article critique




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 working people in the United States become targets of abusive supervision, or nonphysical hostility perpetrated by employees’ immediate superiors (Schat, Frone, & Kelloway, 2006). Examples of be- haviors that fall within the abusive supervision content domain include undermining, public den- igration, and explosive outbursts (Tepper, 2007). Sustained exposure to abusive supervision is asso- ciated with serious negative outcomes for victims and employers, including psychological distress (Tepper, 2000), problem drinking (Bamberger & Bacharach, 2006), and aggression directed against a victim’s supervisor (Dupre, Inness, Connelly, Barling, & Hoption, 2006; Inness, Barling, & Turner, 2005), employer (Duffy, Ganster, & Pagon, 2002), and family (Hoobler & Brass, 2006). These conse- quences translate into annual losses of an estimated $23.8 billion in increased health care costs, work- place withdrawal, and lost productivity (Tepper, Duffy, Henle, & Lambert, 2006).

Far less is known about the conditions that pre- dict the occurrence of abusive supervision (Tepper, 2007). Indeed, only three published studies have

investigated the antecedents of abusive supervision (i.e., Aryee, Chen, Sun, & Debrah, 2007; Hoobler & Brass, 2006; Tepper et al., 2006). In all three, re- searchers framed abusive supervision as a response to supervisor perceptions of mistreatment by their employer. This research has helped explain why supervisors may be inclined to downward hostility in general, but little is known about the reasons supervisors abuse specific subordinates.

To explore that question, we invoke concepts described in the moral exclusion literature (Opo- tow, 1990a, 1995), which examines the factors that influence whether moral considerations apply to specific social targets. According to Opotow, each person has a scope of justice, a psychological boundary separating targets that are perceived as deserving fair treatment and to which moral rules apply (i.e., those morally included in the scope of justice) and targets for which justice concerns are perceived to be irrelevant (i.e., those morally ex- cluded from the scope of justice). As Opotow and Weiss put it, “Norms, moral rules, and concerns about rights and fairness govern our conduct to- ward those inside our scope of justice,” [but those who are morally excluded are perceived to be] “ex- pendable, undeserving, exploitable, and irrelevant” (2000: 478). Morally excluded targets, in turn, be- come likely candidates for “exclusionary prac-

We thank Micki Kacmar and three anonymous review- ers for many helpful comments that they provided on earlier versions of this article.

� Academy of Management Journal 2011, Vol. 54, No. 2, 279–294.


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.

tices,” various forms of hostility and mistreatment that range from mild (e.g., rudeness [Opotow, 2001]) to severe (e.g., violations of fundamental human rights [Opotow, 1990a]). Of direct relevance to the focus of our research, supervisors execute abusive acts against targets who are morally ex- cluded from the supervisors’ scope of justice (Opotow, 1995). Hence, a promising approach to modeling the abuse of specific subordinates involves incorporating the precursors of moral exclusion.

What factors cause targets to become morally ex- cluded from an agent’s scope of justice and, in turn, to become targets for mistreatment? The moral ex- clusion literature focuses on three recurring themes: (1) perceived dissimilarity to targets, (2) conflict with the targets, and (3) the targets’ useful- ness or “utility” (Hafer & Olson, 2003). An individ- ual becomes a target of hostile behavior when a perpetrator perceives the target to be dissimilar, when the perpetrator is in conflict with the target, and when the target is not useful or is even injuri- ous to the perpetrator. We identified supervisor- subordinate analogs of these three factors to ex- plore the predictors of abusive supervision. The respective analogs for dissimilarity, conflict, and utility are a supervisor’s perceived deep-level dis- similarity with a subordinate, perceived relation- ship conflict with the subordinate, and evaluation of the subordinate’s performance. Perceived deep- level dissimilarity refers to the perception that the subordinate’s values and attitudes differ from the supervisor’s (Harrison & Klein, 2007). Perceived relationship conflict with the subordinate refers to negative social interactions, interpersonal incom- patibility, and negative affect in the form of frustra- tion, irritation, and annoyance (Jehn & Mannix, 2001). Subordinate performance evaluation cap-

tures the supervisor’s perception that the subordi- nate meets performance standards on required or in-role tasks. Because our interest is in predicting supervisors’ mistreatment of subordinates and be- cause people react more strongly to their percep- tion of their environment rather than to its objec- tive features (Lewin, 1951), we focus on supervisor perceptions of dissimilarity, relationship conflict, and subordinate performance.

The moral exclusion theory (Opotow, 1990a, 1995) prediction is that abusive supervision is pos- itively related to perceived deep-level dissimilarity and supervisor perceptions of relationship conflict with subordinates, and negatively related to super- visor evaluations of subordinate performance. However, contributions to three bodies of theory and research suggest a more complicated set of interrelationships and corresponding predictions. These research streams are the workplace diversity literature, which examines the effects of supervi- sor-subordinate dissimilarity on dyadic and indi- vidual attitudes and well-being (Harrison & Klein, 2007; Williams & O’Reilly, 1998); the work on con- flict in interpersonal relations, which examines the emergence and consequences of frustration and negative interaction between coworkers (Jehn & Mannix, 2001); and the victim precipitation litera- ture, which examines the characteristics that put people at risk of mistreatment (Elias, 1986; Olweus, 1978). We invoked and integrated concepts from these literatures to develop and test a new model of abusive supervision. According to our model, which is depicted in Figure 1, supervisors experi- ence relationship conflict with and assign lower performance evaluations to subordinates who are perceived to be dissimilar, which, in turn, is asso- ciated with subordinates’ reports that they have been the target of abusive supervision. The model

FIGURE 1 Hypothesized Model

Perceived Deep-Level Dissimilarity

with Subordinate

Perceived Relationship


Supervisor Evaluations of Subordinates’ Performance

Abusive Supervision

280 AprilAcademy of Management Journal

also specifies that the indirect effect of perceived deep-level dissimilarity through relationship con- flict is stronger when supervisors perceive subor- dinates to be poorer performers.

We contribute to management theory by devel- oping and testing a model of abusive supervision that incorporates factors that have been identified in the moral exclusion literature and processes de- scribed in theory and research on workplace diver- sity, relationship conflict, and victim precipitation. In so doing, we demonstrate the usefulness of using moral exclusion concepts to predict a costly form of antisocial organizational behavior. We also enrich and sharpen moral exclusion theory by introducing fine-grained predictions regarding the roles that dissimilarity, conflict, and utility play in explain- ing supervisors’ exclusionary practices. In the fol- lowing sections, we explain the conceptual bases for the linkages depicted in our hypothesized model. We then present the results of a study in which we tested our predictions using two-wave data collected from matched pairs of supervisors and subordinates.


Perceived Deep-Level Dissimilarity and Abusive Supervision

Extant studies of diversity in supervisor-subordi- nate dyads have examined relationships between dissimilarity and an assortment of attitudinal, rela- tional, and behavioral outcomes. Some of these studies have focused on surface-level dissimilarity, comprising objective differences in age, gender, and race; others have focused on deep-level dissim- ilarity, comprising perceived differences in values, attitudes, and personality. As we noted above, we restrict our analysis to perceived deep-level dissim- ilarity—specifically, the perceptions of supervisors that their attitudes and values differ from those of focal subordinates. As Pulakos and Wexley said, “Actual similarity may not be as important to the process of manager-subordinate interpersonal rela- tions as is perceived similarity of the other person” (1983: 130). The preponderance of relevant empir- ical work suggests that perceived supervisor-subor- dinate dissimilarity is associated with unfavorable outcomes, such as lower levels of job satisfaction (Turban & Jones, 1988), lower-quality relationships with supervisors (Liden, Wayne, & Stilwell, 1993; Wayne & Liden, 1995), and lower performance evaluations (Huang & Iun, 2006; Liden et al., 1993; Turban & Jones, 1988; Wayne & Liden, 1995).

