Hypothesis 2). Model 3 was the same model as Model 1 but included
ego depletion in the regression equation (Hypothesis 3). We tested
the mediated-moderation prediction using Edwards and Lambert’s
(2007) approach. Accordingly, Models 2 and 3 jointly define a
mediated-moderation model, in which the first stage of the mediated (i.e., indirect) effect of abusive supervision on deviance is
moderated by DJ (Edwards & Lambert, 2007, Equations 5 and 6).
Mediated moderation occurs if (a) DJ moderates the relationship
between abusive supervision and deviance in Model 2 and (b) the
mediating (i.e., indirect) effect defined through Models 2 and 3
varies according to the level of DJ. The strength and the significance of the indirect effect were calculated with simple effects
analyses (Edwards & Lambert, 2007, Equation 19).
Indirect effects involve the calculation of product terms across
regression equations, which often have a nonnormal distribution.
Conventional significance tests for nonnormal distributions in ordinary least squares lead to a high Type I error rate (Shrout &
Bolger, 2002). We used the multiple regression module in MLwiN
(Version 2.02; Rashbash, Charlton, Browne, Healy, & Cameron,
2005) and bootstrapped 1,000 samples to obtained bias-corrected
parameter estimates and associated confidence intervals (CIs).3
The results are reported in Tables 4 (for supervisor-directed deviance) and 5 (for organizational deviance). The simple effects
analyses detailing indirect, direct, and total effects of abusive
supervision at high and low levels of DJ are reported in Tables 6
(for supervisor-directed deviance) and 7 (for organizational deviance).
Tables 4 and 5 also report supplemental analyses that tested the
fit of the hypothesized mediated-moderation model against the fit
of an alternative mediation model (Model 4). In Model 4, DJ
additionally moderated the effect of ego depletion (the mediator)
on deviance (this effect is referred to as second-stage moderation;
Model 4 is Equation 20 in Edwards & Lambert, 2007). To establish whether adding parameters to a particular regression model
significantly increased model fit, we calculated chi-square-based
likelihood ratio tests.
The results of Model 1 (see Tables 4 and 5) show a significant
Abusive Supervision ⫻ DJ interaction on supervisor-directed deviance (b ⫽ .07, p ⬍ .01) and organizational deviance (b ⫽ .06,
p ⬍ .05). We plotted the interactions and calculated the simple
slopes of the deviant behaviors on abusive supervision one standard deviation above and one standard deviation below the mean
of DJ. The results support Hypothesis 1b and show that the abusive
supervision–supervisor-directed deviance relationship was more
positive for high DJ (b ⫽ .44, p ⬍ .001) than low DJ (b ⫽ .30, p ⬍
.001), and the abusive supervision– organizational deviance rela-
tionship was more positive for high DJ (b ⫽ .33, p ⬍ .001) than
low DJ (b ⫽ .21, p ⬍ .001; see Figures 2 and 3).
Model 2 shows a significant Abusive Supervision ⫻ DJ interaction
on ego depletion for both supervisor-directed deviance (b ⫽ .11, p ⬍
.01) and organizational deviance (b ⫽ .11, p ⬍ .01). The simple
effects analysis of this first-stage moderation effect is provided in
Tables 6 (for supervisor-directed deviance) and 7 (for organizational
deviance). The abusive supervision– ego depletion relationship was
more positive for high DJ (b ⫽ .31, p ⬍ .001) than low DJ (b ⫽
.10, p ⬍ .05) on supervisor-directed deviance and was more
positive for high DJ (b ⫽ .32, p ⬍ .001) than low DJ (b ⫽ .10, ns)
on organizational deviance. These results support Hypothesis 2,
and the pattern of the interaction is shown in Figure 4 (the pattern
was nearly identical for both samples).
Model 3’s results show that the Abusive Supervision ⫻ DJ
interaction on supervisor-directed deviance dropped in significance (b ⫽ .06, p ⬍ .05) when the effects of ego depletion were
included in the regression model, but ego depletion had a significant and positive effect (b ⫽ .07, p ⬍ .05). For organizational
deviance, the Abusive Supervision ⫻ DJ interaction turned nonsignificant (b ⫽ .04, ns), but ego depletion had a significant and
positive effect (b ⫽ .18, p ⬍ .001).
The results of the simple effects analyses further support Hypothesis 3. The indirect effect of abusive supervision on
supervisor-directed deviance through ego depletion was stronger at
high DJ (P ⫽ .02, p ⬍ .01) than low DJ (P ⫽ .01, p ⬍ .05), and
the difference between the two effects was significant ( p ⬍ .05;
see Table 6). Similarly, the indirect effect of abuse on organizational deviance was stronger at high DJ (P ⫽ .06, p ⬍ .01) than
low DJ (P ⫽ .02, p ⬍ .05), and the difference between the two
effects was significant ( p ⬍ .01).
Overall, these results provide evidence for our mediation hypothesis, but they do not show whether a more complex, mediatedmoderation model provides a better explanation for our data (Edwards
& Lambert, 2007). Therefore, we next compared Model 3 with Model
4, where DJ also moderated the direct effect of ego depletion on
deviance. The results show that there were no significant interactions
between ego depletion and DJ on both supervisor-directed (b ⫽ .03,
ns) and organizational deviance (b ⫽ .01, ns). Moreover, adding this
second-stage moderation effect to Model 3 did not improve model fit.
These results further support Hypothesis 3, as the more complex
mediated-moderation model is not a better fit for our data than the one
we hypothesized.
Supplemental analyses. Finally, in supplemental analyses we
explored whether the indirect effect we found was suggestive of
partial or complete mediation (R. M. Baron & Kenny, 1986). The
simple effects analyses reported in Tables 6 and 7 show that the
direct effects of supervisor abuse on deviance was significant,
which is consistent with partial mediation; the effects of abuse on
deviance are partially transmitted through ego depletion (R. M.
Baron & Kenny, 1986; MacKinnon, Lockwood, Hoffman, West, &
Sheets, 2002). Additionally, we compared the fit of a complete
mediation model (in which all parameters involving abuse were
removed) with the fit of a partial mediation model (i.e., Model 3).
3
The direction and significance of the regression parameter estimates
for Studies 2 and 3 without the bootstrap procedure lead to the same
substantive conclusions.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
1017
Table 4
Study 2: Mediated-Moderation Effects of Abusive Supervision ⫻ Distributive Justice and Ego Depletion on Supervisor-Directed
Deviance
Supervisor-directed deviance
Supervisor-directed
deviance (Model 1)
Variable
b
Age
Tenure with supervisor
Gendera
Procedural justice
⌬R2
LR ␹2 test
Abusive supervision
Distributive justice
⌬R2
LR ␹2 test
Abusive Supervision ⫻ Distributive Justice
⌬R2
LR ␹2 test
Ego depletion
⌬R2
LR ␹2 test
Ego Depletion ⫻ Distributive Justice
⌬R2
LR ␹2 test
R2
⫺.01ⴱⴱ
.02ⴱ
⫺.15ⴱ
⫺.07
95% CI
[⫺.01, ⫺.00]
[.01, 03]
[⫺.24, ⫺.05]
[⫺.14, .00]
.12
70.31ⴱⴱⴱ
.38ⴱⴱⴱ [.32, .43]
.03
[⫺.03, .09]
.23
123.69ⴱⴱⴱ
.07ⴱⴱ
[.02, .12]
.01
11.01ⴱⴱ
.36
Ego depletion (Model 2)
b
95% CI
⫺.01ⴱⴱⴱ
.00
.11
⫺.16ⴱⴱⴱ
[⫺.02, ⫺.01]
[⫺.01, .01]
[⫺.03, .23]
[⫺.25, ⫺.07]
.18
89.56ⴱⴱⴱ
.21ⴱⴱⴱ [.13, .29]
⫺.09ⴱ
[⫺.17, ⫺.01]
.05
23.17ⴱⴱⴱ
.11ⴱⴱ
[.04, .18]
.02
10.11ⴱⴱ
.24
Model 3
b
95% CI
⫺.01ⴱⴱ
.01ⴱ
⫺.16ⴱⴱ
⫺.06ⴱ
[⫺.01, ⫺.00]
[.00, .03]
[⫺.25, ⫺.06]
[⫺.13, ⫺.01]
.12
70.31ⴱⴱⴱ
.36ⴱⴱⴱ [.30, .41]
.03
[⫺.03, .09]
.23
123.69ⴱⴱⴱ
.06ⴱ
[.01, .11]
.01
11.01ⴱⴱ
.07ⴱ
[.01, .14]
.01
5.26ⴱ
.37
Model 4
b
95% CI
⫺.01ⴱⴱ
.01ⴱ
⫺.16ⴱⴱ
⫺.06
[⫺.01, ⫺.00]
[.00, .03]
[⫺.25, ⫺.07]
[⫺.13, .02]
.12
70.31ⴱⴱⴱ
.36ⴱⴱⴱ [.30, .41]
.03
[⫺.03, .09]
.23
123.69ⴱⴱⴱ
.06ⴱ
[.01, .11]
.01
11.01ⴱⴱ
.07ⴱ
[.01, .14]
.01
5.26ⴱ
.03
[⫺.03, .09]
.00
1.01
.37
Note. N ⫽ 375. Effects are unstandardized regression coefficients. Likelihood ratio (LR) ␹2 test indicates model fit increase via log-likelihood difference
test. CI ⫽ bias-corrected confidence interval.
a
0 ⫽ male, 1 ⫽ female.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001, two-tailed.
The comparison test showed that for both supervisor-directed
deviance, ␹2(2) ⫽ 126.16, p ⬍ .001, and organizational deviance,
␹2(2) ⫽ 56.60, p ⬍ .001, the less parsimonious, partial mediation
model fit the data better than the complete mediation model.
Consequently, the results in Study 2 can be described as evidence
for partial mediation.
Discussion
Study 2 provides more direct evidence of the self-regulation
impairment view of deviant responses to supervisor abuse. We
replicated the Abusive Supervision ⫻ DJ interaction on
supervisor-directed deviance found in Study 1, and the interaction
effect on organizational deviance was significant. These findings
provide additional empirical support for Hypothesis 1b in a different sample of working adults, from diverse industries and jobs.
These results suggest that victimized employees do not discriminate across targets of deviance when they received high rather than
low levels of fair outcomes. The results also provide support for
Hypothesis 2, as supervisor abuse was more strongly related to ego
depletion when DJ was high rather than low. Last, the results
support Hypothesis 3 because DJ moderated the first stage of the
indirect effect of abusive supervision on employee deviance.
