Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/23045081?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The
Academy of Management Journal
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
® Academy of Management Journal
2011, Vol. 54, No. 2, 279-294.
BENNETT J. TEPPER
Georgia State University
SHERRY E. MOSS
Wake Forest University
MICHELLE K. DUFFY
University of Minnesota
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder's express
written permission. Users may print, download or email articles for individual use only.
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
tices," various forms of hostility and mistreatment tures the supervisor's perception that the subordi
that range from mild (e.g., rudeness [Opotow, nate meets performance standards on required or
2001]) to severe (e.g., violations of fundamental in-role tasks. Because our interest is in predicting
human rights [Opotow, 1990a]). Of direct relevance supervisors' mistreatment of subordinates and be
to the focus of our research, supervisors execute cause people react more strongly to their percep
abusive acts against targets who are morally ex tion of their environment rather than to its objec
cluded from the supervisors' scope of justice tive features (Lewin, 1951), we focus on supervisor
(Opotow, 1995). Hence, a promising approach to perceptions of dissimilarity, relationship conflict,
modeling the abuse of specific subordinates and subordinate performance.
involves incorporating the precursors of moral The moral exclusion theory (Opotow, 1990a,
exclusion.
1995) prediction is that abusive supervision is pos
What factors cause targets to become morally ex itively related to perceived deep-level dissimilarity
cluded from an agent's scope of justice and, in turn, and supervisor perceptions of relationship conflict
to become targets for mistreatment? The moral ex with subordinates, and negatively related to super
clusion literature focuses on three recurring visor evaluations of subordinate performance.
themes: (1) perceived dissimilarity to targets, (2) However, contributions to three bodies of theory
conflict with the targets, and (3) the targets' useful and research suggest a more complicated set of
ness or "utility" (Hafer & Olson, 2003). An individ interrelationships and corresponding predictions.
ual becomes a target of hostile behavior when a These research streams are the workplace diversity
perpetrator perceives the target to be dissimilar, literature, which examines the effects of supervi
when the perpetrator is in conflict with the target, sor-subordinate dissimilarity on dyadic and indi
and when the target is not useful or is even injuri vidual attitudes and well-being (Harrison & Klein,
ous to the perpetrator. We identified supervisor 2007; Williams & O'Reilly, 1998); the work on con
subordinate analogs of these three factors to ex flict in interpersonal relations, which examines the
plore the predictors of abusive supervision. The emergence and consequences of frustration and
respective analogs for dissimilarity, conflict, and negative interaction between coworkers (Jehn &
utility are a supervisor's perceived deep-level dis Mannix, 2001); and the victim precipitation litera
similarity with a subordinate, perceived relation ture, which examines the characteristics that put
ship conflict with the subordinate, and evaluation people at risk of mistreatment (Elias, 1986; Olweus,
of the subordinate's performance. Perceived deep 1978). We invoked and integrated concepts from
level dissimilarity refers to the perception that the these literatures to develop and test a new model of
subordinate's values and attitudes differ from the abusive supervision. According to our model,
supervisor's (Harrison & Klein, 2007). Perceived which is depicted in Figure 1, supervisors experi
relationship conflict with the subordinate refers to ence relationship conflict with and assign lower
negative social interactions, interpersonal incom performance evaluations to subordinates who are
patibility, and negative affect in the form of frustra perceived to be dissimilar, which, in turn, is asso
tion, irritation, and annoyance (Jehn & Mannix, ciated with subordinates' reports that they have
2001). Subordinate performance evaluation cap been the target of abusive supervision. The model
FIGURE 1
Hypothesized Model
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
also specifies that the indirect dinates? The effectmoral exclusion of literature
perceived suppo
deep-level dissimilarity this very thesis. relationship
through Opotow invoked socialcon identi
flict is stronger when supervisors theory to argue that "moral perceive subor
exclusion results fro
dinates to be poorer performers. our innate tendency to differentiate object
We contribute to management (1990a: 7). People categorize others
theory by devel as similar o
oping and testing a model dissimilarofand demonstrate
abusive favoritism toward sim
supervision
that incorporates factorsilar others and
that havederogation
been toward dissimilar others
identified
in the moral exclusion literature (Brewer, 1999; Hewstone, and Rubin,processes
& Willis, 2002; de
scribed in theory and research Tajfel & Turner, on1979).workplace
Perceived dissimilarity diver
sity, relationship conflict, evokesand feelings of unconnectedness
victim and indiffer
precipitation.