Could perceived deep-level dissimilarity also be associated with supervisory mistreatment of subor-

dinates? The moral exclusion literature supports this very thesis. Opotow invoked social identity theory to argue that “moral exclusion results from our innate tendency to differentiate objects” (1990a: 7). People categorize others as similar or dissimilar and demonstrate favoritism toward sim- ilar others and derogation toward dissimilar others (Brewer, 1999; Hewstone, Rubin, & Willis, 2002; Tajfel & Turner, 1979). Perceived dissimilarity evokes feelings of unconnectedness and indiffer- ence to potential threats to focal others’ well-being (Deutsch, 1973; Williams & O’Reilly, 1998), atti- tudes that align with the concept of moral exclu- sion. Indeed, Opotow (1990a, 1995) argued that perpetrators of hostile and aggressive acts against dissimilar others consciously or unconsciously ex- clude these dissimilar victims from their scope of justice which, in turn, produces exclusionary prac- tices—acts of mistreatment and harm doing. Hence, the moral exclusion literature provides the basis for predicting that supervisors will abuse subordinates they perceive to be dissimilar to themselves.

However, recent contributions to the workplace diversity literature suggest that the role perceived dissimilarity plays in predicting abusive supervi- sion may be less straightforward. This work sug- gests that under some circumstances diversity may be valued rather than derogated (Homan, Hollen- beck, Humphrey, Van Knippenberg, Ilgen, & Van Kleef, 2008; Homan, van Knippenberg, Van Kleef, & De Dreu, 2007) and that greater propensity may exist to promote similar others than to derogate those who are different (Halevy, Bornstein, & Sagiv, 2008). Hence, this literature suggests that perceived deep-level dissimilarity may not necessarily be as- sociated with abusive supervision. In the sections that follow, we reconcile this work with moral ex- clusion theory by proposing that perceived deep- level dissimilarity plays a role in predicting abu- sive supervision, but not precisely the role that moral exclusion theory specifies. Specifically, we propose that the relationship between perceived deep-level dissimilarity and abusive supervision is indirect, operating through supervisor perceptions of relationship conflict and supervisor evaluations of subordinates’ performance.

Mediating Effects of Relationship Conflict

In a review and analysis of the workplace diver- sity literature, Harrison and Klein (2007) argued that the effects of diversity depend on diversity type. They argue that “separation on an attribute,” a form of diversity that captures deep-level differ- ences, is associated with negative outcomes includ- ing distrust, reduced cohesiveness, and conflict.

2011 281Tepper, Moss, and Duffy

Perceived deep-level dissimilarity may increase the risk of relationship conflict because dissimilar peo- ple are less likely to validate their counterparts’ beliefs and values (Byrne, 1971) and because per- ceived dissimilarity can lead to fundamental differ- ences of opinion vis-à-vis important work-related tasks and goals (Harrison & Klein, 2007; Hobman & Bordia, 2006). In keeping with these arguments, research exploring relationship conflict in groups suggests that perceived deep-level dissimilarity is associated with higher levels of relationship con- flict (Jehn & Mannix, 2001; Mohammed & Angell, 2004). We predict that a similar effect occurs at the supervisor-subordinate dyad level: supervisors will perceive that the relationship with a focal subordi- nate involves greater interpersonal conflict when they perceive the subordinate to be dissimilar. In- direct support for this prediction comes from stud- ies suggesting that supervisors are more attracted to and form higher-quality relationships with subor- dinates they perceive to be similar to themselves (Liden et al., 1993; Schaubroeck & Lam, 2002) and that relationship conflict is lower in higher- quality supervisor-subordinate relationships (Tur- ban, Dougherty, & Lee, 2002).

Perceived relationship conflict with a subordi- nate will, in turn, be associated with hostility to- ward the subordinate. According to moral exclu- sion theory (Opotow, 1990a, 1995), a person involved in conflict develops very different atti- tudes toward those on the same side of the conflict (i.e., allies) and those on the opposing side (i.e., adversaries). Whereas allies are included in the person’s scope of justice and are afforded fair treat- ment, adversaries are excluded from the person’s scope of justice and become probable targets for exclusionary practices. We theorize that in super- visor-subordinate dyads this nexus plays out in an association between relationship conflict and abu- sive supervision: supervisors position adversarial subordinates beyond the scope of justice, which is associated with exclusionary practices in the form of abusive supervision. The link between relation- ship conflict and abusive supervision is also con- sistent with studies suggesting that exposure to in- terpersonal conflict increases “state negative affect,” which translates into hostile actions (e.g., Bruk-Lee & Spector, 2006; Spector, Fox, Penney, Bruursema, Goh, & Kessler, 2006).

Summarizing these arguments, we predict that the relationship conflict that perceived deep-level dissimilarity evokes is related to interpersonal hos- tility in the form of abusive supervision. That is, relationship conflict mediates the effect of per- ceived deep-level dissimilarity on abusive supervi- sion. Furthermore, because we envision multiple

pathways by which perceived deep-level dissimi- larity leads to abusive supervision, we conceptual- ize relationship conflict as a partial mediator of the effect of perceived deep-level dissimilarity. Figure 1, our hypothesized model, shows these relationships.

Hypothesis 1. A supervisor’s perceptions of re- lationship conflict partially mediate the rela- tionship between the supervisor’s perceived deep-level dissimilarity with a subordinate and abusive supervision directed toward that subordinate.

Mediating Effects of Supervisor Evaluations of Subordinate Performance

We noted earlier that several studies have docu- mented a relationship between supervisor percep- tions of deep-level dissimilarity with subordinates and supervisor evaluations of subordinates’ perfor- mance; supervisors evaluate subordinates who are perceived to be similar to themselves more highly (Liden et al., 1993; Pulakos & Wexley, 1983; Turban & Jones, 1988). Similar subordinates are likely to be evaluated more favorably than their dissimilar counterparts for two reasons. First, in line with the similarity-attraction hypothesis (e.g., Byrne, 1971), individuals are attracted to and render favorable evaluations of similar others because the attitudes of similar others validate the individuals’ beliefs. Supervisors evaluate similar subordinates more fa- vorably because doing so is esteem-enhancing. Sec- ond, according to social identity and moral exclu- sion theory, similar subordinates are likely to be perceived as more deserving of fair outcomes and rewards. In contrast, dissimilar subordinates are likely to trigger competitive and discriminatory cognitions along with the perception that they are less deserving of positive outcomes (Clayton & Opotow, 2003; Hogg & Abrams, 1988; Turner, 1982).

Subordinates who are perceived as having lower performance (hence, “lower performers”), in turn become more likely targets for supervisory hostil- ity. From a moral exclusion perspective, supervi- sors perceive lower performers to be potentially harmful and threatening and to thus have low util- ity. Empirical evidence from the leadership litera- ture supports this thesis: lower performers are more likely to make supervisors look bad, interfere with their capacity to accomplish their work, and take up more of their time addressing the fallout poor performance causes (Bass, 1990). The low utility of subordinates who are perceived to be lower per- formers positions them beyond their supervisors’

282 AprilAcademy of Management Journal

scopes of justice, which puts them at risk of exclu- sionary practices such as abusive supervision.

The link between performance evaluations and abusive supervision is also consistent with the vic- tim precipitation literature, which examines the individual characteristics and behaviors that put people at risk of becoming the targets of aggressive and hostile responses (Elias, 1986). Potential ag- gressors choose targets strategically, focusing their hostility on people who seem difficult to like and/or those who appear to be vulnerable and un- able to defend themselves. Olweus’s (1978) concept of provocative victims explains why supervisors might target subordinates with low performance. According to Olweus, people become good targets for victimization when they are perceived to be difficult to work with. In the context of supervisor- subordinate relationships, supervisors should per- ceive subordinates who are lower performers to be frustrating, aggravating, and annoying—character- istics that align well with the provocative victim profile. Hence, the perception that employees are lower performers has the potential to evoke vic- timization in the form of abusive supervision. Integrating these arguments with those suggest- ing a relationship between perceived deep-level dissimilarity and supervisor evaluations of sub- ordinate performance leads us to propose that the unfavorable performance evaluations that per- ceived deep-level dissimilarity evokes is associ- ated with subordinates’ perceptions of abusive supervision.

Hypothesis 2. A supervisor’s perceptions of a subordinate’s performance partially mediate the relationship between the supervisor’s per- ceived deep-level dissimilarity with the subor- dinate and abusive supervision directed to- ward that subordinate.