Study 2 has notable strengths. It shows that the results found in
Study 1 generalize across both samples and methods. Replicating
the interaction pattern predicted by the self-regulation impairment
view speaks to the robustness of the phenomenon. We also directly
assessed self-regulation impairment, and our results show that it
mediated the effects of abusive supervision on employee deviance.
Overall, Study 2’s results provide a full test of the predictions
made by the self-regulation impairment perspective about employee reactions to abuse.
Despite these strengths, Study 2 is not without limitations. Scholars
contend the focus of self-regulation impairment is on psychological
processes within individuals (e.g., Baumeister, 1998; Beal, Weiss,
Barros, & MacDermid, 2005; Wegner, 1994). Although the betweenindividuals results of Studies 1 and 2 are consistent with selfregulation impairment, whether the findings generalize to withinindividual relationships is unknown (e.g., Chen et al., 2005; Dalal et
al., 2009). To make this inference, abuse, self-regulation impairment,
and deviance should be assessed in a longitudinal design (Singer &
Willett, 2003). Also, past research suggests that controlling for employees’ trait hostility is important when examining responses to
victimization because people with this trait are more likely to be
victimized (for a review, see Aquino & Thau, 2009), engage in
deviance (Judge, Scott, & Ilies, 2006), and feel more injustices
(Aquino, Lewis, & Bradfield, 1999), and so we controlled for trait
hostility in Study 3. Finally, Study 2 measured self-regulation impairment through an ego depletion measure, which past research (e.g.,
Finkel & Campbell, 2001; Gailliot et al., 2006; Sassenberg et al.,
2007; Twenge et al., 2004) suggests adequately captures relentless
attempts to understand, process, and interpret the causes and consequences of the harm. It does not measure actual cognitive activity.
Some theorists contend that a more direct measure is an assessment of
psychological phenomena that occur when resources are drained—
THAU AND MITCHELL
1018
Table 5
Study 2: Mediated-Moderation Effects of Abusive Supervision ⫻ Distributive Justice and Ego Depletion on Organizational Deviance
Organizational deviance
Organizational deviance
(Model 1)
Variable
b
Age
Tenure with supervisor
Gendera
Procedural justice
⌬R2
LR ␹2 test
Abusive supervision
Distributive justice
⌬R2
LR ␹2 test
Abusive Supervision ⫻ Distributive
Justice
⌬R2
LR ␹2 test
Ego depletion
⌬R2
LR ␹2 test
Ego Depletion ⫻ Distributive Justice
⌬R2
LR ␹2 test
R2
Ego depletion (Model 2)
95% CI
b
ⴱⴱⴱ
⫺.01
.01
⫺.12ⴱ
.01
[⫺.02, ⫺.01]
[.00, .02]
[⫺.22, ⫺.03]
[⫺.06, .09]
.10
60.92ⴱⴱⴱ
.27ⴱⴱⴱ
[.22, .32]
⫺.03
[⫺.10, ⫺.02]
.13
67.86ⴱⴱⴱ
.06ⴱ
95% CI
ⴱⴱⴱ
⫺.01
.00
.12
⫺.17ⴱ
[⫺.02, ⫺.01]
[⫺.01, .01]
[⫺.00, .25]
[⫺.27, ⫺.07]
.17
88.02ⴱⴱⴱ
.21ⴱⴱⴱ
[.13, .29]
⫺.08
[⫺.17, .01]
.05
24.18ⴱⴱⴱ
.11ⴱ
[.01, .11]
.01
6.58ⴱ
.24
[.04, .18]
.02
10.94ⴱⴱ
Model 3
b
Model 4
95% CI
ⴱⴱⴱ
[⫺.01, ⫺.01]
[⫺.00, .02]
[⫺.23, ⫺.05]
[⫺.03, .11]
.10
60.92ⴱⴱⴱ
.23ⴱⴱⴱ
[.17, .29]
⫺.02
[⫺.08, .04]
.13
67.86ⴱⴱⴱ
b
95% CI
ⴱⴱⴱ
⫺.01
.01
⫺.14ⴱⴱ
.04
⫺.01
.01
⫺.15ⴱⴱ
.04
[⫺.01, ⫺.01]
[⫺.00, .02]
[⫺.23, ⫺.05]
[⫺.03, .12]
.10
60.92ⴱⴱⴱ
.23ⴱⴱⴱ
[.17, .28]
⫺.02
[⫺.08, .04]
.13
67.86ⴱⴱⴱ
.04
.04
[⫺.01, .09]
.01
6.58ⴱ
.18ⴱⴱⴱ
[.11, .25]
.05
27.51ⴱⴱⴱ
.24
[⫺.01, .08]
.01
6.58ⴱ
.18ⴱⴱⴱ
[.11, .24]
.05
27.51ⴱⴱⴱ
.01
[⫺.03, .06]
.00
⫺0.93
.29
.29
Note. N ⫽ 371. Effects are unstandardized regression coefficients. Likelihood ratio (LR) ␹2 test indicates model fit increase via log-likelihood difference
test. CI ⫽ bias-corrected confidence interval.
a
0 ⫽ male, 1 ⫽ female.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001, two-tailed.
those reflective of a cognitively taxed or exhausted self (Baumeister et
al., 1998; Fischer, Greitemeyer, & Frey, 2007, 2008; Tice, Baumeister, Shmueli, & Muraven, 2007). We adopted these ideas in Study 3.
Study 3: Intrusive Thoughts as Mediator of the
Relationship Between Abusive Supervision and
Employee Deviance
We designed Study 3 to further test the robustness of our
findings, while addressing the limitations identified above. Studies
2 and 3 support the self-regulation impairment view about the
supervisor abuse– employee deviance relationship between individuals: Employees who experience high levels of abusive super-
vision tend to experience self-regulation impairment (particularly
at high levels of DJ), and in turn, this influences the likelihood of
deviant behavior. Although the between-individuals results support the self-regulation impairment explanation, they leave open
questions about the explanatory power of a particular theory (Chen
et al., 2005).
Researchers have noted that the validity, parsimony, and
breadth of a theory can be evaluated by showing that the
theory’s predictions are generalizable not only across samples
but also across levels of analysis (Chen et al., 2005; Dalal et al.,
2009; Klein, Dansereau, & Hall, 1994). On the basis of the data
we have presented so far, we do not know whether on occasions
in which a given employee experiences abusive supervision he
Table 6
Study 2: Simple Effects Analyses for Supervisor-Directed Deviance
First-stage moderation
Variable
DJ high
DJ low
Difference
PMX
ⴱⴱⴱ
.31
.10ⴱ
.21ⴱⴱ
95% CI
[.20, .43]
[.01, .21]
[.07, .36]
Direct effect
PYX
ⴱⴱ
.42
.29ⴱⴱ
.13ⴱ
95% CI
[.34, .50]
[.22, .36]
[.03, .23]
Indirect effect
PYMPMX
ⴱⴱ
.02
.01ⴱ
.02ⴱ
95% CI
[.00, .05]
[.00, .02]
[.00, .04]
Total effect
(PYX ⫹ PYMPMX)
ⴱⴱ
.44
.30ⴱⴱ
.14ⴱⴱ
95% CI
[.36, .52]
[.23, .37]
[.04, .25]
Note. PMX ⫽ path from abusive supervision to ego depletion; PYX ⫽ path from abusive supervision to supervisor-directed deviance; PYM ⫽ path from
ego depletion to supervisor-directed deviance; CI ⫽ bias-corrected confidence interval; DJ ⫽ distributive justice.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
1019
Table 7
Study 2: Simple Effects Analyses for Organizational Deviance
First-stage moderation
Direct effect
Indirect effect
Total effect
Variable
PMX
95% CI
PYX
95% CI
PYMPMX
95% CI
(PYX ⫹ PYMPMX)
95% CI
DJ high
DJ low
Difference
.32ⴱⴱⴱ
.10
.22ⴱⴱ
[.20, .44]
[⫺.02, .19]
[.08, .37]
.27ⴱⴱ
.19ⴱⴱ
.08
[.18, .35]
[.12, .26]
[⫺.03, .18]
.06ⴱⴱ
.02ⴱ
.04ⴱⴱ
[.03, .10]
[.00, .04]
[.02, .08]
.33ⴱⴱ
.21ⴱⴱ
.12ⴱ
[.24, .41]
[.14, .28]
[.01, .22]
Note. PMX ⫽ path from abusive supervision to ego depletion; PYX ⫽ path from abusive supervision to organizational deviance; PYM ⫽ path from ego
depletion to organizational deviance; CI ⫽ bias-corrected confidence interval; DJ ⫽ distributive justice.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
(McCullough, Bellah, Kilpatrick, & Johnson, 2001). Most people experience intrusions on an unconscious level, but when
self-regulation is impaired, intrusions register more frequently
(Matthews, Schwean, Campbell, Saklofske, & Mohamed,
2000). Indeed, research shows that people who try to suppress
unwanted thoughts show subsequently more self-regulation impairment (Muraven, Tice, & Baumeister, 1998; J. L. Robinson
& Demaree, 2007). In short, intrusive thoughts are an indicator
of self-regulation impairment. The more intrusive thoughts
experienced, the fewer self-resources will be available for
abused employees to consider the consequences of their behavior, making deviance more likely.
Thus, Study 3’s design uses a multimeasure approach. In addition to the different measure of self-regulation impairment, we
assessed deviance with a general measure of antisocial behaviors
(S. L. Robinson & O’Leary-Kelly, 1998) and DJ with Niehoff and
Moorman’s (1993) measure. Doing so allowed us to test whether
the results found in Studies 1 and 2 are measure specific. Moreover, we assessed employees’ trait hostility as a betweenparticipants control variable in Study 3 to further strengthen the
statistical conclusion validity of our findings.
1.2
Supervisor-Directed Deviance
1.4
1.6
1.8
or she will experience self-regulation impairment (particularly
at high levels of DJ) and whether this will influence his or her
deviant behavior. Study 3 addresses this concern with a multiwave, longitudinal design. Over multiple measurement occasions, employees reported their perceptions of supervisor abuse
and their engaged deviance as well as their experienced selfregulation impairment (the mediator). Finally, we measured DJ
as a between-individuals variable to test whether DJ moderates
the within-individual relationships between abuse and selfregulation impairment.