In so doing, we demonstrate ence to potential
the threats to focal others' well-being
usefulness of using
moral exclusion concepts (Deutsch, to predict 1973; Williams & O'Reilly, 1998),
a costly form atti of
antisocial organizationaltudes that align with the
behavior. We concept of moral
also exclu
enrich
and sharpen moral exclusion sion. Indeed,
theoryOpotow (1990a, 1995) argued that
by introducing
fine-grained predictions perpetrators
regarding of hostile and the
aggressive acts against
roles that
dissimilarity, conflict, and dissimilarutility
others consciously
play or unconsciously
in explain ex
ing supervisors' exclusionary clude these dissimilar victims from In
practices. their scope
the of fol
lowing sections, we explain justice which,thein turn, produces exclusionary bases
conceptual prac
for the linkages depicted tices—acts in our hypothesized
of mistreatment and harm doing. Hence,
model. We then present thethe
moral exclusion
results literature ofprovides
a the basis for
study in
which we tested our predictions predicting that supervisorsusing two-wave
will abuse subordinates
data collected from matched they perceive pairs of
to be dissimilar supervisors
to themselves.
and subordinates. However, recent contributions to the workplace
diversity literature suggest that the role perceived
dissimilarity plays in predicting abusive supervi
THEORETICAL BACKGROUND sion may be less straightforward. This work sug
gests that under some circumstances diversity may
Perceived Deep-Level Dissimilarity and
be valued rather than derogated (Homan, Hollen
Abusive Supervision
beck, Humphrey, Van Knippenberg, Ilgen, & Van
Kleef, 2008; Homan, van Knippenberg, Van Kleef, &
Extant studies of diversity in supervisor-subordi
nate dyads have examined relationships between De Dreu, 2007) and that greater propensity may
dissimilarity and an assortment of attitudinal,exist
relato promote similar others than to derogate
tional, and behavioral outcomes. Some of these those who are different (Halevy, Bornstein, & Sagiv,
2008). Hence, this literature suggests that perceived
studies have focused on surface-level dissimilarity,
deep-level dissimilarity may not necessarily be as
comprising objective differences in age, gender,
and race; others have focused on deep-level sociated
dissim with abusive supervision. In the sections
that
ilarity, comprising perceived differences in values, follow, we reconcile this work with moral ex
attitudes, and personality. As we noted above, wetheory by proposing that perceived deep
clusion
level dissimilarity plays a role in predicting abu
restrict our analysis to perceived deep-level dissim
sive supervision, but not precisely the role that
ilarity—specifically, the perceptions of supervisors
moral
that their attitudes and values differ from those ofexclusion theory specifies. Specifically, we
focal subordinates. As Pulakos and Wexleypropose said, that the relationship between perceived
"Actual similarity may not be as importantdeep-level
to the dissimilarity and abusive supervision is
process of manager-subordinate interpersonal rela operating through supervisor perceptions
indirect,
of relationship conflict and supervisor evaluations
tions as is perceived similarity of the other person"
of
(1983: 130). The preponderance of relevant empir subordinates' performance.
ical work suggests that perceived supervisor-subor
dinate dissimilarity is associated with unfavorable
Mediating Effects of Relationship Conflict
outcomes, such as lower levels of job satisfaction
In a review and analysis of the workplace diver
(Turban & Jones, 1988), lower-quality relationships
with supervisors (Liden, Wayne, & Stilwell,sity literature, Harrison and Klein (2007) argued
1993;
Wayne & Liden, 1995), and lower performance that the effects of diversity depend on diversity
evaluations (Huang & Iun, 2006; Liden et al., type.
1993;They argue that "separation on an attribute,"
Turban & Jones, 1988; Wayne & Liden, 1995). a form of diversity that captures deep-level differ
ences,
Could perceived deep-level dissimilarity also beis associated with negative outcomes includ
associated with supervisory mistreatment of ingsubor distrust, reduced cohesiveness, and conflict.
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
Perceived deep-level dissimilarity may increase the pathways by which perceived deep-level dissimi
risk of relationship conflict because dissimilar peo larity leads to abusive supervision, we conceptual
ple are less likely to validate their counterparts' ize relationship conflict as a partial mediator of
beliefs and values (Byrne, 1971) and because per the effect of perceived deep-level dissimilarity.
ceived dissimilarity can lead to fundamental differ Figure 1, our hypothesized model, shows these
ences of opinion vis-a-vis important work-related relationships.
tasks and goals (Harrison & Klein, 2007; Hobman &
Bordia, 2006). In keeping with these arguments, Hypothesis 1. A supervisor's perceptions of re
research exploring relationship conflict in groups lationship conflict partially mediate the rela
suggests that perceived deep-level dissimilarity is tionship between the supervisor's perceived
associated with higher levels of relationship con deep-level dissimilarity with a subordinate and
flict (Jehn & Mannix, 2001; Mohammed & Angell, abusive supervision directed toward that
subordinate.