Although researchers have failed to link supervi- sor-subordinate relationship conflict with supervi- sor perceptions of subordinate performance, extant research suggests that subordinates in higher-quality working relationships with their supervisors are evaluated more favorably than those who are in- volved in lower-quality relationships with their su- pervisors (e.g., Huang & Iun, 2006; Judge & Ferris, 1993; Kolodinsky, Treadway, & Ferris, 2007). The rationale for this link is that lower-quality working relationships involve more conflictive interaction, which colors individuals’ perceptions of their counterparts’ capability and effectiveness. In su- pervisor-subordinate relationships characterized by higher levels of relationship conflict, supervi- sors should be more likely to “see” performance problems because of “negative halo” and inability

to reconcile a conflictive relationship with good performance. Thus, supervisors are likely to per- ceive employees with whom they have conflict as being poor performers. In addition, based on the work suggesting that conflictive relationships mo- tivate people to injure their counterparts (Struch & Schwartz, 1989), it can be argued that supervisors use performance evaluation processes to injure subordinates with whom they are in conflict. We therefore expect that the relationship conflict that perceived deep-level dissimilarity evokes will, in turn, be associated with lower evaluations of sub- ordinate performance. Integrating these arguments with those suggesting a relationship between su- pervisor evaluations of subordinate performance and abusive supervision (i.e., the conceptual basis for Hypothesis 2), we propose that the lower per- formance evaluations relationship conflict evokes are associated with abusive supervision. This line of reasoning is depicted in Figure 1 as a mediation chain in which perceived deep-level dissimilarity is positively associated with relationship conflict; relationship conflict is negatively related to super- visor evaluations of subordinate performance; and supervisor evaluations of subordinate performance are, in turn, negatively related to abusive supervi- sion. In the parlance of Kenny, Kashy, and Bolger (1998) and Fletcher (2006), relationship conflict is a distal mediator and subordinate performance is a proximal mediator in the relationship between perceived deep-level dissimilarity and abusive supervision.

Hypothesis 3. Relationship conflict is a distal partial mediator and supervisor evaluation of subordinate performance is a proximal partial mediator of the relationship between perceived deep-level dissimilarity and abu- sive supervision.

Moderating Effects of Subordinate Performance

In the conceptual frame associated with Hypoth- esis 1, we argue that supervisors exclude from their scope of justice subordinates they perceive to be adversaries—those with whom they have relation- ship conflict. This dynamic, in turn, is associated with hostility in the form of abusive supervision. Here we qualify that prediction by proposing that some subordinate adversaries are included in a su- pervisor’s scope of justice and therefore do not become targets of abusive supervision. Specifically, we argue that subordinate adversaries are per- ceived to deserve fair treatment when they are good (“higher”) performers.

Supervisors should process relationship conflict with subordinates differently when they respect

2011 283Tepper, Moss, and Duffy

their ability and competence. Adversaries who are perceived to be capable should evoke more respect than adversaries who appear less competent. This is because parties to relationship conflict are more likely to perceive competent adversaries’ positions as deriving from reasoned and thoughtful processes (Bush & Folger, 1994). Indeed, one way of inducing cooperation between adversaries is to encourage one party’s perception that the other party is capa- ble and worthy (Ting-Toomey, 1999). We do not mean to imply that supervisors will have no ani- mus toward adversarial subordinates who perform well, only that they will view those subordinates as more deserving of fair treatment than low perform- ers subordinate adversaries. Hence, when relation- ship conflict is higher, supervisors are more likely to execute exclusionary acts such as abusive super- vision against low-performing subordinates.

These arguments imply that the mediated effect captured in Hypothesis 1 varies over levels of su- pervisor perceptions of subordinate performance. The positive relationship between perceived rela- tionship conflict (which perceived deep-level dis- similarity evokes) and abusive supervision is stron- ger when supervisors perceive subordinates to be lower performers. Predictions of this sort are re- ferred to as second-stage moderation models in which a mediated effect varies over levels of a moderator that operates at the second stage of the mediated relationship (Edwards & Lambert, 2007). Stated formally:

Hypothesis 4. Supervisor perceptions of subor- dinate performance moderate the indirect ef- fect of perceived deep-level dissimilarity on abusive supervision (through relationship con- flict); the mediated effect is stronger when a supervisor perceives a subordinate as having lower performance.


Participants and Procedures

Participants were recruited from seven health care organizations, including hospitals, long-term care facilities, and outpatient facilities, located in the southeastern United States. We contacted key individuals with administrative responsibilities (e.g., unit directors, directors of human resources, and chief operating officers) and asked them for assistance in identifying and recruiting supervisor- subordinate dyads in their organizations.

Though the procedures varied slightly across or- ganizations, the overall data collection strategy was as follows: Volunteers from the ranks of both man- agerial and nonmanagerial employees were sought

through e-mail notifications, announcements at staff meetings, and/or the distribution of flyers. Once volunteers were identified, we matched each supervisor with a volunteering subordinate. When multiple subordinates matched with a supervisor, we chose one participating subordinate randomly. Participants completed surveys during work hours on two occasions separated by six weeks. One of the authors was present to explain the purpose of the study and to administer surveys. In the rare cases in which no author was available during for- mal survey administration, a contact in the organ- ization assisted in administering the survey.

Supervisors and subordinates completed the sur- veys at separate times in separate locations. Before administering the survey, we told each group that we were conducting a study on various aspects of supervisor-subordinate relationships. Supervisors were told that one of their subordinates would be participating, and subordinates were told that their supervisor would be participating. Although each knew the identity of the other, all participants were assured that their individual results would not be shared with their supervisors or anyone else in their organization. They were told that supervisors and subordinates from multiple health care organ- izations would be participating and that we would use only aggregated data.

These data collection procedures produced us- able data from 183 independent supervisor-subor- dinate dyads. Across the seven organizations from which we collected data, the participation rate of those in supervisory positions ranged from 42 to 95 percent, with an average of 61 percent. The super- visors’ average age was 46 years; 69 percent were women; 28 percent identified themselves as white; 29 percent, as Hispanic; 39 percent, as black; 3 percent, as Asian; and 1 percent, as “other.” They averaged 8.67 years working in the organization and were employed as physicians, nursing super- visors, technical department managers (e.g., imag- ing, lab services), physical plant supervisors, de- partment directors, and higher-level hospital administrators. “Executive/upper management po- sitions” were reported by 34 percent; 45 percent were in “middle management”; and 21 percent held “lower-level” supervisory positions.

The subordinates’ average age was 44 years; 77 percent were women; 17 percent identified them- selves as white; 39 percent, as Hispanic; 39 percent, as black; 4 percent, as Asian; and less than 2 per- cent, as “other.” Their average tenure in the or- ganization was 8.45 years, and they averaged 4.33 years of working for their supervisor. They occu- pied a variety of positions, including nurse, tech- nician, food service employee, physical plant em-

284 AprilAcademy of Management Journal

ployee, and higher-level employee reporting directly to upper and middle management supervisors.


Perceived dissimilarity with subordinate. At time 1, supervisors completed Turban and Jones’s (1988) three-item measure of perceived similarity with subordinate and two additional items written and used by Liden et al. (1993). The items, prefaced with the phrase “This subordinate and I . . .,” read as follows: “are similar in terms of our outlook, perspective, and values,” “analyze problems in a similar way,” “think alike in terms of coming up with a similar solution for a problem,” “are alike in a number of areas,” and “see things in much the same way” (1 � “strongly disagree,” � 7, “strongly agree”). We recoded ratings on the items so that higher ratings captured higher levels of perceived dissimilarity and averaged the answers on items to form total scores for perceived dissimilarity (� � .96).

Relationship conflict. At time 1, supervisors re- sponded to three items from Jehn and Mannix (2001). This scale, typically used to measure intra- group conflict, was adapted for use in this study. The items were preceded with the instruction, “Please refer to your employee, ____, when answer- ing the following questions.” The three items were, “How much relationship tension is there between you and this employee?”; “How often do you and this employee get angry while working?”; “How much emotional conflict is there between you and this employee?” (1 � “none,” 5 � “a lot”). We averaged the item scores to form total scores for relationship conflict (� � .81).