Study 3 extends the findings of Study 2 by assessing victims’
cognitively taxed or exhausted self (e.g., Baumeister et al.,
1998; Fischer et al., 2007, 2008; Tice et al., 2007) as a measure
of self-regulation impairment. Self-regulation impairment is
assessed with a measure of intrusive thoughts about the supervisor’s behavior. A high degree of intrusions reflects drained
self-resources (Baumeister & Heatherton, 1996; Oaten, Williams, Jones, & Zadro, 2008; Wegner, 1994) because intrusions
are consistent, unwanted thoughts and imagery (Horowitz, Wilner, & Alvarez, 1979), making it hard for people to turn their
attention away from the victimization experience and move on
-1
-.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Figure 2.
deviance.
Low (1 SD below mean)
Study 2: Interaction between abusive supervision and distributive justice on supervisor-directed
THAU AND MITCHELL
1.4
Organizational Deviance
1.6
1.7
1.5
1.8
1.9
1020
-1
-.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Low (1 SD below mean)
Figure 3. Study 2: Interaction between abusive supervision and distributive justice on organizational deviance.
Method
2.1
2.2
Ego-Depletion
2.3
2.4
2.5
2.6
Sample and procedure. We recruited 60 part- and full-time
working adults through a behavioral research laboratory web site.
We advertised the study as being about “work experiences.” Once
prospective participants signed up, they received detailed instructions on the study procedure. They were told that there was no
deception involved in the study and that participation was voluntary. At this point, participants filled in a web-based survey,
providing us with their demographics and reporting the hostile
emotions they experienced during the last year. We then sent
e-mail to all participants over a period of 5 weeks on Mondays and
Thursdays to remind them to fill in the web-based survey. These
10 measures assessed their perceptions of supervisor abuse, intrusive thoughts about their supervisor, and antisocial behaviors. We
assessed DJ in a final survey to constructively replicate the
Abuse ⫻ DJ interaction found in Studies 1 and 2. The supervisor
is often the one who communicates news about rewards and
compensation; so abuse may color DJ perceptions (Tepper, 2000).
This possible link between abuse and DJ judgments should make
it less likely to find moderation, providing a conservative test of
our interaction hypothesis (Cohen et al., 2003). At the end of the
-1
-.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Low (1 SD below mean)
Figure 4. Study 2: Interaction between abusive supervision and distributive justice on ego depletion (selfregulation impairment).
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
study, we debriefed participants about the purpose of our study and
what we suspected to find with it. Participants received $45 on
completion of the study.
Participants’ average age was 25.66 years (SD ⫽ 7.84); average
tenure with the supervisor was 11.32 months (SD ⫽ 15.96); 68%
of the sample were women. We used observations in the analysis
sample only when participants indicated that they had worked in
the last 2 days so that employees who did not work as frequently
as others would not report the same experience twice (the results
reported subsequently did not change when including observations
from participants who had worked in the last 3 days). The analysis
sample included 352 total observations nested within 50 participants. The maximum number of observations (per participant) was
10 (10 participants), the minimum number was one (one participant), and the average was seven (eight participants with two to
four observations and 31 with five to nine).
Measures. Abusive supervision was measured with the same
items in Studies 1 and 2. However, we asked respondents to
consider their supervisors’ behaviors during their past few workdays.
DJ was assessed with Niehoff and Moorman’s (1993) measure
(e.g., I think that my level of pay is fair), set on a 7-point scale (1 ⫽
strongly disagree, 7 ⫽ strongly agree).
Self-regulation impairment was measured with the Impact of
Events Scale (Horowitz et al., 1979), adapted to specific events
(e.g., McCullough et al., 2001) and here to the supervisor’s behaviors (e.g., I thought about my supervisor’s behavior when I
didn’t mean to; I had trouble falling asleep or staying asleep
because of pictures or thoughts about my supervisor’s behavior
that came to my mind). Respondents indicated how often they
experienced these thoughts in their past few workdays on a 4-point
scale (1 ⫽ not at all, 4 ⫽ often).
Employee deviance was assessed with a seven-item antisocial
behaviors measure (S. L. Robinson & O’Leary-Kelly, 1998). Respondents indicated the frequency with which they engaged in the
behaviors listed (e.g., Griped with coworkers; Deliberately bent or
broke a rule) on a 7-point scale (1 ⫽ never, 7 ⫽ very often).
As in Studies 1 and 2, we controlled for age, tenure with
supervisor, and gender. We did not assess PJ but instead controlled
for employees’ trait hostility (six items: angry, irritable, hostile,
scornful, disgusted, and loathing). Respondents indicated the extent they felt these emotions during the past year on a 7-point scale
(1 ⫽ not at all, 7 ⫽ extremely). Following common practice in
longitudinal data analyses, we controlled for measurement occasion because across participants there may be, on average, a time
trend in the dependent variables, but the slopes for time may differ
per participant (Singer & Willett, 2003; Snijders, 1996).
Results
Descriptive statistics. Table 8 shows variables’ means, standard deviations, between-individuals zero-order correlations, and
reliability coefficients. The measures displayed acceptable reliabilities. The results also show that between individuals, the average level of supervisor abuse was significantly and positively
related to the average level of intrusions, and the average level of
intrusions was significantly and positively related to employee
deviance. The between-individuals correlations among the other
study variables failed to reach statistical significance. Within in-
1021
dividuals, supervisor abuse was significantly and positively related
to intrusions and deviance. Intrusions were significantly and positively related to employee deviance. DJ was significantly and
negatively related to intrusions.
Variance partitioning and analysis. We used multilevel
analysis to analyze our data because Study 3’s purpose was to
analyze within-individuals relationships between abuse and deviance over time. Multilevel analysis is appropriate because the data
structure in Study 3 was nested (i.e., repeated measurements nested
within individuals; Snijders, 1996). To further evaluate whether
multilevel analysis is appropriate, we estimated so-called null
models to examine the amount of within-individual (␴2) and
between-individuals (␶20) variance in the dependent variables. The
ratio of between individuals to total variance defines the intraclass
correlation coefficient (ICC; Snijders & Bosker, 1999). For employee deviance, we found a significant amount of both withinindividual (␴2 ⫽ .11, 95% CI [.10, .13]) and between-individuals
(␶20 ⫽ .20, 95% CI [.13, .29]) variance (ICC ⫽ .65). The withinindividual (␴2 ⫽ .11, 95% CI [.9, .12]) and between-individuals
(␶20 ⫽ .07, 95% CI [.05, .11]) variance in intrusions was also
significant (ICC ⫽ .39). The positive ICC values indicate statistical dependence in the data, which lead to bias in ordinary-leastsquares-type analyses and suggest that multilevel analysis is appropriate. Multilevel analysis considers statistical dependence in
the estimation of parameters and standard errors (Snijders &
Bosker, 1999). Another reason why multilevel analysis is appropriate is that compared with other types of longitudinal data
analyses (e.g., repeated measures analysis of variance, structural
equation modeling), multilevel analysis allows participants to have
missing observation data (Hox, 2002; Snijders, 1996).
We used MLwiN (Rashbash et al., 2005) to estimate bootstrapped, multilevel, mixed-effects regression models with bias
correction. Measurement occasions were defined as lower level
units and participants as higher level units in the multilevel models. As in Study 2, we bootstrapped 1,000 samples. We chose the
so-called standard error specification for a longitudinal multilevel
model (Singer & Willett, 2003). In this specification, the composite error term is homoscedastic across people and both heteroscedastic and correlated over time within people. This specification is
parsimonious and adaptive to many empirical situations (Singer &
Willett, 2003).4 We used within-person centering for the abuse and
intrusions variables because our conceptual interest was to explain
the relationships among these variables and employee deviance
within individuals. This centering technique can lead to underspecified hierarchical models, and so the average level of abuse
and intrusions was controlled for when entering the within-person
centered variables into the multilevel regression models (Kreft, de
Leeuw, & Aiken, 1995; Singer & Willett, 2003). All other measured variables were grand mean centered; accordingly, gender
and measurement occasion were not centered. In keeping with
Singer and Willett’s (2003) recommendations to test for crosslevel interaction effects in longitudinal multilevel models, in all
models estimated, variables that varied within individuals (abusive
supervision, intrusions, measurement occasions) were included as
fixed effects with random slopes, and variables that did not vary
4
We explored alternative error structure specifications, which did not
provide a better model fit.
THAU AND MITCHELL
1022
Table 8
Study 3: Summary Statistics and Zero-Order Correlations
Variable
1.
2.
3.
4.
5.
6.
7.
8.
Age
Tenure with supervisor
Gendera
Hostility
Abusive supervision
Distributive justice
Intrusions
Deviant behaviors
M
SD
1
2
25.66
11.32
0.68
3.03
1.38
4.85
1.27
1.35
7.84
15.96
0.47
1.39
0.64
1.47
0.30
0.50
—
.58ⴱⴱ
⫺.12
.21
⫺.05
⫺.26
⫺.04
.12
—
.13
⫺.18
⫺.18
.12
⫺.14
.01
3
4
—
⫺.36ⴱⴱ
⫺.10
.21
⫺.10
.15
.88
.34
⫺.57ⴱⴱ
.44ⴱⴱ
.32ⴱ
5
.95
⫺.48ⴱⴱ
.35ⴱ
.19
6
7
8
⫺.41ⴱⴱⴱ
.93
⫺.19
⫺.07
⫺.03
⫺.11
⫺.07
.29ⴱⴱ
.46ⴱⴱⴱ
⫺.12
.88
.46
.10
.01
.13
.26ⴱ
.26ⴱⴱⴱ
⫺.06
.39ⴱⴱⴱ
.80
Note. Correlations under the diagonal are between-individuals correlations (N ⫽ 50). Abusive supervision, intrusions, and deviant behaviors are averaged
to calculate between-individuals correlations. Correlations above the diagonal are within-individual correlations (N ⫽ 252). Within-individual correlations
were calculated based on standardizing regression coefficients in a multilevel model between one predictor and one criterion. Cronbach’s alpha are shown
in italics along the diagonal.
a
0 ⫽ male, 1 ⫽ female.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
within persons (age, gender, tenure, hostility, DJ, average level of
intrusions, average level of abusive supervision) were included as
fixed effects only. Random slopes were estimated jointly with the
respective variable’s fixed effects. To establish whether adding
parameters to a particular model increased model fit, we estimated
chi-square-based likelihood ratio tests. The results are reported in
Table 9.
