Sei sulla pagina 1di 17

PREDICTORS OF ABUSIVE SUPERVISION: SUPERVISOR PERCEPTIONS OF DEEP-LEVEL

DISSIMILARITY, RELATIONSHIP CONFLICT, AND SUBORDINATE PERFORMANCE


Author(s): BENNETT J. TEPPER, SHERRY E. MOSS and MICHELLE K. DUFFY
Source: The Academy of Management Journal, Vol. 54, No. 2 (April 2011), pp. 279-294
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/23045081
Accessed: 12-03-2017 10:56 UTC

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.

PREDICTORS OF ABUSIVE SUPERVISION: SUPERVISOR


PERCEPTIONS OF DEEP-LEVEL DISSIMILARITY,
RELATIONSHIP CONFLICT, AND
SUBORDINATE PERFORMANCE

BENNETT J. TEPPER
Georgia State University

SHERRY E. MOSS
Wake Forest University

MICHELLE K. DUFFY
University of Minnesota

The moral exclusion literature identifies three previously unexamined predictors of


abusive supervision: supervisor perceptions of deep-level dissimilarity, relationship
conflict, and subordinate performance. Invoking theory and research on workplace
diversity, relationship conflict, and victim precipitation, we model the three predictors
as associated with abusive supervision. Path-analytic tests using data collected from
supervisor-subordinate dyads at two time points suggest that supervisor perceptions of
relationship conflict and subordinate performance mediate the relationship between
perceived deep-level dissimilarity and abusive supervision and that relationship con
flict mediates that between perceived deep-level dissimilarity and abusive supervision
when supervisors perceive subordinates as having low performance.

Estimates suggest that more than 13 percent of


investigated the antecedents of abusive supervisio
(i.e., Aryee, Chen, Sun, & Debrah, 2007; Hooble
working people in the United States become targets
Brass, 2006; Tepper et al., 2006). In all three,
of abusive supervision, or nonphysical hostility
perpetrated by employees' immediate superiors
searchers framed abusive supervision as a respo
(Schat, Frone, & Kelloway, 2006). Examplestoof be
supervisor perceptions of mistreatment by th
haviors that fall within the abusive supervision
employer. This research has helped explain w
content domain include undermining, public den
supervisors may be inclined to downward hostilit
igration, and explosive outbursts (Tepper, in2007).
general, but little is known about the reaso
Sustained exposure to abusive supervision supervisors
is asso abuse specific subordinates.
ciated with serious negative outcomes for victims
To explore that question, we invoke concept
and employers, including psychological distress
described in the moral exclusion literature (O
(Tepper, 2000), problem drinking (Bamberger tow, &1990a, 1995), which examines the factors th
Bacharach, 2006), and aggression directed against a
influence whether moral considerations apply
victim's supervisor (Dupre, Inness, Connelly,
specific social targets. According to Opotow, e
Barling, & Hoption, 2006; Inness, Barling, & Turner,
person has a scope of justice, a psychologi
2005), employer (Duffy, Ganster, & Pagon, boundary
2002), separating targets that are perceived
and family (Hoobler & Brass, 2006). These deserving
conse fair treatment and to which moral rul
quences translate into annual losses of an estimated
apply (i.e., those morally included in the scope
$23.8 billion in increased health care costs, work
justice) and targets for which justice concerns
place withdrawal, and lost productivity (Tepper,
perceived to be irrelevant (i.e., those morally
Duffy, Henle, & Lambert, 2006). cluded from the scope of justice). As Opotow a
Far less is known about the conditions that pre put it, "Norms, moral rules, and conce
Weiss
dict the occurrence of abusive supervision (Tepper,
about rights and fairness govern our conduct
2007). Indeed, only three published studiesward havethose inside our scope of justice," [but th
who are morally excluded are perceived to be]
pendable, undeserving, exploitable, and irrelevant
We thank Micki Kacmar and three anonymous review
(2000:
ers for many helpful comments that they provided on 478). Morally excluded targets, in turn,
earlier versions of this article. come likely candidates for "exclusionary p
279