Subordinate performance. At time 2, the super- visors completed a four-item measure of subordi- nates’ performance developed by Liden et al. (1993). The items and response scales were as fol- lows: “My subordinate is superior to other subor- dinates that I’ve supervised before” (1 � “strongly disagree,” 7 � “strongly agree”); “Rate the overall level of performance that you observe for this subordinate” (1 � “unacceptable,” 7 � “out- standing”); “What is your personal view of your subordinate in terms of his or her overall effec- tiveness?” (1 � “very ineffective,” 7 � “very ef- fective”); and “Overall to what extent do you feel your subordinate has been effectively fulfilling his or her roles and responsibilities?” (1 � “not effectively at all,” 7 � “very effectively”). We averaged the item ratings to form total scores for subordinate performance (� � .87).

Abusive supervision. Also at time 2, subordi- nates completed Tepper’s (2000) 15-item measure 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 “He/she uses this behavior very often with me,” to report how often their boss used behaviors such as “tells me my thoughts and feelings are stupid,” “puts me down in front of others,” “ridicules me,” and “makes negative comments about me to oth- ers.” We averaged the item ratings to form total scores for abusive supervision (� � .96).

Control variables. We explored the viability of several control variables that could provide alter- native explanations for the relationships depicted in our model. These included surface demographic diversity (i.e., sex similarity, age similarity, and race similarity), supervisor-subordinate relation- ship tenure, and organization. Although research- ers have shown that surface demographic variables relate to relational outcomes (e.g., Tsui & O’Reilly, 1989), substantial evidence suggests that the effects of surface demography diminish over time as indi- viduals learn more about each other (Harrison, Price, & Bell, 1998; Harrison, Price, Gavin, & Flo- rey, 2002). Even though the average relationship tenure in our sample was almost 4.5 years, some manager-subordinate dyads in the sample had worked together for less time, so it was prudent to control for surface-level demographic characteris- tics as well as relationship tenure. The demo- graphic variables were coded as follows: sex simi- larity was 0 when a supervisor and subordinate were of different sexes and 1 when they were the same; race similarity was coded 0 for different races and 1 for the same; and age similarity was equal to the absolute value of the difference between the supervisor’s age and the subordinate’s age. Rela- tionship tenure was the number of years the subor- dinate had reported directly to the supervisor. Finally, because our sample included seven organ- izations that might have different cultures and ori- entations to diversity, we also created six dummy- coded variables and entered these terms in our analyses before exploring the effects of our substan- tive variables.

Analytical Strategy

We tested our hypotheses using path-analytic procedures for testing mediation models from MacKinnon, Lockwood, Hoffman, West, and Sheets (2002) and Edwards and Lambert’s (2007) elabora- tion of those procedures, which allow examination of second-stage moderation models. The Appendix shows the regression equations that we estimated

2011 285Tepper, Moss, and Duffy

and the specific terms we examined to test Hypoth- eses 1 through 4. The examination of mediated and moderated effects requires estimating product terms, which are not normally distributed (Shrout & Bolger, 2002). Hence, we tested all hypothesized effects by constructing bias-corrected confidence intervals from 10,000 bootstrapped samples (Mooney & Duval, 1993; Shrout & Bolger, 2002).


We conducted a preliminary analysis to assess differences on the substantive variables among the seven organizations in our study. One-way analy- ses of variance suggested no mean differences for perceived deep-level dissimilarity (F[6, 170] � 0.76, n.s.), perceived relationship conflict (F[6, 170] � 0.41, n.s.), supervisor evaluations of subor- dinate performance (F[6, 170] � 1.34, n.s.), and abusive supervision (F[6, 170] � 1.51, n.s.). We also computed intraclass coefficients for the four sub- stantive variables to determine whether it was more appropriate to analyze the data using ordinary least squares (OLS) regression or a multilevel modeling technique that can account for nonindependence among observations. The ICC1s were .00 for per- ceived deep-level dissimilarity and perceived rela- tionship conflict, .02 for subordinates’ performance evaluations, and .03 for abusive supervision. Be- cause no variance in perceived deep-level dissimi- larity or relationship conflict resided between groups, we could assume that nesting within organ- izations had no effect on their relationships with

each other, with subordinate performance evalua- tions, or with abusive supervision. We therefore concluded that it was acceptable to test the hypoth- eses using OLS regression.

Table 1 shows descriptive statistics and correla- tions among the study variables. Perceived dissim- ilarity correlated positively with relationship con- flict (r � .40, p � .01) and abusive supervision (r � .29, p � .01) and negatively with subordinate per- formance evaluations (r � �.29, p � .01); relation- ship conflict correlated negatively with subordi- nate performance evaluations (r � �.31, p � .01) and positively with abusive supervision (r � .37, p � .01); and subordinate performance evaluations correlated negatively with abusive supervision (r � �.39, p � .01). None of the control variables were related to perceived dissimilarity, relationship con- flict, performance evaluations, or abusive supervi- sion. Indeed, including the controls in the regres- sion analyses did not affect the results. We therefore followed Becker’s (2005) recommenda- tion to report the results excluding the control variables.

We centered all variables at their grand means before evaluating the regression equations. Table 2 shows the unstandardized regression results. The equations explained significant variance in rela- tionship conflict (F[1, 175] � 35.49, R2 � .17, p � .01), subordinate performance (F[2, 174] � 14.80, R2 � .15, p � .01), and abusive supervision (F[4, 172] � 35.49, R2 � .37, p � .01). The signs on the unstandardized regression coefficients suggest that perceived dissimilarity positively predicted rela-

TABLE 1 Descriptive Statistics and Correlations among Study Variablesa

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 0.57 0.50 .01 �.02 4. Relationship

length 4.46 5.07 .19** �.11 .00

5. Organization 1 0.08 0.27 .04 .12 .09 �.17* 6. Organization 2 0.10 0.29 .11 .05 .01 �.04 �.10 7. Organization 3 0.03 0.18 �.09 .01 �.03 �.10 �.06 �.06 8. Organization 4 0.12 0.32 .10 .04 �.07 .12 �.11 �.12 �.07 9. Organization 5 0.13 0.33 .16* .17* .02 .02 �.05 �.12 �.07 �.14

10. Organization 6 0.46 0.50 �.19* �.25** �.06 .11 �.27** �.30** �.17* �.34** �.35** 11. Organization 7 0.09 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

conflict 1.42 0.64 .02 �.09 .07 .11 .05 .04 �.03 �.08 �.09 .08 �.02 .40** (.81)

14. Performance 5.61 1.00 �.14 �.01 .09 .09 �.07 .02 �.08 .04 �.13 .07 .08 �.29** �.31** (.87) 15. Abusive

supervision 1.27 0.47 .00 .06 .08 �.09 .03 �.11 .01 �.04 .09 .08 �.10 .29** .37** �.39** (.96)

a Sample sizes on the correlations range from 143 to 178. At time 1, supervisors completed the measures of perceived dissimilarity with subordinate and perceived relationship conflict with subordinate. At time 2, supervisors completed the measure of subordinates’ performance and subordinates completed the measure of abusive supervision.

Alpha internal consistency reliability coefficients appear on the main diagonal. * p � .05

** p � .01

286 AprilAcademy of Management Journal

tionship conflict (b � .15, p � .01), negatively predicted subordinate performance (b � �.12, p � .05), and positively predicted abusive supervision (b � .04, p � .05); relationship conflict negatively predicted subordinate performance (b � �.40, p � .01) and was unrelated to abusive supervision (b � .09, n.s.); and subordinate performance negatively predicted abusive supervision (b � �.11, p � .01). Table 2 also shows that the relationship conflict by subordinate performance cross-product was related to abusive supervision (b � �.17, p � .01).

We used the information from these equations to calculate the mediated effect of perceived deep- level dissimilarity on abusive supervision through relationship conflict (Hypothesis 1), subordinate performance (Hypothesis 2), relationship conflict and subordinate performance (Hypothesis 3), and relationship conflict at higher and lower levels of subordinate performance (Hypothesis 4). Table 3 presents the estimates associated with each hypothesis.

