Employees Behavioral Reactions to Supervisor Aggression ... · Employees Behavioral Reactions to...

Employees’ Behavioral Reactions to Supervisor Aggression:An Examination of Individual and Situational Factors

Marie S. MitchellUniversity of Georgia

Maureen L. AmbroseUniversity of Central Florida

This research examines employees’ behavioral reactions to perceived supervisor aggression. The goal isto understand what makes employees react constructively or destructively to aggression. Three types ofbehavioral reactions are investigated: retaliation, coworker displaced aggression, and problem solving.We suggest employee reactions are influenced by individual and situational characteristics. We test theseideas by examining the moderating effects of 1 individual factor (locus of control) and 2 situationalfactors (fear of retaliation and behavioral modeling) on the relationships between perceived supervisoraggression and employee behaviors. The results of an experiment and 2 field studies provide support forthe predictions and some unexpected findings. Implications for understanding reactions to perceivedsupervisor aggression are presented.

Keywords: abusive supervision, aggression, retaliation, displaced aggression, problem solving

Today’s workplace is often portrayed as arduous and harsh, withorganizational authorities exploiting and abusing employees andemployees engaging in deceitful and insidious work behavior (e.g.,Fisher, 2005; Gurchiek, 2005; Pachter, 2003; Thelen, 2009). Re-cent reports from the Bureau of Justice Statistics (1998, 2005)support these contentions and have shown that incidents of inter-personal aggression (behavior that is carried out with the intent ofinjuring or aggravating another person; Eron, 1987) are on the rise.This is of particular concern, given the negative impact workplaceaggression has on organizations and employees. For example,research shows that workplace aggression impairs employee mo-rale, psychological health, and productive behaviors (e.g., Cortina,Magley, Williams, & Langhout, 2001; Glomb & Liao, 2003;Hershcovis & Barling, 2010; Neuman & Baron, 1998; Tepper,2000). Further, such acts diminish organizational capital and pro-duction and increase insurance premiums and injury compensation(Detert, Trevino, Burris, & Andiappan, 2007; Dunlop & Lee,2004). Given the prevalence and destructive nature of workplaceaggression, research in this area has grown in an effort to under-stand its antecedents and consequences.

One area of inquiry has emphasized consequences of supervi-sors who are aggressive with employees (see Tepper, 2007). Su-pervisor aggression is defined as employees’ perceptions of thesupervisor’s intentionally harmful behavior against them (Schat,Desmarais, & Kelloway, 2005).1 Understanding the impact of

supervisor aggression is important because supervisors are consid-ered a primary source of aggression at work (e.g., Cortina et al.,2001; Neuman & Keashly, 2003). It also represents a particularlystressful, challenging, and threatening situation for employeesbecause supervisors maintain control over things that are of valueto employees’ work lives (e.g., resources, feedback; Tepper,2007). Supervisors who are aggressive undermine subordinates’ability to function effectively at work (Harris, Kacmar, & Ziv-nuska, 2007; Tepper, 2000) and threaten subordinates’ sense ofself (Thau & Mitchell, 2010) and home life (Carlson, Ferguson,Perrewe, & Whitten, 2011; Hoobler & Brass, 2006). Indeed, re-search has shown that supervisor aggression generally provokesquite destructive reactions from employees (e.g., Inness, Barling,& Turner, 2005; Mitchell & Ambrose, 2007; Tepper, Henle, Lam-bert, Giacalone, & Duffy, 2008), with only a handful of studiesshowing employees’ constructive reactions (e.g., Keashly, Trott, &MacLean, 1994; Tepper, Duffy, & Shaw, 2001; Tepper, Moss,Lockhart, & Carr, 2007).