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

ployee, and higher-level employee


Abusive supervision. Also reporting
at time 2, subordi
directly to upper and middle
nates completed management
Tepper's (2000) 15-item measure
supervisors. of abusive supervision. Respondents used a five
point scale ranging from 1, " I cannot remember
him/her ever using this behavior with me," to 5
Measures "He/she uses this behavior very often with me," to
report how often their boss used behaviors such as
Perceived dissimilarity with subordinate. "tells At me my thoughts and feelings are stupid,"
time 1, supervisors completed Turban and Jones's "puts me down in front of others," "ridicules me,"
(1988) three-item measure of perceived similarity and "makes negative comments about me to oth
with subordinate and two additional items written ers." We averaged the item ratings to form total
and used by Liden et al. (1993). The items, prefaced scores for abusive supervision (a = .96).
with the phrase "This subordinate and I. . read Control variables. We explored the viability of
as follows: "are similar in terms of our outlook, several control variables that could provide alter
perspective, and values," "analyze problems in a native explanations for the relationships depicted
similar way," "think alike in terms of coming up in our model. These included surface demographic
with a similar solution for a problem," "are alike in diversity (i.e., sex similarity, age similarity, and
a number of areas," and "see things in much the race similarity), supervisor-subordinate relation
same way" (1 = "strongly disagree," = 7, "strongly ship tenure, and organization. Although research
agree"). We recoded ratings on the items so that ers have shown that surface demographic variables
higher ratings captured higher levels of perceived relate to relational outcomes (e.g., Tsui & O'Reilly,
dissimilarity and averaged the answers on items 1989), substantial evidence suggests that the effects
to form total scores for perceived dissimilarity of surface demography diminish over time as indi
(a = .96). viduals learn more about each other (Harrison,
Relationship conflict. At time 1, supervisors re Price, & Bell, 1998; Harrison, Price, Gavin, & Flo
sponded to three items from Jehn and Mannix rey, 2002). Even though the average relationship
(2001). This scale, typically used to measure intra tenure in our sample was almost 4.5 years, some
group conflict, was adapted for use in this study. manager-subordinate dyads in the sample had
The items were preceded with the instruction, worked together for less time, so it was prudent to
"Please refer to your employee, , when answer control for surface-level demographic characteris
ing the following questions." The three items were, tics as well as relationship tenure. The demo
"How much relationship tension is there between graphic variables were coded as follows: sex simi
you and this employee?"; "How often do you and larity was 0 when a supervisor and subordinate
this employee get angry while working?"; "How were of different sexes and 1 when they were the
much emotional conflict is there between you and same; race similarity was coded 0 for different races
this employee?" (1 = "none," 5 = "a lot"). We and 1 for the same; and age similarity was equal to
averaged the item scores to form total scores for the absolute value of the difference between the
relationship conflict (a = .81). supervisor's age and the subordinate's age. Rela
Subordinate performance. At time 2, the super tionship tenure was the number of years the subor
visors completed a four-item measure of subordi dinate had reported directly to the supervisor.
nates' performance developed by Liden et al. Finally, because our sample included seven organ
(1993). The items and response scales were as fol izations that might have different cultures and ori
lows: "My subordinate is superior to other subor entations to diversity, we also created six dummy
dinates that I've supervised before" (1 = "strongly coded variables and entered these terms in our
disagree," 7 = "strongly agree"); "Rate the overall analyses before exploring the effects of our substan
level of performance that you observe for this tive variables.
subordinate" (1 = "unacceptable," 7 = "out
standing"); "What is your personal view of your
Analytical Strategy
subordinate in terms of his or her overall effec
tiveness?" (1 = "very ineffective," 7 = "very ef We tested our hypotheses using path-analytic
fective"); and "Overall to what extent do you feel procedures for testing mediation models from
your subordinate has been effectively fulfilling MacKinnon, Lockwood, Hoffman, West, and Sheets
his or her roles and responsibilities?" (1 = "not (2002) and Edwards and Lambert's (2007) elabora
effectively at all," 7 = "very effectively"). We tion of those procedures, which allow examination
averaged the item ratings to form total scores for of second-stage moderation models. The Appendix
subordinate performance (a = .87). shows the regression equations that we estimated