2004). We predict that a similar effect occurs at the
supervisor-subordinate dyad level: supervisors will
perceive that the relationship with a focal subordi
Mediating Effects of Supervisor Evaluations of
nate involves greater interpersonal conflict when Subordinate Performance
they perceive the subordinate to be dissimilar. In
direct support for this prediction comes from stud We noted earlier that several studies have docu
mented a relationship between supervisor percep
ies suggesting that supervisors are more attracted to
and form higher-quality relationships with subor tions of deep-level dissimilarity with subordinates
dinates they perceive to be similar to themselves and supervisor evaluations of subordinates' perfor
(Liden et al., 1993; Schaubroeck & Lam, 2002) mance; supervisors evaluate subordinates who are
and that relationship conflict is lower in higherperceived to be similar to themselves more highly
quality supervisor-subordinate relationships (Tur (Liden et al., 1993; Pulakos & Wexley, 1983; Turban
ban, Dougherty, & Lee, 2002). & Jones, 1988). Similar subordinates are likely to be
Perceived relationship conflict with a subordi evaluated more favorably than their dissimilar
nate will, in turn, be associated with hostility to counterparts for two reasons. First, in line with the
ward the subordinate. According to moral exclu similarity-attraction hypothesis (e.g., Byrne, 1971),
sion theory (Opotow, 1990a, 1995), a person individuals are attracted to and render favorable
involved in conflict develops very different atti evaluations of similar others because the attitudes
tudes toward those on the same side of the conflict of similar others validate the individuals' beliefs.
(i.e., allies) and those on the opposing side (i.e., Supervisors evaluate similar subordinates more fa
adversaries). Whereas allies are included in the vorably because doing so is esteem-enhancing. Sec
person's scope of justice and are afforded fair treat ond, according to social identity and moral exclu
ment, adversaries are excluded from the person's sion theory, similar subordinates are likely to be
scope of justice and become probable targets for perceived as more deserving of fair outcomes and
exclusionary practices. We theorize that in super rewards. In contrast, dissimilar subordinates are
visor-subordinate dyads this nexus plays out in an likely to trigger competitive and discriminatory
association between relationship conflict and abu cognitions along with the perception that they are
sive supervision: supervisors position adversarial less deserving of positive outcomes (Clayton &
subordinates beyond the scope of justice, which is Opotow, 2003; Hogg & Abrams, 1988; Turner,
associated with exclusionary practices in the form 1982).
of abusive supervision. The link between relation Subordinates who are perceived as having lower
ship conflict and abusive supervision is also con performance (hence, "lower performers"), in turn
sistent with studies suggesting that exposure to in become more likely targets for supervisory hostil
terpersonal conflict increases "state negative ity. From a moral exclusion perspective, supervi
affect," which translates into hostile actions (e.g., sors perceive lower performers to be potentially
Bruk-Lee & Spector, 2006; Spector, Fox, Penney, harmful and threatening and to thus have low util
Bruursema, Goh, & Kessler, 2006). ity. Empirical evidence from the leadership litera
Summarizing these arguments, we predict that ture supports this thesis: lower performers are more
the relationship conflict that perceived deep-level likely to make supervisors look bad, interfere with
dissimilarity evokes is related to interpersonal hos their capacity to accomplish their work, and take
tility in the form of abusive supervision. That is, up more of their time addressing the fallout poor
relationship conflict mediates the effect of per performance causes (Bass, 1990). The low utility of
ceived deep-level dissimilarity on abusive supervi subordinates who are perceived to be lower per
sion. Furthermore, because we envision multiple formers positions them beyond their supervisors'
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
to reconcile
scopes of justice, which puts them a conflictive
at risk relationship with good
of exclu
sionary practices such as performance.
abusive Thus, supervisors are likely to per
supervision.
ceive employees with
The link between performance whom they have conflict
evaluations and as
being
abusive supervision is also poor performers. Inwith
consistent addition, based
the on vic
the
work suggesting
tim precipitation literature, which that conflictive
examines relationships
themo
individual characteristics and
tivate people behaviors that
to injure their counterparts put
(Struch &
people at risk of becomingSchwartz, 1989), it can be of
the targets argued that supervisors
aggressive
use performance
and hostile responses (Elias, 1986).evaluation processes to injure
Potential ag
subordinates with whom
gressors choose targets strategically, they are in conflict.
focusing their We
hostility on people who therefore
seem expect difficult
that the relationship conflict
to like that
and/or those who appearperceived
to be deep-level dissimilarity evokes
vulnerable and will,
unin
able to defend themselves.turn,Olweus's
be associated with lower evaluations
(1978) conceptof sub
ordinate performance.