Hypothesis tests results. We investigated the same three
models as in Study 2. In Model 1, we tested whether DJ moderated
the within-person relationship between supervisor abuse and employee deviance similar to the between-persons results found in
Studies 1 and 2 (Hypothesis 1b). Model 2 tested whether DJ
moderated the within-person relationship between supervisor
abuse and intrusions (the mediator). Model 3 was the same model
Table 9
Study 3: Mediated-Moderation Effects of Abusive Supervision ⫻ Distributive Justice and Intrusive Thoughts on Employee Deviance
Employee deviance
Employee deviance
(Model 1)
Variable
b
Age
Tenure with supervisor
Gendera
Hostility
Measurement occasion
⌬R2
LR ␹2 test
Average of abusive supervision
Abusive supervision
Distributive justice
⌬R2
LR ␹2 test
Abusive Supervision ⫻ Distributive Justice
⌬R2
LR ␹2 test
Average of intrusive thoughts
Intrusive thoughts
⌬R2
LR ␹2 test
Intrusive Thoughts ⫻ Distributive Justice
⌬R2
LR ␹2 test
R2
.01
.00
.14
.12
⫺.02ⴱⴱ
95% CI
[⫺.01, .03]
[⫺.01, .01]
[⫺.11, .37]
[.02, .22]
[⫺.03, ⫺.01]
.05
20.10ⴱ
.14
[⫺.07, .32]
.32ⴱⴱⴱ [.22, .43]
.05
[⫺.06, .16]
.11
39.72ⴱⴱⴱ
ⴱⴱ
.08
[.02, .14]
.01
16.68ⴱⴱⴱ
.17
Intrusive thoughts
(Model 2)
b
95% CI
⫺.01
.00
.08
.12ⴱⴱ
⫺.01
[⫺.02, .01]
[⫺.00, .00]
[⫺.11, .27]
[.04, .20]
[⫺.03, .00]
.04
20.60ⴱ
.20ⴱ
[.05, .36]
.44ⴱⴱⴱ [.22, .64]
.06
[⫺.01, .13]
.23
37.59ⴱⴱⴱ
ⴱ
.14
[.01, .26]
.01
3.93ⴱ
.28
Model 3
b
95% CI
.02
⫺.00
.27ⴱ
.04
⫺.01ⴱ
[⫺.01, .04]
[⫺.02, .01]
[.01, .52]
[⫺.07, .15]
[⫺.02, ⫺.00]
.05
20.10ⴱ
.03
[⫺.23, .27]
.20
[⫺.01, .40]
.04
[⫺.07, .16]
.11
39.72ⴱⴱⴱ
.06
[⫺.05, .16]
.01
16.68ⴱⴱⴱ
.82ⴱⴱⴱ [0.34, 1.29]
.50ⴱⴱ
[.18, .81]
.56
178.76ⴱⴱⴱ
.73
Model 4
b
95% CI
.02
⫺.00
.28ⴱ
.04
⫺.01ⴱ
[⫺.01, .04]
[⫺.01, .01]
[.01, .56]
[⫺.08, .16]
[⫺.02, ⫺.01]
.05
20.10ⴱ
.03
[⫺.25, .33]
.20
[⫺.00, .41]
.04
[⫺.08, .16]
.11
39.72ⴱⴱⴱ
.08
[⫺.05, .20]
.01
16.68ⴱⴱⴱ
ⴱⴱⴱ
.82
[0.35, 1.30]
.50ⴱⴱ
[.15, .83
.56
178.76ⴱⴱⴱ
⫺.06
[⫺.28, .16]
.00
⫺0.44
.73
Note. N ⫽ 352, k ⫽ 50. Mixed-effects multilevel regression model with persons nested within measurement occasions. Likelihood ratio (LR) ␹2 test
indicates model fit increase via log-likelihood difference test. Within each model, the model fit of the control variables was compared with the fit of a null
(intercept only) model. Random part of the model is not displayed in the table (available from the authors). CI ⫽ bias-corrected confidence interval.
a
0 ⫽ male, 1 ⫽ female.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
moderation model is not a better fit for our data than the one we
hypothesized.
Supplemental analyses. Finally, in supplemental analyses we
explored whether the indirect effect we found was suggestive of
partial or complete mediation (R. M. Baron & Kenny, 1986). The
simple effects analysis reported in Table 10 shows that the direct
effects of supervisor abuse on deviance was not significant, which
is consistent with complete mediation; the effects of abuse on
deviance are fully transmitted through intrusive thoughts (R. M.
Baron & Kenny, 1986; MacKinnon et al., 2002). Additionally, we
compared the fit of a complete mediation model (in which all fixed
and random parameters involving abuse were removed) with the fit
of a partial mediation model (i.e., Model 3). The comparison test
showed that the less parsimonious, partial mediation model did not
fit the data better than the more parsimonious, complete mediation
model, ␹2(7) ⫽ 7.49, ns. Consequently, the results in Study 3 can
be interpreted as evidence for complete mediation.
Discussion
Study 3’s results fully support Hypotheses 1b, 2, and 3. Across
various measurement occasions, abusive supervision was more
strongly related to employee deviance when DJ was high rather
than low. Similarly, abusive supervision was more strongly related
to intrusive thoughts about the supervisor’s behavior when DJ was
high rather than low. DJ moderated the first stage of the indirect
effect of abusive supervision on employee deviance through intrusive thoughts, supporting mediated moderation. These results provide consistent evidence of the self-regulation impairment explanation of why abused employees engage in harmful behaviors.
We believe Study 3’s findings provide more evidence to support
the self-regulation impairment view, using a longitudinal, multimeasure design. The frequency of intrusive thoughts about the
supervisor’s behavior captures self-regulatory cognitive activity
.8
Deviant Employee Behaviors
1.2
1.4
1
1.6
as Model 1, but we included intrusions in the multilevel regression
equation. We then compared the fit of Model 3 with an alternative
model (Model 4), in which DJ also moderated the effect of
intrusions (the mediator) on deviance.
The first model results show a significant cross-level Abuse ⫻
DJ interaction on employee deviance (␥ ⫽ .08, p ⬍ .01; see Table
9). Simple slopes of employee deviance on abusive supervision
were calculated one standard deviation above and below the mean
of DJ. The results support Hypothesis 1b and show that the abusive
supervision– employee deviance relationship was more positive for
high DJ (b ⫽ .44, p ⬍ .001) than low DJ (b ⫽ .20, p ⬍ .01; see
Figure 5).
There was a significant Abuse ⫻ DJ interaction on intrusions
(␥ ⫽ .14, p ⬍ .05; see Model 2). The simple effects analyses (see
Table 10) support Hypothesis 2 and show that the first-stage
moderation effect of DJ on the supervisor abuse– employee deviance relationship was stronger for high DJ (b ⫽ .64, p ⬍ .001) than
low DJ (b ⫽ .24, p ⬍ .01; see Figure 6).
Model 3’s results show that the Abuse ⫻ DJ interaction was no
longer significant (␥ ⫽ .06, ns) when intrusions were included in the
multilevel regression model but show that intrusions significantly
predicted employee deviance (␥ ⫽ .50, p ⬍ .01). Models 2 and 3
jointly define the mediated-moderation model. The simple effects
analyses results (see Table 10) further support Hypothesis 3. The first
stage of the indirect effect of abusive supervision on employee deviance was stronger for high DJ (P ⫽ .32, p ⬍ .01) than low DJ (P ⫽
.12, p ⬍ .05), and this difference was significant ( p ⬍ .05). These
results provide evidence of our mediated-moderation model.
We next compared Model 3 with Model 4, which added the
second-stage moderation effect (Intrusions ⫻ DJ) to Model 3. The
results show that the Intrusions ⫻ DJ interaction was not significant (␥ ⫽ ⫺.06, ns) and adding a second-stage moderation effect
to Model 3 did not improve model fit. These results provide further
support for Hypothesis 3, as the more complex mediated-
1023
-1
-.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Low (1 SD below mean)
Figure 5. Study 3: Interaction between abusive supervision and distributive justice on employee deviant
behaviors.
THAU AND MITCHELL
1024
Table 10
Study 3: Simple Effects Analyses for Deviant Employee Behaviors
First-stage moderation
Direct effect
Indirect effect
Total effect
Variable
PMX
95% CI
PYX
95% CI
PYMPMX
95% CI
(PYX ⫹ PYMPMX)
95% CI
DJ high
DJ how
Difference
.64ⴱⴱⴱ
.24ⴱⴱ
.39ⴱ
[.32, .94]
[.00, .49]
[.00, .75]
.28
.11
.18
[⫺.01, 58]
[⫺.12, .32]
[⫺.14, .50]
.32ⴱⴱ
.12ⴱ
.20ⴱ
[.11, .65]
[.01, .32]
[.02, .55]
.60ⴱⴱ
.23
.37ⴱ
[.28, .98]
[⫺.02, .46]
[.02, .84]
Note. PMX ⫽ path from abusive supervision to intrusive thoughts; PYX ⫽ path from abusive supervision to deviant employee behaviors; PYM ⫽ path from
ego depletion to deviant employee behaviors; CI ⫽ bias-corrected confidence interval; DJ ⫽ distributive justice.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
devoted to understanding, processing, and interpreting the supervisor’s behavior. Intrusions about the supervisor’s behavior measured specifically how the self becomes drained. We also used
alternative measures for employee deviance and DJ. Finally, we
employed a within-person design, showing that Study 2’s findings
generalize across levels of analysis. The results of Study 3 replicate the interactive pattern found in Studies 1 and 2. Furthermore,
by collecting DJ at the end of the periods, we potentially addressed
bias associated with sensitivity to the research purpose. Thinking
about how the supervisor treated participants might have colored
how they then perceived their rewards over time, introducing error
to our interaction test. Consequently, by design, our research
provides for a conservative, comparative test of the self-gain view
against the self-regulation impairment view of the abusive
supervision– employee deviance relationship.
General Discussion
.6
.8
Intrusive Thoughts
1
1.2
1.4
1.6
The goal of the current research was to examine whether a
self-gain or self-regulation view better explains why employees
respond to supervisor abuse with deviance. We argued that DJ
perceptions allowed for a comparative test because each view
makes opposing predictions about DJ’s influence on the abusive
supervision– employee deviance relationship. The self-gain
view predicts that DJ would weaken the effects of abuse on
deviance; compensatory logic suggests that the benefits of DJ
would offset the costs of abuse. The self-regulation impairment
view predicts that DJ would strengthen the effects of abuse on
deviance; receiving inconsistent information about the employees’ organizational utility would further drain the self-resources
needed to consider normative expectations and counterretaliation risks. Across diverse samples, measures, designs, and
statistical techniques, the results show that DJ strengthened the
positive relationship between abusive supervision and employee deviance.