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

Descriptive Statistics and Correlations among Study Variables®

Variable Mean s.d. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

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

Path-Analytic Regression Results" Indirect Effects of Perceived Deep-Level Dissimilarity


on Abusive Supervision®
Dependent Variables
Mediator(s) P Hypothesis
Predictors Conflict Performance Abuse
Tests of hypothesized mediated
Perceived dissimilarity .15** -.12* .04* effects
.01 1
Relationship conflict -.40* .09 Relationship conflict
-.11** Subordinate performance .01** 2
Performance
-.17* Relationship conflict & .01* 3
Conflict X performance
subordinate performance
Equation R2 .17** .15** .37**
Test of hypothesized moderated
a n = 177. Tabled values are unstandardized regression coef indirect effect
4
ficients. All parameter estimates were tested for significance Relationship conflict
Low subordinate performance .04**
using bias-corrected bootstrapped confidence intervals.
High subordinate performance -.01
* p < .05
** p < .01
a p refers to the mediated effect. All estimates were tested f
significance using bias-corrected confidence intervals fr
10,000 bootstrapped samples.
tionship conflict [b = .15, p < .01), negatively * p < .05
predicted subordinate performance [b = -.12, p < ** p < .01
.05), and positively predicted abusive supervision
[b = .04, p < .05); relationship conflict negatively
predicted subordinate performance (b = -.40, p < high and low levels of subordinate performance.
.01) and was unrelated to abusive supervision (b = When subordinate performance was high, the me
.09, n.s.); and subordinate performance negatively diated effect of relationship conflict was not signif
predicted abusive supervision (b = —.11, p < .01). icant (p = —.01, n.s.); however, when subordinate
Table 2 also shows that the relationship conflict by
performance was low, relationship conflict medi
subordinate performance cross-product was related ated the effect of perceived deep-level dissimilarity
to abusive supervision [b = -.17, p < .01). on abusive supervision (p = .04, p < .01). Figure 2
We used the information from these equations to shows the plot of the mediated effects at higher and
calculate the mediated effect of perceived deep lower levels of subordinate performance (Aiken &
level dissimilarity on abusive supervision through West, 1991). The form of the interaction is consis
relationship conflict (Hypothesis 1), subordinate tent with Hypothesis 4; the mediated effect of rela
performance (Hypothesis 2), relationship conflict tionship conflict was stronger when subordinate
and subordinate performance (Hypothesis 3), and performance was perceived to be lower.
relationship conflict at higher and lower levels of
subordinate performance (Hypothesis 4). Table 3
presents the estimates associated with each DISCUSSION
hypothesis.
The first row of results in Table 3 shows that the Our research contributes to a growing body o
mediated effect of perceived deep-level dissimilar research exploring abusive supervision in work
ity on abusive supervision through relationship ganizations. The research to date has shed light
conflict was not significant (p = .01, n.s.). Hence, the deleterious consequences of abusive superv
Hypothesis 1 was not supported. The next row of sion, but little is known about the factors that pr
results shows that subordinate performance medi dict when abusive supervision is likely to occu
ated the relationship between perceived deep-level The few studies that have examined the antece
dissimilarity on abusive supervision [p = .01, p < ents of abusive supervision have focused on th
.01). Hence, Hypothesis 2 was supported. The next treatment supervisors receive and suggest that
row in Table 3 shows that Hypothesis 3 was sup pervisor perceptions of organizational injustice and
contract breach are associated with subordinate
ported; the mediated effect of perceived deep-level
dissimilarity on abusive supervision through rela ports of supervisory abuse (Aryee et al., 200
tionship conflict and subordinate performance was Hoobler & Brass, 2006; Tepper et al., 2006). Our
research shifts the focus from the mistreatment
significant (p = .01, p < .05).
The last two rows in Table 3 show the mediated pervisors experience to concepts described in O
effect of perceived deep-level dissimilarity on abu tow's (1990a, 1995) work on moral exclusion. Th
sive supervision through relationship conflict at approach provides a basic framework for answerin

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

Mediated Effect of Perceived Deep-Level Dissimilarity with Subordinate on Abusive Superv


Relationship Conflict) at High and Low Levels of Subordinate Performance
1.5
—High subordinate
performance
1.4 Low subordinate
performance
Abusive ^ ^
Supervision

1.2

1.1

1 !

-1 s.d. +1 s.d.