of provocative victims explains why Integrating these arguments
supervisors
with those suggesting
might target subordinates with low performance. a relationship between su
pervisor evaluations
According to Olweus, people become good targets of subordinate performance
and abusiveare
for victimization when they supervision (i.e., the conceptual
perceived to basis
be
for Hypothesis 2), we
difficult to work with. In the context of supervisorpropose that the lower per
subordinate relationships,formance evaluations relationship
supervisors shouldconflict evokes
per
are associated with
ceive subordinates who are lower performers to abusive supervision. This
beline
of reasoning is depicted
frustrating, aggravating, and annoying—character in Figure 1 as a mediation
chain
istics that align well with in which
the perceived deep-level dissimilarity
provocative victim
is positivelythat
profile. Hence, the perception associated employees
with relationship conflict;
are
lower performers has the relationship conflict is negatively
potential to evoke related to super
vic
timization in the form of visor evaluations
abusive of subordinate
supervision. performance; and
supervisor evaluations
Integrating these arguments with ofthose subordinate suggest
performance
are, inperceived
ing a relationship between turn, negatively related to abusive supervi
deep-level
sion. In the
dissimilarity and supervisor parlance of Kenny, Kashy,
evaluations ofand Bolger
sub
(1998) us
ordinate performance leads and Fletcher (2006), relationship
to propose thatconflict
the is
unfavorable performance a distalevaluations
mediator and subordinatethatperformance
per is
a proximal mediator
ceived deep-level dissimilarity in the relationship
evokes between
is associ
ated with subordinates'perceived deep-level dissimilarity
perceptions and abusive
of abusive
supervision. supervision.
Hypothesis
Hypothesis 2. A supervisor's 3. Relationship conflictof
perceptions is a distal
a
partial mediator
subordinate's performance partially mediate and supervisor evaluation
the relationship between of subordinate
the performance is a per
supervisor's proximal
partial mediator of
ceived deep-level dissimilarity with the suborthe relationship between
perceived deep-level
dinate and abusive supervision dissimilarity and
directed toabu
ward that subordinate. sive supervision.
Although researchers have failed to link supervi Moderating Effects of Subordinate Performance
sor-subordinate relationship conflict with supervi
sor perceptions of subordinate performance, extant In the conceptual frame associated with Hypoth
research suggests that subordinates in higher-quality esis 1, we argue that supervisors exclude from their
working relationships with their supervisors are scope of justice subordinates they perceive to be
evaluated more favorably than those who are in adversaries—those with whom they have relation
volved in lower-quality relationships with their su ship conflict. This dynamic, in turn, is associated
pervisors (e.g., Huang & Iun, 2006; Judge & Ferris, with hostility in the form of abusive supervision.
1993; Kolodinsky, Treadway, & Ferris, 2007). The Here we qualify that prediction by proposing that
rationale for this link is that lower-quality working some subordinate adversaries are included in a su
relationships involve more conflictive interaction, pervisor's scope of justice and therefore do not
which colors individuals' perceptions of their become targets of abusive supervision. Specifically,
counterparts' capability and effectiveness. In su we argue that subordinate adversaries are per
ceived to deserve fair treatment when they are good
pervisor-subordinate relationships characterized
by higher levels of relationship conflict, supervi ("higher") performers.
sors should be more likely to "see" performance Supervisors should process relationship conflict
problems because of "negative halo" and inability with subordinates differently when they respect
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
their ability and competence. Adversaries who are through e-mail notifications, announcements
perceived to be capable should evoke more respect staff meetings, and/or the distribution of fly
than adversaries who appear less competent. This Once volunteers were identified, we matched each
is because parties to relationship conflict are more supervisor with a volunteering subordinate. Wh
likely to perceive competent adversaries' positions multiple subordinates matched with a supervis
as deriving from reasoned and thoughtful processes we chose one participating subordinate random
(Bush & Folger, 1994). Indeed, one way of inducing Participants completed surveys during work hour
cooperation between adversaries is to encourage on two occasions separated by six weeks. One
one party's perception that the other party is capa the authors was present to explain the purpose
ble and worthy (Ting-Toomey, 1999). We do not the study and to administer surveys. In the ra
mean to imply that supervisors will have no ani cases in which no author was available during
mus toward adversarial subordinates who perform mal survey administration, a contact in the org
well, only that they will view those subordinates as ization assisted in administering the survey.
more deserving of fair treatment than low perform Supervisors and subordinates completed the s
ers subordinate adversaries. Hence, when relation veys at separate times in separate locations. Befor
ship conflict is higher, supervisors are more likely administering the survey, we told each group tha
to execute exclusionary acts such as abusive super we were conducting a study on various aspects
vision against low-performing subordinates. supervisor-subordinate relationships. Superviso
These arguments imply that the mediated effect were told that one of their subordinates would be
captured in Hypothesis 1 varies over levels of su participating, and subordinates were told that their
pervisor perceptions of subordinate performance. supervisor would be participating. Although each
The positive relationship between perceived rela knew the identity of the other, all participants were
tionship conflict (which perceived deep-level dis assured that their individual results would not be
similarity evokes) and abusive supervision is stron shared with their supervisors or anyone else in
ger when supervisors perceive subordinates to be their organization. They were told that supervisors
lower performers. Predictions of this sort are re and subordinates from multiple health care organ
ferred to as second-stage moderation models in izations would be participating and that we would
which a mediated effect varies over levels of a use only aggregated data.