The first row of results in Table 3 shows that the mediated effect of perceived deep-level dissimilar- ity on abusive supervision through relationship conflict was not significant (� � .01, n.s.). Hence, Hypothesis 1 was not supported. The next row of results shows that subordinate performance medi- ated the relationship between perceived deep-level dissimilarity on abusive supervision (� � .01, p � .01). Hence, Hypothesis 2 was supported. The next row in Table 3 shows that Hypothesis 3 was sup- ported; the mediated effect of perceived deep-level dissimilarity on abusive supervision through rela- tionship conflict and subordinate performance was significant (� � .01, p � .05).

The last two rows in Table 3 show the mediated effect of perceived deep-level dissimilarity on abu- sive supervision through relationship conflict at

high and low levels of subordinate performance. When subordinate performance was high, the me- diated effect of relationship conflict was not signif- icant (� � �.01, n.s.); however, when subordinate performance was low, relationship conflict medi- ated the effect of perceived deep-level dissimilarity on abusive supervision (� � .04, p � .01). Figure 2 shows the plot of the mediated effects at higher and lower levels of subordinate performance (Aiken & West, 1991). The form of the interaction is consis- tent with Hypothesis 4; the mediated effect of rela- tionship conflict was stronger when subordinate performance was perceived to be lower.


Our research contributes to a growing body of research exploring abusive supervision in work or- ganizations. The research to date has shed light on the deleterious consequences of abusive supervi- sion, but little is known about the factors that pre- dict when abusive supervision is likely to occur. The few studies that have examined the anteced- ents of abusive supervision have focused on the treatment supervisors receive and suggest that su- pervisor perceptions of organizational injustice and contract breach are associated with subordinate re- ports of supervisory abuse (Aryee et al., 2007; Hoobler & Brass, 2006; Tepper et al., 2006). Our research shifts the focus from the mistreatment su- pervisors experience to concepts described in Opo- tow’s (1990a, 1995) work on moral exclusion. This approach provides a basic framework for answering

TABLE 2 Path-Analytic Regression Resultsa


Dependent Variables

Conflict Performance Abuse

Perceived dissimilarity .15** �.12* .04* Relationship conflict �.40* .09 Performance �.11** Conflict � performance �.17*

Equation R2 .17** .15** .37**

a n � 177. Tabled values are unstandardized regression coef- ficients. All parameter estimates were tested for significance using bias-corrected bootstrapped confidence intervals.

* p � .05 ** p � .01

TABLE 3 Indirect Effects of Perceived Deep-Level Dissimilarity

on Abusive Supervisiona

Mediator(s) � Hypothesis

Tests of hypothesized mediated effects

Relationship conflict .01 1 Subordinate performance .01** 2 Relationship conflict &

subordinate performance .01* 3

Test of hypothesized moderated indirect effect

Relationship conflict 4 Low subordinate performance .04** High subordinate performance �.01

a � refers to the mediated effect. All estimates were tested for significance using bias-corrected confidence intervals from 10,000 bootstrapped samples.

* p � .05 ** p � .01

2011 287Tepper, Moss, and Duffy

the question, Why do supervisors abuse specific subordinates?

Implications for Theory and Research

Our findings suggest a causal framework in which perceived deep-level dissimilarity evokes perceived relationship conflict, which produces lower evaluations of subordinate performance, which, in turn, lead to higher levels of abusive supervision. Our findings also suggest that relation- ship conflict mediates the effect of perceived deep- level dissimilarity and abusive supervision, but only when supervisors perceive subordinates as having low performance. The links between subor- dinate performance evaluations and abusive super- vision and between perceived relationship conflict and abusive supervision (when subordinate perfor- mance is low) are consistent with moral exclusion theory. From a moral exclusion perspective, per- ceived relationship conflict and low subordinate performance put referent subordinates beyond a focal supervisor’s scope of justice, which translates into an exclusionary practice: abusive supervision.

However, although the predictors of abusive su- pervision we examined were restricted to analogs for the key constructs that moral exclusion theory specifies (i.e., dissimilarity, conflict, and utility), our model and arguments augment extant moral exclusion research in two general ways. First, drawing on contributions to the literature on diver- sity in supervisor-subordinate relationships, we proposed that perceived deep-level dissimilarity is related to abusive supervision indirectly, through its effects on supervisor perceptions of relationship conflict and supervisor perceptions of subordinate

performance. Second, the relationships between re- lationship conflict and abusive supervision and be- tween subordinate performance and abusive super- vision, although consistent with a moral exclusion argument, are also consistent with other contribu- tions to the management and psychology litera- tures. The interpersonal conflict literature suggests that people develop negative affect toward adver- saries, which translates into hostility. The victim precipitation literature supports the link between subordinate performance and abusive supervision. Subordinates who are perceived to be lower per- formers may become targets of abusive supervision because they fit the provocative victim profile— annoying, aggravating, and difficult to work with (Olweus, 1978).

Two features of our results may be regarded as unexpected and have implications for theory and research. First, we found that even after we ac- counted for the effects of supervisor perceptions of relationship conflict and subordinate performance, and their interaction, supervisor perceptions of deep-level dissimilarity remained a significant pre- dictor of abusive supervision (see Table 2). Hence, perceived deep-level dissimilarity had both direct and indirect effects on abusive supervision. This finding is consistent with moral exclusion theory and suggests the need for additional research that examines other possible mediating mechanisms, including scope of justice. We return to this in our summary of the study’s limitations.

The second unexpected finding was that Hypoth- esis 1 was not supported; relationship conflict did not mediate the relationship between perceived deep-level dissimilarity and abusive supervision. Of course, the results of subsequent analyses ex-

FIGURE 2 Mediated Effect of Perceived Deep-Level Dissimilarity with Subordinate on Abusive Supervision (through

Relationship Conflict) at High and Low Levels of Subordinate Performance







−1 s.d. +1 s.d.

Perceived Dissimilarity with Subordinate

Abusive Supervision

High subordinate performance Low subordinate performance

288 AprilAcademy of Management Journal

plain why this may have occurred. The test of Hy- pothesis 3 suggested that relationship conflict was a distal mediator of the effect of perceived deep- level dissimilarity on abusive supervision, which operated through supervisor evaluations of subor- dinate performance. The test of Hypothesis 4 sug- gested that relationship conflict played a mediating role, albeit when subordinates were perceived to be lower performers. It appears, then, that models linking perceived relationship conflict and abusive supervision should account for supervisor percep- tions of subordinate performance.

A final and more general implication of our work is that it demonstrates the usefulness of using con- cepts from the moral exclusion literature to explain one of the many forms of antisocial work behavior that researchers have explored. Similar models could be developed to explore other manifestations of antisocial work behavior, such as rank-and-file employees’ acts of workplace deviance directed against their organization, supervisor, and cowork- ers—counternormative actions that threaten organ- izations and their members (Robinson & Bennett, 1995). In extant work, researchers have invoked theories of justice, social learning, and personality to explain employees’ deviance (e.g., Aquino, Douglas, & Martinko, 2004; Skarlicki, Folger, & Tes- luk, 1999). We take the position that value may exist in modeling the occurrence of workplace de- viance by exploring the factors that put the targets of deviant behavior beyond a focal individual’s scope of justice. It should be noted that the scope of justice concept is not reserved to human targets; the framework can be used to explore focal per- sons’ attitudes toward any referent that can be- come the target of maltreatment (e.g., an environ- ment [Clayton & Opotow, 2003]). Hence, researchers should be able to use moral exclusion concepts to model employees’ deviance directed against their organization.

Study Limitations and Recommendations for Future Research

Our study has limitations that should be ac- knowledged and addressed in future research. First, the nature of our research design requires caution in the conclusions we draw from the re- sults. Although we collected the data in two waves and from two sources, we failed to measure all variables at both time points; we therefore cannot rule out the possibility that abusive supervision caused perceived dissimilarity, relationship con- flict, and subordinate performance. In addition, we cannot examine the effects of perceived dissimilar- ity, relationship conflict, and subordinate perfor-

mance on changes in abusive supervision. It is therefore appropriate to conclude that our study results are merely suggestive of the causal ordering our model implies. A natural follow-up to the re- search we have reported here is to examine similar predictions using designs in which all variables are measured several times.