Perceived Dissimilarity with Subord

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

plain why this may have mance on changes in abusive


occurred. Thesupervision.
test It is of Hy
pothesis 3 suggested that therefore appropriate to concludeconflict
relationship that our study wa
a distal mediator of the effect
results of perceived
are merely suggestive of the causal ordering deep
level dissimilarity on abusive our model implies. A natural follow-up to the which
supervision, re
operated through supervisor search we have evaluations
reported here is to examineof similarsubo
dinate performance. The test
predictions using of Hypothesis
designs in which all variables are 4 sug
gested that relationship measured conflict several times.
played a mediating
role, albeit when subordinates Such designs were would also address a second to
perceived limi be
lower performers. It appears, tation: that we cannot then, entirely rule out the
that modelpossi
linking perceived relationship bility that the results
conflictcan be attributed
and to common
abusiv
supervision should account method bias. for This supervisor
is particularly true for percep
the rela
tions of subordinate performance. tionship between perceived deep-level dissimilar
A final and more general ity and relationship conflict, the
implication ofmeasures
our ofwork which
is that it demonstrates the usefulness of using con were both included in the supervisor time 1 survey.
cepts from the moral exclusion However, it is fair to conclude that common
literature to explain
one of the many forms of antisocial work method bias is an unlikely explanationbehavior
for our me
that researchers have diation explored.
results, given that Similar model
all the analyses relied to
could be developed to explore some degree on data collected
other at different time
manifestations
of antisocial work behavior, such
points and/or fromas rank-and-file
different sources (Podsakoff,
employees' acts of workplace deviance MacKenzie, Lee, & Podsakoff, 2003). Hence, for ex
directe
against their organization, ample, the test of Hypothesis 3 was and
supervisor, based on cowor
the
ers—counternormative actions product of the thatlinks between perceived deep-level
threaten organ
izations and their members (Robinson & Bennett, dissimilarity and relationship conflict (which
1995). In extant work, researchers have invoked could have been influenced by common method
theories of justice, social learning, and personality bias), relationship conflict and subordinate perfor
to explain employees' deviance (e.g., Aquino, mance (which were measured at two time points),
Douglas, & Martinko, 2004; Skarlicki, Folger, & Tes and subordinate performance and abusive supervi
luk, 1999). We take the position that value may sion (which were measured from two perspectives).
exist in modeling the occurrence of workplace de A third limitation is that all data were collected
viance by exploring the factors that put the targets from health care employees, so we must cautiously
of deviant behavior beyond a focal individual's render conclusions regarding the generalizability of
scope of justice. It should be noted that the scope of the results to organizations outside the health care
justice concept is not reserved to human targets; delivery sector. It has been argued that abusive
the framework can be used to explore focal per supervision is particularly common in health care,
sons' attitudes toward any referent that can be where the costs of failure are high, yet workers face
come the target of maltreatment (e.g., an environ heavy workloads and time pressures (Richman,
ment [Clayton & Opotow, 2003]). Hence, Flaherty, & Rospenda, 1996; Richman, Flaherty,
researchers should be able to use moral exclusion Rospenda, & Christensen, 1992). A promising di
concepts to model employees' deviance directed rection for future research is to conduct cross-in
against their organization. dustry comparisons to determine whether the re
sults that emerged here generalize to industries not
sharing these features.
Study Limitations and Recommendations for A fourth limitation is that although we investi
Future Research
gated the viability of a number of control variables,
Our study has limitations that should be acwe were unable to control for all variables that may
knowledged and addressed in future research.be related to abusive supervision. For example, we
First, the nature of our research design requiresdid not control for the tendency to respond in a
caution in the conclusions we draw from the re socially desirable manner, nor did we account for
sults. Although we collected the data in two waves supervisors' justice perceptions and trait hostility,
and from two sources, we failed to measure all although previous research has linked both with
variables at both time points; we therefore cannot abusive supervision (Hoobler & Brass, 2006). Our
rule out the possibility that abusive supervision framework may be viewed as a middle-range theory
caused perceived dissimilarity, relationship con (Moore, Johns, & Pinder, 1980), one that does not
flict, and subordinate performance. In addition, we account for all possible influences but instead mod
cannot examine the effects of perceived dissimilar els how one set of factors predict abusive supervi
ity, relationship conflict, and subordinate perfor sion. An important next step would be to simul