moderator that operates at the second stage of the These data collection procedures produced us
mediated relationship (Edwards & Lambert, 2007). able data from 183 independent supervisor-subor
Stated formally: dinate dyads. Across the seven organizations from
which we collected data, the participation rate of
Hypothesis 4. Supervisor perceptions of subor
those in supervisory positions ranged from 42 to 95
dinate performance moderate the indirect ef
percent, with an average of 61 percent. The super
fect of perceived deep-level dissimilarity on
visors' average age was 46 years; 69 percent were
abusive supervision (through relationship con
women; 28 percent identified themselves as white;
flict); the mediated effect is stronger when a
29 percent, as Hispanic; 39 percent, as black; 3
supervisor perceives a subordinate as having
lower performance. percent, as Asian; and 1 percent, as "other." They
averaged 8.67 years working in the organization
and were employed as physicians, nursing super
METHODS
visors, technical department managers (e.g., imag
Participants and Procedures ing, lab services), physical plant supervisors, de
partment directors, and higher-level hospital
Participants were recruited from seven health administrators. "Executive/upper management po
care organizations, including hospitals, long-term
sitions" were reported by 34 percent; 45 percent
care facilities, and outpatient facilities, located
werein in "middle management"; and 21 percent
the southeastern United States. We contacted heldkey
"lower-level" supervisory positions.
individuals with administrative responsibilities The subordinates' average age was 44 years; 77
(e.g., unit directors, directors of human resources,
percent were women; 17 percent identified them
and chief operating officers) and asked them for
selves as white; 39 percent, as Hispanic; 39 percent,
assistance in identifying and recruiting supervisor
as black; 4 percent, as Asian; and less than 2 per
subordinate dyads in their organizations. cent, as "other." Their average tenure in the or
Though the procedures varied slightly across or
ganization was 8.45 years, and they averaged 4.33
ganizations, the overall data collection strategy was of working for their supervisor. They occu
years
as follows: Volunteers from the ranks of bothpied mana variety of positions, including nurse, tech
agerial and nonmanagerial employees were sought nician, food service employee, physical plant em
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
and the specific terms we examined to test Hypoth each other, with subordinate performance evalua
eses 1 through 4. The examination of mediated and tions, or with abusive supervision. We therefore
moderated effects requires estimating product concluded that it was acceptable to test the hypoth
terms, which are not normally distributed (Shrout eses using OLS regression.
& Bolger, 2002). Hence, we tested all hypothesized Table 1 shows descriptive statistics and correla
effects by constructing bias-corrected confidence tions among the study variables. Perceived dissim
intervals from 10,000 bootstrapped samples ilarity correlated positively with relationship con
(Mooney & Duval, 1993; Shrout & Bolger, 2002). flict [r = .40, p < .01) and abusive supervision (r =
.29, p < .01) and negatively with subordinate per
RESULTS formance evaluations (r = -.29, p < .01); relation
ship conflict correlated negatively with subordi
We conducted a preliminary analysis tonate assess
performance evaluations (r = —.31, p < .01)
differences on the substantive variables among and the
positively with abusive supervision (r = .37,
seven organizations in our study. One-way p < analy
.01); and subordinate performance evaluations
ses of variance suggested no mean differences for negatively with abusive supervision (r =
correlated
perceived deep-level dissimilarity (F[6, -.39, 170]p = < .01). None of the control variables were
0.76, n.s.), perceived relationship conflict (F[6,
related to perceived dissimilarity, relationship con
170] = 0.41, n.s.), supervisor evaluations offlict, suborperformance evaluations, or abusive supervi
dinate performance (F[6, 170] = 1.34, n.s.), sion. and
Indeed, including the controls in the regres
abusive supervision (F[6,170] = 1.51, n.s.). We sion also
analyses did not affect the results. We
computed intraclass coefficients for the four sub followed Becker's (2005) recommenda
therefore
stantive variables to determine whether it was more tion to report the results excluding the control
appropriate to analyze the data using ordinary least variables.