Such designs would also address a second limi- tation: that we cannot entirely rule out the possi- bility that the results can be attributed to common method bias. This is particularly true for the rela- tionship between perceived deep-level dissimilar- ity and relationship conflict, the measures of which were both included in the supervisor time 1 survey. However, it is fair to conclude that common method bias is an unlikely explanation for our me- diation results, given that all the analyses relied to some degree on data collected at different time points and/or from different sources (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Hence, for ex- ample, the test of Hypothesis 3 was based on the product of the links between perceived deep-level dissimilarity and relationship conflict (which could have been influenced by common method bias), relationship conflict and subordinate perfor- mance (which were measured at two time points), and subordinate performance and abusive supervi- sion (which were measured from two perspectives).

A third limitation is that all data were collected from health care employees, so we must cautiously render conclusions regarding the generalizability of the results to organizations outside the health care delivery sector. It has been argued that abusive supervision is particularly common in health care, where the costs of failure are high, yet workers face heavy workloads and time pressures (Richman, Flaherty, & Rospenda, 1996; Richman, Flaherty, Rospenda, & Christensen, 1992). A promising di- rection for future research is to conduct cross-in- dustry comparisons to determine whether the re- sults that emerged here generalize to industries not sharing these features.

A fourth limitation is that although we investi- gated the viability of a number of control variables, we were unable to control for all variables that may be related to abusive supervision. For example, we did not control for the tendency to respond in a socially desirable manner, nor did we account for supervisors’ justice perceptions and trait hostility, although previous research has linked both with abusive supervision (Hoobler & Brass, 2006). Our framework may be viewed as a middle-range theory (Moore, Johns, & Pinder, 1980), one that does not account for all possible influences but instead mod- els how one set of factors predict abusive supervi- sion. An important next step would be to simul-

2011 289Tepper, Moss, and Duffy

taneously investigate the relative predictive power of models that depict abusive supervision as the result of mistreatment perpetrated against supervisors and models rooted in the notion that abusive supervisors are themselves perpetrators of mistreatment.

A fifth limitation has to do with our measure of deep-level dissimilarity. We used a general mea- sure drawn from previous studies of perceived deep-level dissimilarity in supervisor-subordinate relationships (Liden et al., 1993; Turban & Jones, 1988). The items comprising this scale capture deep-level dissimilarity that is relevant to supervi- sor-subordinate relationships but, as Harrison et al. (1998) pointed out, deep dissimilarity can manifest along a number of work-related values, attitudes, and personality traits. Our use of a general measure seemed appropriate for our research objective of initially examining the role that perceived deep- level dissimilarity plays in explaining abusive su- pervision. In future work, it would be fruitful to take a finer-grained approach by exploring the di- rect, indirect, and moderated effects of specific deep-level differences on abusive supervision.

A final limitation is that we failed to measure moral exclusion, which constitutes the proximal link in a moral exclusion explanation for abusive supervision. Instead, we interpreted reports of an exclusionary practice (i.e., abusive supervision) as evidence that subordinate targets had been morally excluded from supervisors’ scope of justice. Our approach is consistent with extant published re- search on moral exclusion, which has yet to opera- tionalize moral exclusion directly (Hafer & Olson, 2003). As we said above, several links we tested are consistent with ideas forwarded in other theories (e.g., the relationship between supervisor percep- tions of subordinate performance and abusive su- pervision is consistent with victim precipitation theory), but a logical direction for future research is to measure and model the cognitive rationales that are associated with moral exclusion to determine to what extent moral exclusion explains the effects observed here.

A promising foundation for such work is Ban- dura and colleagues’ theory and research on moral disengagement, which addresses how people deac- tivate the self-regulatory mechanisms that ordinar- ily inhibit them from performing antisocial acts (Bandura, 1990, 1999; Bandura, Barbanelli, Ca- prara, & Pastorelli, 1996). The mechanisms of moral disengagement Bandura (1999) has identified in- clude such mental constructions as construing harm as serving a righteous purpose, and dehuman- izing victims. Future research should explore the possibility that these mechanisms explain the ef-

fects of the predictors studied here as well as the effects of other potential exogenous influences on abusive supervision, such as personality and devel- opmental problems (Evans, 1996; Hoobler & Brass, 2006). Such research would also permit examina- tion of an emerging debate in the literature as to whether these cognitive justifications for mistreat- ment are the causes, the consequences, or the causes and consequences of such behavior (Trevino, Weaver, & Reynolds, 2006). Opotow (1990a, 1995) is clear in positioning moral exclu- sion as an antecedent of exclusionary practices, but conceivably moral exclusion captures perpetrators’ post hoc explanations for their mistreatment of ref- erent others (“If I abuse my subordinates, I must think that they deserve unfair treatment”).

Practical Implications

As workforce composition becomes increasingly diverse, supervisors will more frequently have di- rect reports whose values and attitudes differ from their own (Mohammed & Angell, 2004). Our re- search suggests that the resulting perceived deep- level dissimilarity may be directly and indirectly related to abusive supervision. A more sanguine implication of our research is that it may be possi- ble to reduce the occurrence of abusive supervision by refraining from hiring individuals for manage- rial positions who are dispositionally inclined to have a narrow scope of justice and to execute hos- tile acts. A relevant individual difference is trait empathy, the dispositional tendency to take the perspective of others and to recognize and experi- ence concern for others’ thoughts and feelings (Eisenberg & Miller, 1987). A recent study suggests that low-empathy workers were more likely to per- ceive that others deserve harmful treatment (Detert, Trevino, & Sweitzer, 2008), and a large body of empirical work suggests that low-empathy individ- uals are more hostile than those who are higher in trait empathy (Miller & Eisenberg, 1988). By select- ing personnel for trait empathy, organizations may be able to reduce the pool of managers who are inclined to put subordinates beyond the scope of justice and to abuse them.

Of course, it may not always be possible to use trait empathy as a selection tool. In such cases, organizations should discourage supervisors from perceiving that subordinates fall outside their scope of justice and should promote skills for work- ing constructively with morally excluded subordi- nates. Organizations should broaden supervisors’ scope of justice by endorsing concepts of plural- ism—the notion that there is value in promoting tolerance and appreciation for divergent values

290 AprilAcademy of Management Journal

(Opotow, 1990b). Many organizations have found that simply increasing employee diversity does not by itself facilitate inclusion; rather, successful in- terventions focus on discovering similarities, de- veloping empathy, and identifying ways of capital- izing on differences (Chavez & Weisinger, 2008). In keeping with these ideas, Homan et al. (2007) found that through training, individuals can be en- couraged to adopt prodiversity beliefs, and that this, in turn, leads to healthier interpersonal pro- cesses and superior performance.

Even when organizations embrace pluralism, sit- uations in which supervisors exclude some subor- dinates from their scope of justice may occur. In anticipation of these circumstances, organizations can use justice training to help supervisors interact constructively with all subordinates, particularly those whom the supervisors perceive to be dissim- ilar, those with whom they have relationship con- flict, and those they perceive to be lower perform- ers. Justice training involves coaching supervisors to use techniques that reduce psychological dis- tance with employees, promote the perception that employees have “voice,” and equitably apply deci- sion-making criteria (Greenberg, 2006; Skarlicki & Latham, 2005). For example, an organization may set the goal of supervisors meeting with each direct report, one-on-one, at least monthly. Employees would be invited to use that time to provide feed- back and voice concerns. In essence, supervisors would be trained in techniques to recognize and suspend bias so that they can overcome the ten- dency to abuse subordinates who fall beyond their scope of justice.


Despite its low base rate, abusive supervision is a costly workplace phenomenon in terms of lost pro- ductivity, absenteeism, turnover, and health care expenditures. It is therefore important that re- searchers continue to investigate the antecedent conditions associated with abusive supervision. Our findings are significant because they draw at- tention to previously unexamined antecedents of abusive supervision and provide the bases for prac- tical interventions that have the potential to curb the frequency of abusive supervision.


Aiken, L. S., & West, S. G. 1991. Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage.

Aquino, K., Douglas, S. C., & Martinko, M. J. 2004. Overt anger in response to victimization: Attributional

style and organizational norms as moderators. Jour- nal of Occupational Health Psychology, 9: 152– 164.