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

style and organizational norms


(Opotow, 1990b). Many organizations as moderators.
have found Jour
nal of Occupational Health
that simply increasing employee diversity does not Psychology, 9: 152
164.
by itself facilitate inclusion; rather, successful in
terventions focus on discovering similarities,
Aryee, S., Chen, Z. X., Sun, de
L., & Debrah, Y. A. 2007.
Antecedents and outcomes
veloping empathy, and identifying ways of abusive supervision:
of capital
Test &
izing on differences (Chavez of a Weisinger,
trickle-down model. Journal
2008).of Applied
In
keeping with these ideas, Psychology,
Homan92: 191-201. et al. (2007)
found that through training,
Bamberger,individuals can
P. A., & Bacharach, S. B. be en
2006. Abusive su
couraged to adopt prodiversity beliefs,
pervision and subordinate and Taking
problem drinking: that
resistance, stress,
this, in turn, leads to healthier and subordinate personality pro
interpersonal into
account. Human Relations, 59: 1-30.
cesses and superior performance.
Even when organizations embrace
Bandura, A. 1990. Selectivepluralism,
activation and disengagesit
uations in which supervisors exclude
ment of moral control. Journal of some
Social Issues,subor
46:
dinates from their scope 27-46.
of justice may occur. In
anticipation of these circumstances,
Bandura, A. 1999. Moral organizations
disengagement in the per
can use justice training to help
ation of supervisors interact
inhumanities. Personality and Social
constructively with all subordinates, particularly
chology Review, 3: 193-209.
those whom the supervisors perceive to be dissim
Bandura, A., Barbanelli, C., Caprara, G., & Pastorelli,
ilar, those with whom they have relationship
1996. Mechanisms of moral disengagement con in
flict, and those they perceive to
exercise be lower
of moral perform
agency. Journal of Personali
ers. Justice training involves coaching supervisors
Social Psychology, 71: 364-374.
to use techniques that reduce psychological dis
Bass, B. M. 1990. Bass and Stogdill's handbook of
tance with employees, promote the perception that
ership (3rd ed.). New York: Free Press.
employees have "voice," and equitably apply deci
Becker, T. 2005. Potential
sion-making criteria (Greenberg, 2006; problems in the statistical
Skarlicki &
trol of variables
Latham, 2005). For example, an organization in organizational research:
may A
tative analysis with recommendations. Organ
set the goal of supervisors meeting with each direct
tional Research Methods, 8: 274-289.
report, one-on-one, at least monthly. Employees
would be invited to use Brewer,
that M. B. 1999. The to
time psychology of prejudice: In
provide feed
group love or outgroup
back and voice concerns. In essence, supervisors hate? Journal of Social Is
sues, 55: 429-444.
would be trained in techniques to recognize and
suspend bias so that theyBruk-Lee,
can V., & overcome
Spector, P. E. 2006. Thethesocial stressors
ten
dency to abuse subordinates counterproductivity
who fall work beyondbehaviors link: Are con
their
scope of justice. flicts with supervisors and coworkers the same?
Journal of Occupational Health Psychology, 11:
145-156.
Conclusion
Bush, R., & Folger, J. 1994. The promise of media
Despite its low base rate, abusive supervision is a Responding to conflict through empowerment
costly workplace phenomenon in terms of lost pro recognition. San Francisco: Jossey-Bass.
ductivity, absenteeism, turnover, and health careByrne, D. 1971. The attraction paradigm. New Y
expenditures. It is therefore important that re Academic.
searchers continue to investigate the antecedentChavez, C. I., & Weisinger, J. Y. 2008. Beyond diversity
conditions associated with abusive supervision. training: A social infusion for cultural inclusion.
Our findings are significant because they draw at Human Resource Management, 47: 331-350.
tention to previously unexamined antecedents of
Clayton, S., & Opotow, S. 2003. Justice and identity:
abusive supervision and provide the bases for prac Changing perspectives on what is fair. Personality
tical interventions that have the potential to curb and Social Psychology Review, 7: 298-310.
the frequency of abusive supervision.
Detert, J. R., Trevino, L. K., & Sweitzer, V. L. 2008. Moral
disengagement in ethical decision making: A study
REFERENCES of antecedents and outcomes. Journal of Applied
Psychology, 93: 374-391.
Aiken, L. S., & West, S. G. 1991. Multiple regression:
Testing and interpreting interactions. Thousand Deutsch, M. 1973. The resolution of conflict: Construc
Oaks, CA: Sage. tive and destructive processes. New Haven, CT:
Yale University Press.
Aquino, K., Douglas, S. C., & Martinko, M. J. 2004. Overt
Duffy, M. K., Ganster, D., & Pagon, M. 2002. Social un
anger in response to victimization: Attributional

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.