The findings suggest that the reason why employees respond to
abusive supervision with deviance is because being abused drains
self-resources (particularly at high levels of DJ). This mediatedmoderation effect is consistent with self-regulatory strength
(Baumeister et al., 1998) and victimization models (Aquino &
Thau, 2009; Thau et al., 2007). Abuse prompts self-regulation
-1
-.5
0
Abusive Supervision
.5
1
Distributive Justice
High (1 SD above mean)
Low (1 SD below mean)
Figure 6. Study 3: Interaction between abusive supervision and distributive justice on intrusive thoughts
(self-regulation impairment).
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
impairment by draining self-resources that are needed to consider
the risks associated with retaliating and make deviance more
likely. Our studies programmatically tested these predictions.
Study 1’s design shows that DJ strengthened the abusive
supervision– deviance relationship. The designs of Studies 2 and 3
integrated two self-regulation impairment measures (Study 2: ego
depletion; Study 3: intrusive thoughts about the supervisor’s behavior).
The results in Studies 2 and 3 showed significant indirect effects
of abuse on deviance (particularly at high levels of DJ) that were
consistent with Hypothesis 3’s mediated-moderation prediction.
The indirect effects in Study 2 were comparatively weaker than the
indirect effects in Study 3. Supplemental analyses suggested that
the effects of self-regulation impairment in Study 2 partially mediated the relationship between Abusive Supervision ⫻ DJ on
deviance, whereas self-regulation impairment in Study 3 completely mediated the relationship. That we had found partial mediation in Study 2 and complete mediation in Study 3 further
supports our argument that the measure used in Study 2 (i.e., ego
depletion) did not quite map onto self-regulation impairment in the
context of abuse as well as the measure used in Study 3 (i.e.,
intrusive thoughts about the supervisor’s behavior). Although the
strength of the indirect effects we found differs across studies, the
overall pattern of results suggests that supervisor abuse influences
victims’ experienced self-regulation impairment (particularly at
high levels of DJ), which then influences their engaged deviant
behavior.
Overall, our research challenged the self-gain account of deviant
responses to supervisor abuse and speaks to a more complex
theoretical model— one that considers the consequences of victimization on self-regulation. Importantly, the results generalized
across samples, methods, and levels of analysis, highlighting the
breadth and robustness of the theoretical predictions made by the
self-regulation perspective. Tepper et al. (2009; see also Sitkin,
2007) argued that there is “tremendous value in conducting multiple tests of the same phenomenon” because it is evidence for the
robustness of the phenomenon (p. 165). Our studies also highlight
a key process variable (self-regulation impairment) and moderator
(DJ) that strengthens the effects of supervisor abuse on deviance.
Last, our studies’ results imply that organizational decision makers
cannot consider DJ as an ex ante control mechanism for supervisor
abuse. Fair outcomes do not lessen deviant responses to abusive
supervision; worse, they make deviance more likely.
Theoretical Implications
Traditional exchange explanations (e.g., reciprocity, social exchange, equity) of deviant responses to supervisor abuse share the
underlying assumption that retaliation is motivated by self-gain.
These explanations are parsimonious, but are limited. By violating
normative expectations, victims risk counterretaliation by others.
The risks of counterretaliation may be and often are greater than
the benefits that may come from retaliation (Aquino et al., 2001;
Tepper et al., 2009). If cost– benefit considerations drive deviant
responses to abuse, then additional benefits (high DJ) should make
deviance less likely and additional costs (low DJ) should make
them more likely. Instead, our research shows that the self-gain
view was unsupported and the self-regulation impairment view
was supported. DJ strengthened the relationship between abuse
1025
and employee deviance, and Studies 2 and 3 suggest that the
reason why this effect occurs is because (particularly at high levels
of DJ) supervisor abuse indirectly affects deviance through selfregulation impairment (ego depletion and intrusive thoughts).
Our findings imply the possibility that past research on the
abusive supervision– deviance relationship using traditional selfgain reasoning may be reframed as self-regulation impairment. For
example, Thau et al. (2009) argued that the abusive supervision—
employee deviance relationship was strengthened for those who
experienced high (rather than low) uncertainty in their exchange
relationship with the organization because victims devoted more
attention to the abusive supervisor, heightening the desire to retaliate. The results, however, may also imply that self-regulation
impairment prompted by abuse is magnified by the expenditure of
self-regulatory resources needed to cope with the uncertainty. Still,
we concede that for research that has not examined variables that
specifically account for self-gain, self-regulation impairment,
and/or other processes, understanding why abusive supervision
influences deviance is less clear.
Our test of the self-gain versus self-regulation view makes a
distinct contribution to understanding why victims of supervisor
abuse reciprocate with deviance and the results highlight limitations with traditional exchange reasoning, extending beyond the
supervisor abuse– employee deviance relationship. Some scholars
criticize the rational action core of traditional exchange reasoning
and argue that instrumental motives alone cannot always explain
why people reciprocate or fail to reciprocate (e.g., Clark & Mills,
1979; Elster, 1989; A. P. Fiske, 1991). To be sure, we do not
discount self-gain motives as a simple and powerful explanation of
exchange behavior. We instead argue that by solely relying on
self-gain principles, scholars may not ask questions and test predictions that go beyond the rational actor model (cf. Cropanzano,
Stein, & Goldman, 2007; De Dreu & Nauta, 2009; Korsgaard,
Meglino, Lester, & Jeong, 2010).
Our study shows the impact of social exchanges on the psychology of the self. Employees who are harmed in exchanges with their
supervisor are less able to maintain normative and rationally
appropriate behavior because the victimization experience drains
self-resources. In addition, our findings contribute to the foci
approach of exchange theory, showing that other exchange relations—those that provide inconsistent information about employees’ organizational membership (i.e., high DJ)—further drain selfresources. In short, our findings suggest that there is more to
exchange behavior than rational cost– benefit considerations. A
generalization from our results to theory construction, then, is that
self-gain models are less likely to account for exchange phenomena when individuals are harmed, self-impaired, and/or receive
inconsistent information about their organizational membership.
Whether these generalizations extend to other exchange situations
is an important avenue for future research.
Our research also uniquely contributes to an emerging view on
the role of the self in explaining harmful behaviors and reactions
to victimization (Aquino & Douglas, 2003; Ferris, Brown, &
Heller, 2009; Ferris, Brown, Lian, & Keeping; Thau et al., 2007).
It substantiates the need to consider psychological processes, as the
self is particularly fragile when individuals are victimized. Abused
employees’ self-resources become depleted; they become preoccupied and try to understand and process the transgressor’s behaviors, making behaviors that violate normative standards more
1026
THAU AND MITCHELL
likely. Thus, our research empirically supports Aquino and Thau’s
(2009) contention that “victimized employees may become locked
into a vicious cycle in which the experience of victimization
motivates them to engage in behaviors that invite further victimization” (p. 731).
Practical Implications
Organizational decision makers may believe controlling supervisors’ behavior is sometimes difficult, and so they may have little
faith in truly preventing supervisor abuse. Whereas DJ is generally
described as a well-functioning control mechanism for employee
behavior (Cropanzano & Greenberg, 1997), the current study
suggests that it is counterproductive as a compensatory mechanism
for supervisor abuse. Instead, fair outcomes exacerbate supervisorabused victims’ harmful responses. Consistent costs or consistent
benefits are a better strategy for organizations than inconsistencies
(J. N. Baron & Kreps, 2003). To control employee deviance, the
best strategy is to prevent supervisor abuse and allocate outcomes
fairly.
Indeed, abusive supervision undermines organizational effectiveness (Detert et al., 2007) because it leads to many negative
outcomes. It negatively influences subordinates’ work attitudes
(e.g., Aryee, Chen, Sun, & Debrah, 2007; Tepper, 2000), performance (Detert et al., 2007), and citizenship behaviors (Aryee et al.,
2007; Zellars, Tepper, & Duffy, 2002) and is highly destructive to
employees (e.g., psychological distress, work–family conflict;
Tepper, 2000). Our research provides sobering evidence to support
these arguments. Therefore, identifying abusive supervisors is an
essential first step to mitigating its destructive effects. Moreover,
providing training to supervisors on how to appropriately interact
with their employees is warranted.
Last, decision makers should consider how to help employees
minimize the self-draining effects of abuse. Recent theory and
research extending the self-regulatory strength model provide evidence on how to deal with abusive contexts, and suggest that
organizations may strengthen self-resources with training (e.g.,
Baumeister, Gailliot, DeWall, & Oaten, 2006; Gailliot, Plant, Butz,
& Baumeister, 2007; Muraven & Baumeister, 2000; Muraven,
Baumeister, & Tice, 1999; Oaten & Cheng, 2006). This research
refers to self-resources as akin to a muscle, where initial exercise
may be exhausting (e.g., abuse, victimization) but continuous,
fruitful exercise over time may strengthen it. This implies that
organizations can build interventions to assist employees who deal
with abusive others (i.e., counselors, police officers, customer
service agents). Employees might experience simulated abusive
interactions and be taught interpersonal skills to deal with them.
Over time, such training would strengthen self-resources and allow
employees to better cope with the situation, find constructive
solutions, and reduce destructive reactions.
Limitations
Our article is not without limitations. For example, all our data
are self-reported, which may promote common method variance
bias. Although researchers have found that common method variance does not pose a biasing problem (Spector, 1987, 2006), we do
not know whether we would have found the same results with
another source reporting some of the study variables. Yet, given the
sensitive nature of the phenomena of interest (i.e., abusive supervision, employee deviance), the method seems appropriate (e.g.,
Aquino & Thau, 2009; Dalal, 2005; Tepper, 2007), especially given
recent research evidence showing the limitations of other reports of
deviance (e.g., Dalal, 2005; Sackett, Berry, Wiemann, & Laczo,
2006). In addition, we believe the designs (the time lags) of Studies 2
and 3 lessen the possibility that common method variance biased our
data.
A last limitation is that our data are based on surveys, and so
causal inferences are limited. We did try to handle this by measuring the mediator at Time 2 in Study 2 and through a complex
within- and between-individuals design in Study 3. Ideally, we
would have liked to have a bigger sample in Study 3 to increase
statistical power, but we made a classical trade-off between the
costs of data collection and the likelihood of finding the effects
(Snijders & Bosker, 1999). Still, recent research suggests that the
results of Study 3 were unlikely biased by the sample size, particularly because bootstrapping was also used (Maas & Hox,
2005). Further, we note from social psychology experimental
results that victimization causes self-regulation impairment
(Baumeister et al., 2005), which causes harmful behaviors (Stucke
& Baumeister, 2006). Their evidence, the preliminary evidence we
cited earlier, and our results suggest that our proposed causal chain
is plausible.
Future Research Directions
This article provides ground for several new research questions.