Aryee, S., Chen, Z. X., Sun, L., & Debrah, Y. A. 2007. Antecedents and outcomes of abusive supervision: Test of a trickle-down model. Journal of Applied Psychology, 92: 191–201.

Bamberger, P. A., & Bacharach, S. B. 2006. Abusive su- pervision and subordinate problem drinking: Taking resistance, stress, and subordinate personality into account. Human Relations, 59: 1–30.

Bandura, A. 1990. Selective activation and disengage- ment of moral control. Journal of Social Issues, 46: 27–46.

Bandura, A. 1999. Moral disengagement in the perpetu- ation of inhumanities. Personality and Social Psy- chology Review, 3: 193–209.

Bandura, A., Barbanelli, C., Caprara, G., & Pastorelli, C. 1996. Mechanisms of moral disengagement in the exercise of moral agency. Journal of Personality & Social Psychology, 71: 364–374.

Bass, B. M. 1990. Bass and Stogdill’s handbook of lead- ership (3rd ed.). New York: Free Press.

Becker, T. 2005. Potential problems in the statistical con- trol of variables in organizational research: A quali- tative analysis with recommendations. Organiza- tional Research Methods, 8: 274–289.

Brewer, M. B. 1999. The psychology of prejudice: In- group love or outgroup hate? Journal of Social Is- sues, 55: 429–444.

Bruk-Lee, V., & Spector, P. E. 2006. The social stressors- counterproductivity work behaviors link: Are con- flicts with supervisors and coworkers the same? Journal of Occupational Health Psychology, 11: 145–156.

Bush, R., & Folger, J. 1994. The promise of mediation: Responding to conflict through empowerment and recognition. San Francisco: Jossey-Bass.

Byrne, D. 1971. The attraction paradigm. New York: Academic.

Chavez, C. I., & Weisinger, J. Y. 2008. Beyond diversity training: A social infusion for cultural inclusion. Human Resource Management, 47: 331–350.

Clayton, S., & Opotow, S. 2003. Justice and identity: Changing perspectives on what is fair. Personality and Social Psychology Review, 7: 298–310.

Detert, J. R., Trevino, L. K., & Sweitzer, V. L. 2008. Moral disengagement in ethical decision making: A study of antecedents and outcomes. Journal of Applied Psychology, 93: 374–391.

Deutsch, M. 1973. The resolution of conflict: Construc- tive and destructive processes. New Haven, CT: Yale University Press.

Duffy, M. K., Ganster, D., & Pagon, M. 2002. Social un-

2011 291Tepper, Moss, and Duffy

dermining in the workplace. Academy of Manage- ment Journal, 45: 331–351.

Dupre, K. E., Inness, M., Connelly, C. E., Barling, J., & Hoption, C. 2006. Workplace aggression in teenage part-time employees. Journal of Applied Psychol- ogy, 91: 987–997.

Edwards, J. R., & Lambert, L. S. 2007. Methods for inte- grating moderation and mediation: A general analyt- ical framework using moderated path analysis. Psy- chological Methods, 12: 1–22.

Eisenberg, N., & Miller, P. A. 1987. The relation of em- pathy to prosocial and related behaviors. Psycholog- ical Bulletin, 101: 91–119.

Elias, R. 1986. The politics of victimization: Victims, victimology, and human rights. New York: Oxford University Press.

Evans, P. 1996. The verbally abusive relationship: How to recognize it and how to respond. Avon, MA: Adams Media.

Fletcher, T. 2006. Methods and approaches to assessing distal mediation. Paper presented at the annual meeting of the Academy of Management, Atlanta.

Greenberg, J. 2006. Losing sleep over organizational in- justice: Attenuating insomniac reactions to under- payment inequity with supervisory training in inter- actional justice. Journal of Applied Psychology, 91: 58–69.

Hafer, C. L., & Olson, J. M. 2003. An analysis of empirical research in the scope of justice. Personality and Social Psychology Review, 4: 311–323.

Halevy, N., Bornstein, G., & Sagiv, L. 2008. “In-group love” and “out-group hate” as motives for individual participation in intergroup conflict: A new para- digm. Psychological Science, 19: 405–411.

Harrison, D. A., & Klein, K. J. 2007. What’s the differ- ence? Diversity constructs as separation, variety, or disparity in organizations. Academy of Manage- ment Review, 32: 1199–1228.

Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyond relational demography: Time and the effects of sur- face- and deep-level diversity on work group cohe- sion. Academy of Management Journal, 41: 96– 107.

Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A. 2002. Time, teams, and task performance: Changing effects of surface- and deep-level diversity on group functioning. Academy of Management Journal, 45: 1029–1045.

Hewstone, M., Rubin, M., & Willis, H. 2002. Intergroup bias. In S. T. Fiske, D. L. Schacter, & C. Zahn-Waxler (Eds.), Annual review of psychology, vol. 53: 575– 604. Palo Alto, CA: Annual Reviews.

Hobman, E. V., & Bordia, P. 2006. The role of team identification in the dissimilarity-conflict relation- ship. Group Processes and Intergroup Relations, 9: 483–507.

Hogg, M. A., & Abrams, D. 1988. Social identifications: A social psychology of intergroup relations and group processes. London: Routledge.

Homan, A. C., Hollenbeck, J., Humphrey, S., van Knip- penberg, D., Ilgen, D., & Van Kleef, G. A. 2008. Fac- ing differences with an open mind: Openness to experience, salience of intragroup differences, and performance of diverse workgroups. Academy of Management Journal, 51: 1204–1222.

Homan, A. C., van Knippenberg, D., Van Kleef, G. A., & De Dreu, C. K. W. 2007. Bridging faultlines by valu- ing diversity: The effects of diversity beliefs on in- formation elaboration and performance in diverse work groups. Journal of Applied Psychology, 92: 1189–1199.

Hoobler, J., & Brass, D. 2006. Abusive supervision and family undermining as displaced aggression. Jour- nal of Applied Psychology, 91: 1125–1133.

Huang, X., & Iun, J. 2006. The impact of subordinate- supervisor similarity in growth-need strength on work outcomes: The mediating role of perceived similarity. Journal of Organizational Behavior, 27: 1121–1148.

Inness, M., Barling, J., & Turner, N. 2005. Understanding supervisor-targeted aggression: A within-person be- tween-jobs design. Journal of Applied Psychology, 90: 731–739.

Jehn, K. A., & Mannix, E. A. 2001.The dynamic nature of conflict: A longitudinal study of intragroup conflict and group performance. Academy of Management Journal, 44: 238–251.

Judge, T. A., & Ferris, G. R. 1993. Social context of per- formance evaluation decisions. Academy of Man- agement Journal, 36: 80–105.

Kenny, D. A., Kashy, D. A., & Bolger, N. 1998. Data analysis in social psychology. In D. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psy- chology (4th ed.), vol. 1: 233–265. New York: McGraw-Hill.

Kolodinsky, R. W., Treadway, D. C., & Ferris, G. R. 2007. Political skill and influence effectiveness: Testing portions of an expanded Ferris and Judge (1991) model. Human Relations, 60: 1747–1777.

Lewin, K. 1951. Field theory in social science. New York: Harper & Row.

Liden, R. C., Wayne, S. J., & Stilwell, D. 1993. A longitu- dinal study on the early development of leader- member exchanges. Journal of Applied Psychology, 78: 662–674.

MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. 2002. A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7: 83–104.

Miller, P. A., & Eisenberg, N. 1988. The relation of em- pathy to aggressive and externalizing/antisocial be- havior. Psychological Bulletin, 103: 324–344.

292 AprilAcademy of Management Journal

Mohammed, S., & Angell, L. C. 2004. Surface- and deep- level diversity in workgroups: Examining the mod- erating effects of team orientation and team process on relationship conflict. Journal of Organizational Behavior, 25: 1015–1039.

Mooney, C. Z., & Duval, R. D. 1993. Bootstrapping: A nonparametric approach to statistical inference. Newbury Park, CA: Sage.

Moore, L., Johns, G., & Pinder, C. C. 1980. Toward middle range theory: An overview and perspective. In C. C. Pinder & L. F. Moore (Eds.), Middle range theory and the study of organizations: 1–16. Hingham, MA: Martinus Nijhoff.