Greenberg, J. 2006. Losing sleep over organizational in


Inness, M., Barling, J„ & Turner, N. 2005. Underst
justice: Attenuating insomniac reactions to under
supervisor-targeted aggression: A within-person
payment inequity with supervisory training in inter
tween-jobs design. Journal of Applied Psycho
actional justice. Journal of Applied Psychology, 91: 90: 731-739.
58-69.
Jehn, K. A., & Mannix, E. A. 2001.The dynamic nat
Hafer, C. L., & Olson, J. M. 2003. An analysis of empirical
conflict: A longitudinal study of intragroup co
research in the scope of justice. Personality and
and group performance. Academy of Manag
Social Psychology Review, 4: 311-323.
Journal, 44: 238-251.
Halevy, N., Bornstein, G., & Sagiv, L. 2008. "In-group
Judge, T. A., & Ferris, G. R. 1993. Social context
love" and "out-group hate" as motives for individual
formance evaluation decisions. Academy of
participation in intergroup conflict: A new para
agement Journal, 36: 80-105.
digm. Psychological Science, 19: 405—411.
Kenny, D. A., Kashy, D. A., & Bolger, N. 1998.
Harrison, D. A., & Klein, K. J. 2007. What's the differ
analysis in social psychology. In D. Gilbert,
ence? Diversity constructs as separation, variety, or
Fiske, & G. Lindzey (Eds.), Handbook of socia
disparity in organizations. Academy of Manage
ment Review, 32: 1199-1228. chology (4th ed.), vol. 1: 233-265. New York:
McGraw-Hill.
Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyond
Kolodinsky, R. W., Treadway, D. C., & Ferris, G. R. 2007.
relational demography: Time and the effects of sur
Political skill and influence effectiveness: Testing
face- and deep-level diversity on work group cohe
sion. Academy of Management Journal, 41: 96 portions of an expanded Ferris and Judge (1991)
107.
model. Human Relations, 60: 1747-1777.

Lewin, K. 1951. Field theory in social science. New


Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A.
York: Harper & Row.
2002. Time, teams, and task performance: Changing
effects of surface- and deep-level diversity on group Liden, R. C., Wayne, S. J., & Stilwell, D. 1993. A longitu
functioning. Academy of Management Journal, 45: dinal study on the early development of leader
1029-1045. member exchanges. Journal of Applied Psychology,
78: 662-674.
Hewstone, M., Rubin, M., & Willis, H. 2002. Intergroup
bias. In S. T. Fiske, D. L. Schacter, & C. Zahn-Waxler
MacKinnon, D. P., Lockwood, C. M., Hoffman,
(Eds.), Annual review of psychology, vol. 53: 575 S. G., & Sheets, V. 2002. A compariso
West,
604. Palo Alto, CA: Annual Reviews. methods to test mediation and other interve
Hobman, E. V., & Bordia, P. 2006. The role of team variable effects. Psychological Methods, 7: 83
identification in the dissimilarity-conflict relation Miller, P. A., & Eisenberg, N. 1988. The relation
ship. Group Processes and Intergroup Relations, 9: pathy to aggressive and externalizing/antisoc
483-507.
havior. Psychological Bulletin, 103: 324-344.