For example, future research may examine the conditions under
which exchanges are governed by a self-gain versus self-regulation
impairment logic. We already generalized our results to theoretical
predictions under which the self-gain logic is unlikely to hold (e.g.,
experiencing harm, self-regulation impairment, inconsistent utility). A potential test of this idea is to investigate classical exchange
situations that are explained by self-gain motives (e.g., cooperation
in iterated vs. one-shot interactions; Axelrod, 1984), manipulate
one of the three generalized conditions under which self-gain is
unlikely to hold, and test for their effects.
Another research problem to investigate is whether the effects
we found are more pronounced in people who are particularly
vulnerable to victimization. For example, it may be that employees
whose self-worth is staked on their organizational membership are
more likely to undergo self-regulation impairment when they are
abused (Ferris, Brown, Lian, & Keeping, 2009). It may also be that
certain individuals hold characteristics (e.g., low trait self-control;
Marcus & Schuler, 2004) that make them more susceptible to
impairment following abuse.
Further, researchers might examine effects of self-regulation
impairment on behaviors other than deviance. Self-regulatory
strength research shows that self-regulation impairment influences
a variety of behaviors (see Baumeister, Heatherton, & Tice, 1994).
For instance, self-regulation impairment influences behaviors requiring regulatory impulse control with long-term costs (e.g.,
alcohol consumption, gambling, unnecessary spending, smoking,
unhealthy eating). Self-regulation impairment also influences behaviors that require attention to norms (e.g., helping or harming).
Potentially, impaired self-resources may also increase behaviors
that are typically thought of as “positive” (e.g., helping behaviors),
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
as long as the situational context defines helping as normatively
inappropriate. We leave these ideas for future study.
In addition, it would be interesting to see whether environmental
factors influence self-regulatory abilities and the abusive
supervision– employee deviance relationship. We argued that fair
distributions further drain self-resources when employees are
abused because they provide inconsistent information about the
utility of the organizational membership. We hope future researchers explore the effects of other situational cues that may provide
for inconsistencies. For example, it is possible that psychologically
safe climates intensify the effects of abuse. The effects of abuse
might also intensify if employees report to more than one supervisor and the other supervisor is highly supportive. We hope
researchers explore these ideas further.
Understanding other causes of self-regulation impairment is
needed. According to the regulatory depletion model, any activity
that involves self-regulation depletes self-resources. This may be
particularly true for employees who are forced to fake expressions
at work. J. L. Robinson and Demaree’s (2007) work shows that
experienced versus expressed emotions (feeling sad but forced to
smile) impair self-regulatory abilities, which worsen task performance. Similarly, self-regulation impairment may result from
holding inconsistent values. Research shows that employees put on
“facades of conformity” when their values do not match their
employers’ values, and these facades influence turnover intentions
and emotional exhaustion (Hewlin, 2003, 2009). Our research
suggests that the reason why may be self-regulation impairment.
Finally, researchers might consider examining in more detail the
dynamics of dissonance. We believe our results provide suggestive
evidence of the dissonance experience. However, measuring dissonance directly and exploring the ways by which employees
attempt to reduce dissonance when they receive inconsistent information about their organizational membership might provide
insight on the process and when dissonance-reducing attempts are
effective and ineffective.
Conclusion
Self-gain assumptions provide a parsimonious starting point to a
theory of retaliation, but they cannot explain why— given the
probable risks of counterretaliation—abused employees tend to
respond with deviant behaviors. To account for this puzzle, assumptions about how victimization affects the self are necessary.
Our research showed how the self-regulation impairment perspective embraces assumptions that can explain why victims of abuse
fail to consider the likely costs of retaliation, and these predictions
were supported in studies involving different data and methodological approaches.
More generally, we believe our study speaks to the issue of
when the assumption of rational actors is useful and when it is not
in explaining organizational behavior. A general heuristic could be
to start off with the parsimonious assumption of self-gain and
retain it as long as the theory does not meet empirical facts
implying “irrational behavior” (Lindenberg, 1992). Our findings
suggest that when we observe seemingly irrational phenomena, we
need to extend self-gain models with assumptions about how the
particular social context in which the puzzling behavior takes place
affects the psychology of self-regulation. In this spirit, we chal-
1027
lenge future research to consider the role of the self in explaining
reciprocity in social exchanges.
References
Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 2, pp. 267–299). New
York, NY: Academic Press.
Aquino, K., & Douglas, S. (2003). Identity threat and antisocial behavior
in organizations: The moderating effects of individual differences, aggressive modeling, and hierarchical status. Organizational Behavior and
Human Decision Processes, 90, 195–208.
Aquino, K., Lewis, M. U., & Bradfield, M. (1999). Justice constructs,
negative affectivity, and employee deviance: A proposed model and
empirical test. Journal of Organizational Behavior, 20, 1073–1091.
Aquino, K., & Thau, S. (2009). Workplace victimization: Aggression from
the target’s perspective. Annual Review of Psychology, 60, 717–741.
Aquino, K., Tripp, T. M., & Bies, R. J. (2001). How employees respond to
personal offense: The effects of blame attribution, victim status, and
offender status on revenge and reconciliation in the workplace. Journal
of Applied Psychology, 86, 52–59.
Aronson, E. (1968). Dissonance theory: Progress and problems. In R. P.
Abelson et al. (Eds.), Theories of cognitive consistence: A sourcebook
(pp. 5–27). Chicago, IL: Rand McNally.
Arrow, K. J. (1963). Uncertainty and the welfare economics of medical
care. American Economic Review, 52, 941–973.
Aryee, S., Chen, Z., Sun, L.-Y., & Debrah, Y. A. (2007). Antecedents and
outcomes of abusive supervision: Test of a trickle-down model. Journal
of Applied Psychology, 92, 191–201.
Axelrod, R. (1984). The evolution of cooperation. New York, NY: Basic
Books.
Barnard, C. I. (1938). The functions of the executive. Cambridge, MA:
Harvard University Press.
Baron, J. N., & Kreps, D. M. (2003). Strategic human resources: Frameworks for general managers. New York, NY: Wiley.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Baumeister, R. F. (1998). The self. In D. T. Gilbert, S. T. Fiske, & G.
Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 1, pp.
680 –740). New York, NY: McGraw-Hill.
Baumeister, R. F. (2001). Ego depletion, the executive function, and
self-control: An energy model of the self in personality. In B. W. Roberts
& R. Hogan (Eds.), Personality psychology in the workplace: Decade of
behavior (pp. 299 –316). Washington, DC: American Psychological
Association.
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998).
Ego depletion: Is the active self a limited resource? Journal of Personality and Social Psychology, 74, 1252–1265.
Baumeister, R. F., DeWall, C. N., Ciarocco, N. J., & Twenge, J. M. (2005).
Social exclusion impairs self-regulation. Journal of Personality and
Social Psychology, 88, 589 – 604.
Baumeister, R. F., Gailliot, M., DeWall, C. N., & Oaten, M. (2006).
Self-regulation and personality: How interventions increase regulatory
success, and how depletion moderates the effects of traits on behavior.
Journal of Personality, 74, 1773–1801.
Baumeister, R. F., & Heatherton, T. F. (1996). Self-regulation failure: An
overview. Psychological Inquiry, 7, 1–15.
Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1994). Losing control:
How and why people fail at self-regulation. San Diego, CA: Academic
Press.
Baumeister, R. F., & Vohs, K. D. (2007). Self-regulation, ego depletion,
and motivation. Social and Personality Psychology Compass, 1, 115–
128.
1028
THAU AND MITCHELL
Beal, D. J., Weiss, H. M., Barros, E., & MacDermid, S. M. (2005). An
episodic process model of affective influences on performance. Journal
of Applied Psychology, 90, 1054 –1068.
Bennett, R. J., & Robinson, S. L. (2000). The development of a measure of
workplace deviance. Journal of Applied Psychology, 85, 349 –360.
Bies, R. J. (1987). The predicament of injustice: The management of moral
outrage. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 9, pp. 289 –319). Greenwich, CT: JAI Press.
Bies, R. J., & Tripp, T. M. (1995). The use and abuse of power: Justice as
social control. In R. S. Cropanzano & K. M. Kacmar (Eds.), Organizational politics, justice, and support: Managing the social climate of the
workplace (pp. 131–145). New York, NY: Quorum Press.
Bies, R. J., & Tripp, T. M. (1996). Beyond distrust: “Getting even” and the
need for revenge. In R. M. Kramer & T. R. Tyler (Eds.), Trust in
organizations: Frontiers of theory and research (pp. 246 –260). Thousand Oaks, CA: Sage.
Blau, P. M. (1964). Exchange and power in social life. New York, NY:
Wiley.
Bowling, N. A., & Beehr, T. A. (2006). Workplace harassment from the
victim’s perspective: A theoretical model and meta-analysis. Journal of
Applied Psychology, 91, 998 –1012.
Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcissism, self-esteem, and direct and displaced aggression: Does self-love or
self-hate lead to violence? Journal of Personality and Social Psychology, 75, 219 –229.
Chen, G., Bliese, P. D., & Mathieu, J. E. (2005). Conceptual framework
and statistical procedures for delineating and testing multilevel theories
of homology. Organizational Research Methods, 8, 375– 409.
Clark, M. S., & Mills, J. (1979). Interpersonal attraction in exchange and
communal relationships. Journal of Personality and Social Psychology,
37, 12–24.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple
regression/correlation analysis for the behavioral sciences (3rd ed.).
Mahwah, NJ: Erlbaum.
Cohen-Charash, Y., & Spector, P. E. (2001). The role of justice in organizations: A meta-analysis. Organizational Behavior and Human Decision Processes, 86, 278 –321.
Coleman, J. S. (1990). Foundations of social theory. Cambridge, MA:
Belknap.
Colquitt, J. A. (2001). On the dimensionality of organizational justice: A
construct validation of a measure. Journal of Applied Psychology, 86,
386 – 400.
Colquitt, J. A., Conlon, D. E., Wesson, M. J., Porter, C. O. L. H., & Ng,
K. Y. (2001). Justice at the millennium: A meta-analytic review of 25
years of organizational justice research. Journal of Applied Psychology,
86, 425– 445.
Colquitt, J. A., & Greenberg, J. (2003). Organizational justice: A fair
assessment of the state of the literature. In J. Greenberg (Ed.), Organizational behavior: The state of the science (2nd ed., pp. 165–210).
Mahwah, NJ: Erlbaum.
Cook, K. S., & Emerson, R. M. (1978). Power, equity and commitment in
exchange networks. American Sociological Review, 43, 721–739.