squares (OLS) regression or a multilevel modeling We centered all variables at their grand means
technique that can account for nonindependence before evaluating the regression equations. Table 2
among observations. The ICCls were .00 for per shows the unstandardized regression results. The
ceived deep-level dissimilarity and perceived rela equations explained significant variance in rela
tionship conflict, .02 for subordinates' performance tionship conflict (F[l, 175] = 35.49, R2 = .17, p <
evaluations, and .03 for abusive supervision. Be .01), subordinate performance (F[2, 174] = 14.80,
cause no variance in perceived deep-level dissimi R2 = .15, p < .01), and abusive supervision (F[4,
larity or relationship conflict resided between 172] = 35.49, R2 = .37, p < .01). The signs on the
groups, we could assume that nesting within organ unstandardized regression coefficients suggest that
izations had no effect on their relationships with perceived dissimilarity positively predicted rela
TABLE 1
1. Sex similarity
0.72 0.45
2. Age difference
9.58 8.67 -.04
3. Race similarity 0.50
0.57 .01 -.02
4. Relationship4.46 5.07 .19** -.11 .00
length
5. Organization 1 0.08 0.27 .04 .12 .09 -.17*
6. Organization 2
0.10 0.29 .11 .05 .01 -.04 -.10
7. Organization0.03 3 0.18 -.09 .01 -.03 -.10 -.06 -.06
8. Organization0.12 4 0.32 .10 .04 -.07 .12 -.11 -.12 -.07
9. Organization 0.135 0.33 .16* .17* .02 .02 -.05 -.12 -.07 -.14
10. Organization 0.466 0.50 -.19* -.25** -.06 .11 -.27** -.30** -.17* -.34** -.35**
11. Organization0.09 7 0.29 -.07 .03 .08 -.11 -.09 -.10 -.06 -.12 -.12 -.29**
12. Dissimilarity 3.52 1.75 .14 .13 -.02 .08 -.05 .08 -.01 -.03 .07 .02 -.11 (.96)
13. Relationship 1.42 0.64 .02 -.09 .07 .11 .05 .04 -.03 -.08 -.09 .08 -.02 .40** (.81)
conflict
14. Performance 5.61 1.00 -.14 -.01 .09 .09 -.07 .02 -.08 .04 -.13 .07 .08 -.29** -.31* (.87)
15. Abusive 1.27 0.47 .00 .06 .08 -.09 .03 -.11 .01 -.04 .09 .08 .29**
-.10 .37* -.39** (.96)
supervision
a Sample sizes on the correlations range from 143 to 178. At time 1, supervisors completed the measures of perceived dissimilarity w
subordinate and perceived relationship conflict with subordinate. At time 2, supervisors completed the measure of subordin
performance and subordinates completed the measure of abusive supervision.
Alpha internal consistency reliability coefficients appear on the main diagonal.
* p < .05
** p < .01
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
TABLE 2 TABLE 3
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
FIGURE 2
1.2
1.1
1 !
-1 s.d. +1 s.d.
the question, Why do supervisors abuse specific performance. Second, the relationships between re
subordinates? lationship conflict and abusive supervision and be
tween subordinate performance and abusive super
vision, although consistent with a moral exclusion
Implications for Theory and Research
argument, are also consistent with other contribu
Our findings suggest a causal framework in tions to the management and psychology litera
which perceived deep-level dissimilarity evokes tures. The interpersonal conflict literature suggests
perceived relationship conflict, which produces that people develop negative affect toward adver
lower evaluations of subordinate performance, saries, which translates into hostility. The victim
which, in turn, lead to higher levels of abusive precipitation literature supports the link between
supervision. Our findings also suggest that relation subordinate performance and abusive supervision.
ship conflict mediates the effect of perceived deep Subordinates who are perceived to be lower per
level dissimilarity and abusive supervision, but formers may become targets of abusive supervision
only when supervisors perceive subordinates as because they fit the provocative victim profile—
having low performance. The links between subor annoying, aggravating, and difficult to work with
dinate performance evaluations and abusive super (Olweus, 1978).
vision and between perceived relationship conflict Two features of our results may be regarded as
and abusive supervision (when subordinate perfor unexpected and have implications for theory and
mance is low) are consistent with moral exclusion research. First, we found that even after we ac
theory. From a moral exclusion perspective, per counted for the effects of supervisor perceptions of
ceived relationship conflict and low subordinate relationship conflict and subordinate performance,
performance put referent subordinates beyond a and their interaction, supervisor perceptions of
focal supervisor's scope of justice, which translates deep-level dissimilarity remained a significant pre
into an exclusionary practice: abusive supervision. dictor of abusive supervision (see Table 2). Hence,
However, although the predictors of abusive su perceived deep-level dissimilarity had both direct
pervision we examined were restricted to analogs and indirect effects on abusive supervision. This
for the key constructs that moral exclusion theory finding is consistent with moral exclusion theory
specifies (i.e., dissimilarity, conflict, and utility), and suggests the need for additional research that
our model and arguments augment extant moral examines other possible mediating mechanisms,
exclusion research in two general ways. First, including scope of justice. We return to this in our
drawing on contributions to the literature on diver summary of the study's limitations.
sity in supervisor-subordinate relationships, we The second unexpected finding was that Hypoth
proposed that perceived deep-level dissimilarity is esis 1 was not supported; relationship conflict did
related to abusive supervision indirectly, through not mediate the relationship between perceived
its effects on supervisor perceptions of relationship deep-level dissimilarity and abusive supervision.