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

Mohammed, S., & Angell, L. workplace


C. 2004.violence: 47-89. Thousand
Surface- andOaks,
deepCA:
Sage.
level diversity in workgroups: Examining the mod
erating effects of team orientation and
Schaubroeck, J., & Lam, S. K. 2002 Howteam
similarity to process
on relationship conflict. Journal peers and supervisorof Organizational
influences organizational ad
Behavior, 25: 1015-1039. vancement in different cultures. Academy of Man
Mooney, C. Z., & Duval, R. D. 1993. Bootstrapping: A agement Journal, 45: 1120-1136.
nonparametric approach to statistical inference. Shrout, P. E., & Bolger, N. 2002. Mediation in experimen
Newbury Park, CA: Sage. tal and nonexperimental studies: New procedures
Moore, L., Johns, G., & Pinder, C. C. 1980. Toward middle and recommendations. Psychological Methods, 7:
422-445.
range theory: An overview and perspective. In C. C.
Pinder & L. F. Moore (Eds.), Middle range theory Skarlicki, D. P., Folger, R. G., & Tesluk, P. 1999.
and the study of organizations: 1-16. Hingham, ality as a moderator in the relationship between
MA: Martinus Nijhoff. ness and retaliation. Academy of Management
Olweus, D. 1978. Aggression in schools: Bullies and Journal, 42: 100-108.
whipping boys. Washington, DC: Hemisphere. Skarlicki, D. P., & Latham, G. P. 2005. How can training
Opotow, S. 1990a. Moral exclusion and injustice: An be used to foster organizational justice? In J. Green
introduction. Journal of Social Issues, 46(1): 1-20. berg & J. A. Colquitt (Eds.), Handbook of organiza
tional justice: 499-522. Mahwah, NJ: Erlbaum.
Opotow, S. 1990b. Deterring moral exclusion. Journal of
Social Issues, 46(1): 173-182. Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh,
A., & Kessler, S. 2006. The dimensionality of coun
Opotow, S. 1995. Drawing the line: Social categorization,
terproductivity: Are all counterproductive behaviors
moral exclusion, and the scope of justice. In B. Bun
created equal? Journal of Vocational Behavior, 68:
ker & J. Rubin (Eds.), Conflict, cooperation, and 446-460.
justice: Essays inspired by the work of Morton
Deutsch: 347-369. San Francisco: Jossey-Bass. Struch, N., & Schwartz, S. H. 1989. Intergroup aggr
Its predictors and distinctiveness from in-grou
Opotow, S. 2001. Reconciliation in times of impunity: Journal of Personality 6- Social Psychology
Challenges for social justice. Social Justice Re 364-373.
search, 14: 149-170.
Tajfel, H., & Turner, J. C. 1979. An integrative the
Opotow, S., & Weiss, L. 2000. Denial and the process of intergroup conflict. In W. G. Austin & S. Wo
moral exclusion in environmental conflict. Journal
(Eds.), The social psychology of intergroup
of Social Issues, 56(3): 475-490. tions: 94-109. Monterey, CA: Brooks/Cole.
Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsa
Tepper, B. J. 2000. Consequences of abusive supervi
koff, N. P. 2003. Common method biases in behav
Academy of Management Journal, 43: 178-19
ioral research: A critical review of the literature and
recommended remedies. Journal of Applied Psy Tepper, B. J. 2007. Abusive supervision in work
zations: Review, synthesis, and directions for fu
chology, 88: 879-903.
research. Journal of Management, 33: 261-28
Pulakos, E. D., & Wexley, K. N. 1983. The relationship
Tepper, B. J., Duffy, M. K., Henle, C. A., & Lambert
among perceptual similarity, sex, and performance
2006. Procedural injustice, victim precipitation,
ratings in manager-subordinate dyads. Academy of
abusive supervision. Personnel Psychology
Management Journal, 26: 129-139.
101-123.
Richman, J. A., Flaherty, J. A., & Rospenda, K. M. 1996.
Perceived workplace harassment experiences and Ting-Toomey, S. 1999. Communicating across cult
New York: Guilford.
problem drinking among physicians: Broadening the
stress/alienation paradigm. Addiction, 91: 391-403. Trevino, L., K., Weaver, G. R., & Reynolds, S. K. 2006.
Behavioral ethics in organizations: A review. Jour
Richman, J. A., Flaherty, J. A., Rospenda, K. M., & Chris
tensen, M. 1992. Mental health consequences and nal of Management, 32: 951-990.
Tsui, A. S., & O'Reilly, C. A., Ill 1989. Beyond simple
correlates of medical student abuse. Journal of the
American Medical Association, 267: 692-694. demographic effects: The importance of relational
demography in superior-subordinate dyads. Acad
Robinson, S. L., & Bennett, R. }. 1995. A typology of
emy of Management Journal, 32: 402-423.
workplace deviant behaviors: A multidimensional
scaling study. Academy of Management Journal, Turban, D. B, Dougherty, T. W., & Lee, F. K. 2002. Gen
38: 555-572. der, race and perceived similarity effects in develop
mental relationships: The moderating role of rela
Schat, A. C. H., Frone, M. R., & Kelloway, E. K. 2006.
tionship duration. Journal of Vocational Behavior,
Prevalence of workplace aggression in the U.S. work 61: 240-262.
force: Findings from a national study. In E. K. Kel
Turban,
loway, J. Barling, & J. J. Hurrell (Eds.), Handbook of D. B., & Jones, A. P. 1988. Supervisor-subor

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

Potrebbero piacerti anche