Cropanzano, R., & Ambrose, M. L. (2001). Procedural and distributive
justice are more similar than you think: A monistic perspective and
research agenda. In J. Greenberg & R. Cropanzano (Eds.), Advances in
organizational justice (pp. 119 –151). Palo Alto, CA: Stanford University Press.
Cropanzano, R., & Greenberg, J. (1997). Progress in organizational justice:
Tunneling through the maze. In C. L. Cooper & I. T. Robertson (Eds.),
International review of industrial and organizational psychology (Vol.
12, pp. 317–372). New York, NY: Wiley.
Cropanzano, R., & Mitchell, M. S. (2005). Social exchange theory: An
interdisciplinary review. Journal of Management, 31, 874 –900.
Cropanzano, R., & Rupp, D. E. (2008). Social exchange theory and
organizational justice: Job performance, citizenship behaviors, multiple
foci, and a historical integration of two literatures. In S. W. Gilliland,
D. P. Skarlicki, & D. D. Steiner (Eds.), Justice, morality, and social
responsibility (pp. 63–99). Greenwich, CT: Information Age.
Cropanzano, R., Rupp, D. E., Mohler, C., & Schminke, M. (2001). Three
roads to organizational justice. In J. Martocchio, H. Liao, & A. Joshi
(Eds.), Research in personnel and human resources management (Vol.
20, pp. 1–113). Greenwich, CT: JAI Press.
Cropanzano, R., Stein, J., & Goldman, B. M. (2007). Individual aesthetics:
Self-interest. In E. H. Kessler & J. R. Bailey (Eds.), Handbook of
organizational and managerial wisdom (pp. 181–221). Thousand Oaks,
CA: Sage.
Dalal, R. S. (2005). A meta-analysis of the relationship between organizational citizenship behavior and counterproductive work behavior.
Journal of Applied Psychology, 90, 1241–1255.
Dalal, R. S., Lam, H., Weiss, H. M., Welch, E. R., & Hulin, C. L. (2009).
A within-person approach to work behavior and performance: Concurrent and lagged citizenship– counterproductivity associations, and dynamic relationships with affect and overall job performance. Academy of
Management Journal, 52, 1051–1066.
De Dreu, C. K. W., & Nauta, A. (2009). Self-interest and other-orientation
in organizational behavior: Implications for job performance, prosocial
behavior, and personal initiative. Journal of Applied Psychology, 94,
913–926.
DeJordy, R. (2008). Just passing through: Stigma, passing, and identity
decoupling in the work place. Group & Organization Management, 33,
504 –531.
Detert, J. R., Treviño, L. K., Burris, E. R., & Andiappan, M. (2007).
Managerial modes of influence and counterproductivity in organizations: A longitudinal business-unit-level investigation. Journal of Applied Psychology, 92, 993–1005.
Deutsch, M. (1975). Equity, equality, and need: What determines which
value will be used as the basis for distributive justice? Journal of Social
Issues, 31, 137–149.
DeWall, C. N., Baumeister, R. F., Stillman, T. F., & Gailliot, M. T. (2007).
Violence restrained: Effects of self-regulation and its depletion on aggression. Journal of Experimental Social Psychology, 43, 62–76.
Dornbusch, S. M., & Scott, W. R. (1975). Evaluation and the exercise of
authority. San Francisco, CA: Jossey-Bass.
Duffy, M. K., Ganster, D. C., & Pagon, M. (2002). Social undermining in
the workplace. Academy of Management Journal, 45, 331–351.
Duffy, M. K., Ganster, D. C., Shaw, J. D., Johnson, J. L., & Pagon, M.
(2006). The social context of undermining behavior at work. Organizational Behavior and Human Decision Processes, 101, 105–126.
Edwards, J. R., & Lambert, L. S. (2007). Methods for integrating moderation and mediation: A general analytical framework using moderated
path analysis. Psychological Methods, 12, 1–22.
Eisenberger, R., Cummings, J., Armeli, S., & Lynch, P. (1997). Perceived
organizational support, discretionary treatment, and job satisfaction.
Journal of Applied Psychology, 82, 812– 820.
Ekeh, P. P. (1974). Social exchange theory: The two traditions. Cambridge,
MA: Harvard University Press.
Elster, J. (1989). The cement of society: A study of social order. Cambridge, MA: Cambridge University Press.
Emerson, R. M. (1976). Social exchange theory. Annual Review of Sociology, 2, 335–362.
Erdogan, B., & Enders, J. (2007). Support from the top: Supervisors’
perceived organizational support as a moderator of leader–member
exchange to satisfaction and performance relationships. Journal of Applied Psychology, 92, 321–330.
Ferris, D. L., Brown, D. J., & Heller, D. (2009). Organizational supports
and organizational deviance: The mediating role of organization-based
self-esteem. Organizational Behavior and Human Decision Processes,
108, 279 –286.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
Ferris, D. L., Brown, D. J., Lian, H., & Keeping, L. M. (2009). When does
self-esteem relate to deviant behavior? The role of contingencies of
self-worth. Journal of Applied Psychology, 94, 1345–1353.
Festinger, L. (1957). A theory of cognitive dissonance. Palo Alto, CA:
Stanford University Press.
Finkel, E. J., & Campbell, W. K. (2001). Self-control and accommodation
in close relationships: An interdependence analysis. Journal of Personality and Social Psychology, 81, 263–277.
Fischer, P., Greitemeyer, T., & Frey, D. (2007). Ego depletion and positive
illusions: Does the construction of positivity require regulatory resources? Personality and Social Psychology Bulletin, 33, 1306 –1321.
Fischer, P., Greitemeyer, T., & Frey, D. (2008). Self-regulation and selective exposure: The impact of depleted self-regulation resources on
confirmatory information processing. Journal of Personality and Social
Psychology, 94, 382–395.
Fiske, A. P. (1991). Structures of social life: The four elementary forms of
human relations. New York, NY: Free Press.
Fiske, S. T. (2004). Developing a program of research. In C. Sansone, C.
Morf, & A. T. Panter (Eds.), The Sage handbook of methods in social
psychology (pp. 71–90). Thousand Oaks, CA: Sage.
Fiske, S. T., & Taylor, S. E. (2008). Social cognition: From brains to
culture. New York, NY: McGraw-Hill.
Gailliot, M. T., Plant, E. A., Butz, D. A., & Baumeister, R. F. (2007).
Increasing self-regulatory strength can reduce the depleting effects of
suppressing stereotypes. Personality and Social Psychology Bulletin, 33,
281–294.
Gailliot, M. T., Schmeichel, B. J., & Baumeister, R. F. (2006). Selfregulatory processes defend against the threat of death: Effects of
self-control depletion and trait self-control on thoughts and fears of
dying. Journal of Personality and Social Psychology, 91, 49 – 62.
Gergen, K. J. (1980). Exchange theory: The transient and the enduring. In
K. J. Gergen, M. S. Greenberg, & R. H. Wills (Eds.), Social exchange:
Advances in theory and research (pp. 261–280). New York, NY: Plenum
Press.
Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement.
American Sociological Review, 25, 161–178.
Greenberg, J. (1988). Equity and workplace status: A field experiment.
Journal of Applied Psychology, 73, 606 – 613.
Greenwald, A. G., & Ronis, D. L. (1978). Twenty years of cognitive
dissonance: Case study of the evaluation of a theory. Psychological
Review, 85, 53–57.
Heider, F. (1958). The psychology of interpersonal relations. New York,
NY: Wiley.
Hewlin, P. F. (2003). And the award for best actor goes to . . . : Facades of
conformity in organizational settings. Academy of Management Review,
28, 633– 642.
Hewlin, P. F. (2009). Wearing the cloak: Antecedents and consequences of
creating facades of conformity. Journal of Applied Psychology, 94,
727–741.
Homans, G. C. (1961). Social behavior: Its elementary forms. New York,
NY: Harcourt, Brace & World.
Horowitz, M. J., Wilner, N., & Alvarez, W. (1979). Impact of Event Scale:
A measure of subjective stress. Psychosomatic Medicine, 41, 209 –218.
Hox, J. J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum.
Judge, T. A., Scott, B. A., & Ilies, R. (2006). Hostility, job attitudes, and
workplace deviance: Test of a multilevel model. Journal of Applied
Psychology, 91, 126 –138.
Kelley, H. H. (1968). Interpersonal accommodation. American Psychologist, 23, 399 – 410.
Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory
of interdependence. New York, NY: Wiley.
Klein, K. J., Dansereau, F., & Hall, R. J. (1994). Levels issues in theory
1029
development, data collection, and analysis. Academy of Management
Review, 19, 195–229.
Knutson, B. (2004). Sweet revenge? Science, 305, 1246 –1247.
Korsgaard, M. A., Meglino, B. M., Lester, S. W., & Jeong, S. S. (2010).
Paying you back or paying me forward: Understanding rewarded and
unrewarded organizational citizenship behavior. Journal of Applied Psychology, 95, 277–290.
Kreft, I. G. G., de Leeuw, J., & Aiken, L. S. (1995). The effect of different
forms of centering in hierarchical linear models. Multivariate Behavioral
Research, 30, 1–21.
Lavelle, J. J., Rupp, D. E., & Brockner, J. (2007). Taking a multifoci
approach to the study of justice, social exchange, and citizenship behavior: The target similarity model. Journal of Management, 33, 841– 866.
Lindenberg, S. (1992). The method of decreasing abstraction. In J. S.
Coleman & T. J. Fararo (Eds.), Rational choice theory: Advocacy and
critique (pp. 3–20). Newbury Park, CA: Sage.
Locke, E. A. (1996). Using programmatic research to build a grounded
theory. In P. J. Frost & M. S. Taylor (Eds.), Rhythms of academic life:
Personal accounts of careers in academia (pp. 99 –106). Thousand
Oaks, CA: Sage.
Locke, E. A. (2007). The case for inductive theory building. Journal of
Management, 33, 867– 890.
Lykken, D. T. (1968). Statistical significance in psychological research.
Psychological Bulletin, 70, 151–159.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel
modeling. Methodology, 1, 86 –92.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., &
Sheets, V. (2002). A comparison of methods to test mediation and other
intervening variable effects. Psychological Methods, 7, 83–104.
Marcus, B., & Schuler, H. (2004). Antecedents of counterproductive behavior at work: A general perspective. Journal of Applied Psychology,
89, 647– 660.
Matthews, G., Schwean, V. L., Campbell, S. E., Saklofske, D. H., &
Mohamed, A. A. R. (2000). Personality, self-regulation, and adaptation:
A cognitive-social framework. In M. Boekaerts, P. R. Pintrich, & M.