conflict and supervisor perceptions of subordinate Of course, the results of subsequent analyses ex
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
taneously investigate the relative predictive fects of the predictors studied here as well as the
power of models that depict abusive supervision effects of other potential exogenous influences on
as the result of mistreatment perpetrated against abusive supervision, such as personality and devel
supervisors and models rooted in the notion that opmental problems (Evans, 1996; Hoobler & Brass,
abusive supervisors are themselves perpetrators 2006). Such research would also permit examina
of mistreatment. tion of an emerging debate in the literature as to
A fifth limitation has to do with our measure of whether these cognitive justifications for mistreat
deep-level dissimilarity. We used a general mea ment are the causes, the consequences, or the
sure drawn from previous studies of perceived causes and consequences of such behavior
deep-level dissimilarity in supervisor-subordinate (Trevino, Weaver, & Reynolds, 2006). Opotow
relationships (Liden et al., 1993; Turban & Jones, (1990a, 1995) is clear in positioning moral exclu
1988). The items comprising this scale capture sion as an antecedent of exclusionary practices, but
deep-level dissimilarity that is relevant to supervi conceivably moral exclusion captures perpetrators'
sor-subordinate relationships but, as Harrison et al. post hoc explanations for their mistreatment of ref
(1998) pointed out, deep dissimilarity can manifest erent others ("If I abuse my subordinates, I must
along a number of work-related values, attitudes, think that they deserve unfair treatment").
and personality traits. Our use of a general measure
seemed appropriate for our research objective of
Practical Implications
initially examining the role that perceived deep
level dissimilarity plays in explaining abusive su As workforce composition becomes increasingly
pervision. In future work, it would be fruitful to diverse, supervisors will more frequently have di
take a finer-grained approach by exploring the di rect reports whose values and attitudes differ from
rect, indirect, and moderated effects of specific their own (Mohammed & Angell, 2004). Our re
deep-level differences on abusive supervision. search suggests that the resulting perceived deep
A final limitation is that we failed to measure level dissimilarity may be directly and indirectly
moral exclusion, which constitutes the proximal related to abusive supervision. A more sanguine
link in a moral exclusion explanation for abusive implication of our research is that it may be possi
supervision. Instead, we interpreted reports of ble an to reduce the occurrence of abusive supervision
exclusionary practice (i.e., abusive supervision) by as refraining from hiring individuals for manage
evidence that subordinate targets had been morally rial positions who are dispositional^ inclined to
excluded from supervisors' scope of justice. Our have a narrow scope of justice and to execute hos
approach is consistent with extant published re tile acts. A relevant individual difference is trait
search on moral exclusion, which has yet to opera empathy, the dispositional tendency to take the
tionalize moral exclusion directly (Hafer & Olson, perspective of others and to recognize and experi
2003). As we said above, several links we tested are ence concern for others' thoughts and feelings
consistent with ideas forwarded in other theories (Eisenberg & Miller, 1987). A recent study suggests
(e.g., the relationship between supervisor percepthat low-empathy workers were more likely to per
tions of subordinate performance and abusive suceive that others deserve harmful treatment (Detert,
pervision is consistent with victim precipitationTrevino, & Sweitzer, 2008), and a large body of
theory), but a logical direction for future research isempirical work suggests that low-empathy individ
to measure and model the cognitive rationales thatuals are more hostile than those who are higher in
are associated with moral exclusion to determine to trait empathy (Miller & Eisenberg, 1988). By select
what extent moral exclusion explains the effects ing personnel for trait empathy, organizations may
observed here. be able to reduce the pool of managers who are
A promising foundation for such work is Ban inclined to put subordinates beyond the scope of
dura and colleagues' theory and research on moral justice and to abuse them.
disengagement, which addresses how people deac Of course, it may not always be possible to use
tivate the self-regulatory mechanisms that ordinar trait empathy as a selection tool. In such cases,
ily inhibit them from performing antisocial acts organizations should discourage supervisors from
(Bandura, 1990, 1999; Bandura, Barbanelli, Ca perceiving that subordinates fall outside their
prara, & Pastorelli, 1996). The mechanisms of moral scope of justice and should promote skills for work
disengagement Bandura (1999) has identified in ing constructively with morally excluded subordi
clude such mental constructions as construing nates. Organizations should broaden supervisors'
harm as serving a righteous purpose, and dehuman scope of justice by endorsing concepts of plural
izing victims. Future research should explore the ism—the notion that there is value in promoting
possibility that these mechanisms explain the ef tolerance and appreciation for divergent values
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
dermining in the workplace. Academy of Manage Hogg, M. A., & Abrams, D. 1988. Social identific
ment Journal, 45: 331-351. A social psychology of intergroup relations
group processes. London: Routledge.
Dupre, K. E., Inness, M., Connelly, C. E., Barling, J., &
Hoption, C. 2006. Workplace aggression in teenage Homan, A. C., Hollenbeck, J., Humphrey, S., va
part-time employees. Journal of Applied Psychol penberg, D., Ilgen, D., & Van Kleef, G. A. 200
ogy, 91: 987-997. ing differences with an open mind: Openne
Edwards, J. R., & Lambert, L. S. 2007. Methods for inte experience, salience of intragroup difference
grating moderation and mediation: A general analyt performance of diverse workgroups. Academ
ical framework using moderated path analysis. Psy Management Journal, 51: 1204-1222.
chological Methods, 12: 1-22. Homan, A. C., van Knippenberg, D., Van Kleef, G.