Olweus, D. 1978. Aggression in schools: Bullies and whipping boys. Washington, DC: Hemisphere.

Opotow, S. 1990a. Moral exclusion and injustice: An introduction. Journal of Social Issues, 46(1): 1–20.

Opotow, S. 1990b. Deterring moral exclusion. Journal of Social Issues, 46(1): 173–182.

Opotow, S. 1995. Drawing the line: Social categorization, moral exclusion, and the scope of justice. In B. Bun- ker & J. Rubin (Eds.), Conflict, cooperation, and justice: Essays inspired by the work of Morton Deutsch: 347–369. San Francisco: Jossey-Bass.

Opotow, S. 2001. Reconciliation in times of impunity: Challenges for social justice. Social Justice Re- search, 14: 149–170.

Opotow, S., & Weiss, L. 2000. Denial and the process of moral exclusion in environmental conflict. Journal of Social Issues, 56(3): 475–490.

Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsa- koff, N. P. 2003. Common method biases in behav- ioral research: A critical review of the literature and recommended remedies. Journal of Applied Psy- chology, 88: 879–903.

Pulakos, E. D., & Wexley, K. N. 1983. The relationship among perceptual similarity, sex, and performance ratings in manager-subordinate dyads. Academy of Management Journal, 26: 129–139.

Richman, J. A., Flaherty, J. A., & Rospenda, K. M. 1996. Perceived workplace harassment experiences and problem drinking among physicians: Broadening the stress/alienation paradigm. Addiction, 91: 391–403.

Richman, J. A., Flaherty, J. A., Rospenda, K. M., & Chris- tensen, M. 1992. Mental health consequences and correlates of medical student abuse. Journal of the American Medical Association, 267: 692–694.

Robinson, S. L., & Bennett, R. J. 1995. A typology of workplace deviant behaviors: A multidimensional scaling study. Academy of Management Journal, 38: 555–572.

Schat, A. C. H., Frone, M. R., & Kelloway, E. K. 2006. Prevalence of workplace aggression in the U.S. work- force: Findings from a national study. In E. K. Kel- loway, J. Barling, & J. J. Hurrell (Eds.), Handbook of

workplace violence: 47–89. Thousand Oaks, CA: Sage.

Schaubroeck, J., & Lam, S. K. 2002 How similarity to peers and supervisor influences organizational ad- vancement in different cultures. Academy of Man- agement Journal, 45: 1120–1136.

Shrout, P. E., & Bolger, N. 2002. Mediation in experimen- tal and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7: 422–445.

Skarlicki, D. P., Folger, R. G., & Tesluk, P. 1999. Person- ality as a moderator in the relationship between fair- ness and retaliation. Academy of Management Journal, 42: 100–108.

Skarlicki, D. P., & Latham, G. P. 2005. How can training be used to foster organizational justice? In J. Green- berg & J. A. Colquitt (Eds.), Handbook of organiza- tional justice: 499–522. Mahwah, NJ: Erlbaum.

Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh, A., & Kessler, S. 2006. The dimensionality of coun- terproductivity: Are all counterproductive behaviors created equal? Journal of Vocational Behavior, 68: 446–460.

Struch, N., & Schwartz, S. H. 1989. Intergroup aggression: Its predictors and distinctiveness from in-group bias. Journal of Personality & Social Psychology, 56: 364–373.

Tajfel, H., & Turner, J. C. 1979. An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup rela- tions: 94–109. Monterey, CA: Brooks/Cole.

Tepper, B. J. 2000. Consequences of abusive supervision. Academy of Management Journal, 43: 178–190.

Tepper, B. J. 2007. Abusive supervision in work organi- zations: Review, synthesis, and directions for future research. Journal of Management, 33: 261–289.

Tepper, B. J., Duffy, M. K., Henle, C. A., & Lambert, L. S. 2006. Procedural injustice, victim precipitation, and abusive supervision. Personnel Psychology, 59: 101–123.

Ting-Toomey, S. 1999. Communicating across cultures. New York: Guilford.

Trevino, L., K., Weaver, G. R., & Reynolds, S. K. 2006. Behavioral ethics in organizations: A review. Jour- nal of Management, 32: 951–990.

Tsui, A. S., & O’Reilly, C. A., III 1989. Beyond simple demographic effects: The importance of relational demography in superior-subordinate dyads. Acad- emy of Management Journal, 32: 402–423.

Turban, D. B, Dougherty, T. W., & Lee, F. K. 2002. Gen- der, race and perceived similarity effects in develop- mental relationships: The moderating role of rela- tionship duration. Journal of Vocational Behavior, 61: 240–262.

Turban, D. B., & Jones, A. P. 1988. Supervisor-subordi-

2011 293Tepper, Moss, and Duffy

nate similarity: Types, effects and mechanisms. Journal of Applied Psychology, 73: 228–234.

Turner, J. C. 1982. Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations: 15–45. Cambridge, U.K.: Cam- bridge University Press.

Wayne, S. J., & Liden, R. C. 1995. Effects of impression management on performance ratings: A longitudinal study. Academy of Management Journal, 38: 232– 260.

Williams, K. Y., & O’Reilly, C. A. 1998. Demography and diversity in organizations: A review of 40 years of research. In B. M. Staw (Ed.), Research in organiza- tional behavior, vol. 20: 77–140. Greenwich, CT: JAI Press.


Estimation of Regression Equations for Hypothesis Tests

The first equation estimates the direct effect of per- ceived dissimilarity on relationship conflict:

M � a0 � a1X � eM, (1)

where M is relationship conflict and X is perceived dissimilarity.

The second equation estimates the direct effects of perceived dissimilarity (X) and relationship conflict (M) on subordinates’ performance (Z):

Z � b0 � b1X � b2M � eZ. (2)

The third equation estimates the direct effects of per- ceived dissimilarity (X), relationship conflict (M), and subordinates’ performance (Z), as well as the interaction between relationship conflict and subordinate perfor- mance (MZ), on abusive supervision (Y).

Y � c0 � c1X � c2M � c3Z � c4MZ � eY. (3)

Substituting Equation 1 for M in Equations 2 and 3 and substituting Equation 2 for Z for in Equation 3 produces Equation 4:

Y � c0 � c1X � c2(a0 � a1X � eM) � c3(b0 � b1X�b2�a0

� a1X � eM� � eZ) � c4(a0 � a1X�eM) Z � eY. (4)

Rearranging the terms in Equation 4 produces the fol- lowing reduced-form equation:

Y � c0 � c2a0 � c3b0 � c3b2a0 � c4a0Z � X(c1 � c2a1

� c3b1 � c3b2a1 � c4a1Z) � c2eM�c3b2eM � c3ez � c4eM

� eY. (5)

Equation 5 shows that the effect of perceived dissimi- larity (X) on abusive supervision (Y) consists of a direct effect (c1), an indirect effect through relationship conflict (c2a1), an indirect effect through subordinates’ perfor- mance (c3b1), an indirect effect through relationship con- flict and subordinate performance (c3b2a1), and an indi- rect effect through relationship conflict that is moderated by subordinate performance (c4a1Z) . The examination of the four indirect effects in Equation 5 corresponds with the tests of Hypotheses 1, 2, 3, and 4, respectively.

Bennett J. Tepper ([email protected]) is a professor of managerial sciences in the J. Mack Robinson College of Business at Georgia State University. He received his Ph.D. in organizational psychology from the University of Miami. His current research interests include leader- ship, behavioral ethics, and psychological well-being.

Sherry E. Moss ([email protected]) is an associ- ate professor of organizational studies in the School of Business at Wake Forest University. She received her Ph.D. in organizational behavior and theory from Florida State University. Her research interests include attribu- tion theory, feedback exchanges and supervisor-subordi- nate relations.

Michelle K. Duffy ([email protected]) is a professor in the Carlson School of Management at the University of Minnesota. She received her Ph.D. from the University of Arkansas. Her research interests include social under- mining, moral disengagement, and affect and emotions at work.

294 AprilAcademy of Management Journal

Copyright of Academy of Management Journal is the property of Academy of Management and its content may

not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.