Zeidner (Eds.), Handbook of self-regulation (pp. 171–207). San Diego,
CA: Academic Press.
McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin, 114,
376 –390.
McCullough, M. E., Bellah, C. G., Kilpatrick, S. D., & Johnson, J. L.
(2001). Vengefulness: Relationships with forgiveness, rumination, wellbeing, and the Big Five. Personality and Social Psychology Bulletin, 27,
601– 610.
McFarlin, D. B., & Sweeney, P. D. (1992). Distributive and procedural
justice as predictors of satisfaction with personal and organizational
outcomes. Academy of Management Journal, 35, 626 – 637.
Meeker, B. F. (1971). Decisions and exchange. American Sociological
Review, 36, 485– 495.
Mitchell, M. S., & Ambrose, M. L. (2007). Abusive supervision and
workplace deviance and the moderating effects of negative reciprocity
beliefs. Journal of Applied Psychology, 92, 1159 –1168.
Molm, L. D., Quist, T. M., & Wiseley, P. A. (1994). Imbalanced structures,
unfair strategies: Power and justice in social exchange. American Sociological Review, 59, 98 –121.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of
limited resources: Does self-control resemble a muscle? Psychological
Bulletin, 126, 247–259.
Muraven, M., Baumeister, R. F., & Tice, D. M. (1999). Longitudinal
improvement of self-regulation through practice: Building self-control
strength through repeated exercise. Journal of Social Psychology, 139,
446 – 457.
Muraven, M., Tice, D. M., & Baumeister, R. F. (1998). Self-control as a
1030
THAU AND MITCHELL
limited resource: Regulatory depletion patterns. Journal of Personality
and Social Psychology, 74, 774 –789.
Niehoff, B. P., & Moorman, R. H. (1993). Justice as a mediator of the
relationship between methods of monitoring and organizational citizenship behavior. Academy of Management Journal, 36, 527–556.
Oaten, M., & Cheng, K. (2006). Longitudinal gains in self-regulation from
regular physical exercise. British Journal of Health Psychology, 11,
717–733.
Oaten, M., Williams, K. D., Jones, A., & Zadro, L. (2008). The effects of
ostracism on self-regulation in the socially anxious. Journal of Clinical
and Social Psychology, 27, 471–504.
Platt, J. R. (1964). Strong inference: Certain systematic methods of scientific thinking may produce much more rapid progress than others.
Science, 146, 347–353.
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003).
Common method biases in behavioral research: A critical review of the
literature and recommended remedies. Journal of Applied Psychology,
88, 879 –903.
Prendergast, C., & Topel, R. (1993). Discretion and bias in performance
evaluation. European Economic Review, 37, 355–365.
Pruitt, D. G. (1968). Reciprocity and credit building in a laboratory dyad.
Journal of Personality and Social Psychology, 8, 143–147.
Rashbash, J., Charlton, C., Browne, W. J., Healy, M., & Cameron, B.
(2005). MLwiN Version 2.02 [Software]. Bristol, England: Centre for
Multilevel Modelling, University of Bristol.
Richards, J. M., & Gross, J. J. (1999). Composure at any cost? The
cognitive consequences of emotion suppression. Personality and Social
Psychology Bulletin, 25, 1033–1044.
Robinson, J. L., & Demaree, H. A. (2007). Physiological and cognitive
effects of expressive dissonance. Brain and Cognition, 63, 70 –78.
Robinson, S. L., & Bennett, R. J. (1995). A typology of deviant workplace
behaviors: A multidimensional scaling study. Academy of Management
Journal, 38, 555–572.
Robinson, S. L., & O’Leary-Kelly, A. M. (1998). Monkey see, monkey do:
The influence of work groups on the antisocial behavior of employees.
Academy of Management Journal, 41, 658 – 672.
Sackett, P. R., Berry, C. M., Wiemann, S. A., & Laczo, R. M. (2006).
Citizenship and counterproductive behavior: Clarifying relations between the two domains. Human Performance, 19, 441– 464.
Sassenberg, K., Moskowitz, G. B., Jacoby, J., & Hansen, N. (2007). The
carry-over effect of competition: The impact of competition on prejudice
towards uninvolved outgroups. Journal of Experimental Social Psychology, 43, 529 –538.
Saul, J. R., Simon, L., & Colquitt, J. A. (2008, April). When ethical leaders
undermine: A complex supervisor–subordinate interaction. Paper presented at the annual meeting of the Society for Industrial and Organizational Psychology, San Francisco, CA.
Schmader, T., & Johns, M. (2003). Converging evidence that stereotype
threat reduces working memory capacity. Journal of Personality and
Social Psychology, 85, 440 – 452.
Schmeichel, B. J., & Baumeister, R. F. (2004). Self-regulatory strength. In
R. F. Baumeister & K. D. Vohs (Eds.), Handbook of self-regulation:
Research, theory, and applications (pp. 84 –98). New York, NY: Guilford Press.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422– 445.
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis:
Modeling change and event occurrence. New York, NY: Oxford University Press.
Sitkin, S. B. (2007). Promoting a more generative and sustainable organizational science. Journal of Organizational Behavior, 28, 841– 848.
Skarlicki, D. P., & Folger, R. (1997). Retaliation in the workplace: The
roles of distributive, procedural, and interactional justice. Journal of
Applied Psychology, 82, 434 – 443.
Skitka, L. J., Winquist, J., & Hutchinson, S. (2003). Are outcome fairness
and outcome favorability distinguishable psychological constructs? A
meta-analytic review. Social Justice Research, 16, 309 –341.
Sluss, D. M., Klimchak, M., & Holmes, J. J. (2008). Perceived organizational support as a mediator between relational exchange and organizational identification. Journal of Vocational Behavior, 73, 457– 464.
Snijders, T. A. B. (1996). Analysis of longitudinal data using the hierarchical linear model. Quality and Quantity, 30, 405– 426.
Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, England:
Sage.
Spector, P. E. (1987). Method variance as an artifact in self-reported affect
and perceptions at work: Myth or significant problem? Journal of
Applied Psychology, 72, 438 – 443.
Spector, P. E. (2006). Method variance in organizational research: Truth or
urban legend? Organizational Research Methods, 9, 221–232.
Spencer, S., & Rupp, D. E. (2009). Angry, guilty, and conflicted: Injustice
toward coworkers heightens emotional labor through cognitive and
emotional mechanisms. Journal of Applied Psychology, 94, 429 – 444.
Stucke, T. S., & Baumeister, R. F. (2006). Ego depletion and aggressive
behavior: Is the inhibition of the aggression a limited resource? European Journal of Social Psychology, 36, 1–13.
Tangirala, S., Green, S. G., & Ramanujam, R. (2007). In the shadow of the
boss’s boss: Effects of supervisors’ upward exchange relationships on
employees. Journal of Applied Psychology, 92, 309 –320.
Tepper, B. J. (2000). Consequences of abusive supervision. Academy of
Management Journal, 43, 178 –190.
Tepper, B. J. (2007). Abusive supervision in work organizations: Review,
synthesis, and research agenda. Journal of Management, 33, 261–289.
Tepper, B. J., Carr, J. C., Breaux, D. M., Geider, S., Hu, C., & Hua, W.
(2009). Abusive supervision, intentions to quit, and employees’ workplace deviance: A power/dependence analysis. Organizational Behavior
and Human Decision Processes, 109, 156 –167.
Tepper, B. J., Duffy, M. K., Henle, C. A., & Lambert, L. S. (2006).
Procedural injustice, victim precipitation, and abusive supervision. Personnel Psychology, 59, 101–123.
Tepper, B. J., Henle, C. A., Lambert, L. S., Giacalone, R. A., & Duffy,
M. K. (2008). Abusive supervision and subordinates’ organizational
deviance. Journal of Applied Psychology, 93, 721–732.
Thau, S., Aquino, K., & Poortvliet, P. M. (2007). Self-defeating behaviors in
organizations: The relationship between thwarted belonging and interpersonal work behaviors. Journal of Applied Psychology, 92, 840 – 847.
Thau, S., Bennett, R. J., Mitchell, M. S., & Marrs, M. B. (2009). How
management style moderates the relationship between abusive supervision and workplace deviance: An uncertainty management theory perspective. Organizational Behavior and Human Decision Processes, 108,
79 –92.
Tice, D. M., Baumeister, R. F., Shmueli, D., & Muraven, M. (2007).
Restoring the self: Positive affect helps improve self-regulation following ego depletion. Journal of Experimental Social Psychology, 43,
379 –384.
Tripp, T. M., & Bies, R. J. (1997). What’s good about revenge? The
avenger’s perspective. In R. J. Lewicki, R. J. Bies, & B. H. Sheppard
(Eds.), Research on negotiation in organizations (Vol. 6, pp. 145–160).
Greenwich, CT: JAI Press.
Tripp, T. M., Bies, R. J., & Aquino, K. (2002). Poetic justice or petty
jealousy? The aesthetics of revenge. Organizational Behavior and Human Decision Processes, 89, 966 –984.
Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001).
If you can’t join them, beat them: Effects of social exclusion on
aggressive behavior. Journal of Personality and Social Psychology,
81, 1058 –1069.
SELF-GAIN OR SELF-REGULATION IMPAIRMENT?
Twenge, J. M., Muraven, M., & Tice, D. M. (2004). Measuring state
self-control: Reliability, validity, and correlations with physical and
psychological stress. Unpublished manuscript, San Diego State University, San Diego, CA.
Walster, E., Walster, G. W., & Berscheid, E. (1978). Equity: Theory and
research. Boston, MA: Allyn & Bacon.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological
Review, 101, 34 –52.
1031
Zellars, K. L., Tepper, B. J., & Duffy, M. K. (2002). Abusive supervision
and subordinates’ organizational citizenship behavior. Journal of Applied Psychology, 87, 1068 –1076.
Received August 25, 2009
Revision received June 1, 2010
Accepted June 8, 2010 䡲
Correction to Thau and Mitchell (2010)
In the Online First Publication of the article “Self-Gain or Self-Regulation Impairment? Tests of
Competing Explanations of the Supervisor Abuse and Employee Deviance Relationship Through
Perceptions of Distributive Justice,” by Stefan Thau and Marie S. Mitchell (Journal of Applied
Psychology, doi:10.1037/a0020540), the note to Table 10 on p. 1024 inadvertently referred to PYM
as the path from ego depletion to deviant employee behaviors. The note instead should have referred
to PYM as the path from intrusive thoughts to deviant employee behaviors.
DOI: 10.1037/a0021629