Eisenberg, N., & Miller, P. A. 1987. The relation of em De Dreu, C. K. W. 2007. Bridging faultlines by
pathy to prosocial and related behaviors. Psycholog ing diversity: The effects of diversity beliefs o
ical Bulletin, 101: 91-119. formation elaboration and performance in di
work groups. Journal of Applied Psychology
Elias, R. 1986. The politics of victimization: Victims, 1189-1199.
victimology, and human rights. New York: Oxford
University Press. Hoobler, J., & Brass, D. 2006. Abusive supervis
family undermining as displaced aggression
Evans, P. 1996. The verbally abusive relationship: How nal of Applied Psychology, 91: 1125-1133.
to recognize it and how to respond. Avon, MA:
Adams Media. Huang, X., & Iun, J. 2006. The impact of subo
supervisor similarity in growth-need stren
Fletcher, T. 2006. Methods and approaches to assessing work outcomes: The mediating role of per
distal mediation. Paper presented at the annual
similarity. Journal of Organizational Behavi
meeting of the Academy of Management, Atlanta. 1121-1148.
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
2011 Tepper, Moss, and Duffy
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms
Academy of Management Journal
nate similarity: Types, effects and mechanisms. Y = c0 + caX + c2(a0 + a^X + eM) + c3(b0 + baX+b2[a0
Journal of Applied Psychology, 73: 228-234.
+ ajX + eM] + ez) + c4(a0 + a 1X+eM)Z + ey. (4)
Turner, J. C. 1982. Towards a cognitive redefinition of the
social group. In H. Tajfel (Ed.), Social identity and Rearranging the terms in Equation 4 produces the fol
intergroup relations: 15-45. Cambridge, U.K.: Cam lowing reduced-form equation:
bridge University Press.
Y= c0 + c2a0 + c3b0 + c3b2a0 + c4a0Z + + c^
Wayne, S. J., & Liden, R. C. 1995. Effects of impression
management on performance ratings: A longitudinal + c3b! + c3b2a1 + c4ajZ) + c2eM+c3b2eM + c3ez + c4eM
study. Academy of Management Journal, 38: 232
260. + ey. (5)
Williams, K. Y., & O'Reilly, C. A. 1998. Demography and Equation 5 shows that the effect of perceived dissimi
diversity in organizations: A review of 40 years of larity (X) on abusive supervision [Y] consists of a direct
research. In B. M. Staw (Ed.), Research in organizaeffect (Cj), an indirect effect through relationship conflict
tional behavior, vol. 20: 77-140. Greenwich, CT: JAI (023!), an indirect effect through subordinates' perfor
Press. mance (CabJ, an indirect effect through relationship con
flict and subordinate performance (c3b2a1), and an indi
rect effect through relationship conflict that is moderated
APPENDIX by subordinate performance (c4a1Z). The examination of
the four indirect effects in Equation 5 corresponds with
Estimation of Regression Equations forthe tests of Hypotheses 1, 2, 3, and 4, respectively.
Hypothesis Tests
Ha
The first equation estimates the direct effect of per
ceived dissimilarity on relationship conflict:
Bennett J. Tepper (btepper@gsu.edu) is a professor of
M = a0 + arX + eM, (1) managerial sciences in the J. Mack Robinson College of
Business at Georgia State University. He received his
where M is relationship conflict and Ph.D.Xin organizational
is perceived psychology from the University
of Miami. His current research interests include leader
dissimilarity.
The second equation estimates the direct effects of ship, behavioral ethics, and psychological well-being.
perceived dissimilarity (X) and relationship conflict [M] Sherry E. Moss (sherry.moss@mba.wfu.edu) is an associ
on subordinates' performance (Z): ate professor of organizational studies in the School of
Business at Wake Forest University. She received her
Z ~ b0 + b-[X -t- b2M + ez. (2)
Ph.D. in organizational behavior and theory from Florida
State University. Her research interests include attribu
The third equation estimates the direct effects of per
tion theory, feedback exchanges and supervisor-subordi
ceived dissimilarity (X), relationship conflict (M), and
nate relations.
subordinates' performance (Z), as well as the interaction
between relationship conflict and subordinate perfor
Michelle K. Duffy (duffylll@umn.edu) is a professor in
mance (MZ), on abusive supervision (Y). the Carlson School of Management at the University of
Minnesota. She received her Ph.D. from the University of
Y = c0 + CjX + c2M + c3Z + c4MZ + eY■ (3) Arkansas. Her research interests include social under
mining, moral disengagement, and affect and emotions at
Substituting Equation 1 for M in Equations 2 and 3 and work.
substituting Equation 2 for Z for in Equation 3 produces
Equation 4: M
This content downloaded from 111.68.97.226 on Sun, 12 Mar 2017 10:56:03 UTC
All use subject to http://about.jstor.org/terms