Yet, what makes employees react destructively versus construc-tively is not entirely clear. That is, research shows that of victimsof supervisor aggression, not all act destructively and not all actconstructively (Bies & Tripp, 1996, 1998). Given that workplaceaggression is escalating (Bureau of Justice Statistics, 1998, 2005)and employees identify supervisors as a dominant source of ag-gression (Cortina et al., 2001; Neuman & Keashly, 2003), exam-ination of which factors enhance or mitigate different reactions tosupervisor aggression is needed. The impact of supervisor aggres-sion is significant; it is estimated to cost U.S. corporations billions

1 Supervisor aggression can be considered akin to abusive supervision(see Hershcovis et al., 2007; Tepper, 2007). Both constructs involve thesupervisor being perceived by subordinates as engaging in hostilitiestoward them. However, abusive supervision, as defined by Tepper (2000),includes sustained hostile behavior by the supervisor. Definitions of su-pervisor aggression do not emphasize this ongoing hostility. This distinc-tion is important for our operationalization in Study 1.

This article was published Online First July 30, 2012.Marie S. Mitchell, Department of Management, University of Georgia;

Maureen L. Ambrose, Department of Management, University of CentralFlorida.

We thank Thomas Tripp and Marshall Schminke for their help onprevious versions of this article.

Correspondence concerning this article should be addressed to Marie S.Mitchell, Department of Management, Terry College of Business, Univer-sity of Georgia, 425 Brooks Hall, Athens, GA 30621. E-mail: [email protected]

Journal of Applied Psychology © 2012 American Psychological Association2012, Vol. 97, No. 6, 1148–1170 0021-9010/12/$12.00 DOI: 10.1037/a0029452

1148

of dollars annually (Tepper, Duffy, Henle, & Lambert, 2006).Additionally, understanding how organizations can minimize de-structive and enhance constructive behavior is of practical valuefor decision makers. Doing so may reduce costs of destructivework behavior and potentially foster an effective and productivework environment for employees.

Our purpose in this article is to develop and test a model ofemployees’ behavioral reactions to perceived supervisor aggres-sion. We investigate three types of reactions: retaliation (aggres-sive reactions against the supervisor), displaced aggression (ag-gressive reactions against coworkers), and problem solving(constructive efforts made to address the supervisor aggression).Research and theory suggest each of these types of reactions occur,but little research attention has been given to understanding whenvictims are likely to respond to supervisor aggression destructivelyversus constructively. Thus, identifying moderators is an essentialnext step (Tepper, 2007). To this end, our research identifies andtests one individual factor (locus of control [LOC]) and twosituational factors (fear of retaliation and behavioral modeling)that theory (e.g., social learning theory, Bandura, 1973;frustration–aggression theory, Dollard, Doob, Miller, Mowrer, &Sears, 1939) suggests influence reactions to perceived threats, inparticular, aggression from individuals in authority.

Our research provides an extension to the literature and isalso practically significant. First, our work meets the call toexamine various boundary effects of employees’ reactions tosupervisor aggression (Tepper, 2007). Because research hasshown not all victims of supervisor aggression react in the sameway (Bies & Tripp, 1996, 1998), understanding what motivatesdestructive and constructive reactions is needed. Our workextends previous research by considering factors that heightenemployees’ sense of control (i.e., LOC, fear of retaliation),which allows for more constructive ways of responding to anaggressive supervisor. Additionally, we theorize that there arework factors that increase employees’ salience about the typesof behaviors that are accepted in the work environment andinvoke behavioral learning (i.e., behavioral modeling, fear ofretaliation) that influences employees’ reactions. Second, byconsidering constructive reactions (i.e., problem solving), ourresearch extends current workplace aggression models, whichhave traditionally focused on the negative consequences ofaggression to organizations and its members (e.g., Baron, 2004;Hershcovis & Barling, 2010; Hershcovis et al., 2007; Neuman& Baron, 1998). Third, previous research has primarily exam-ined reactions to supervisor aggression via survey methodol-ogy. Our research uses both experimental and survey ap-proaches. The enhanced internal validity of the experimentaldesign complements the dominant survey method. Finally, theresults of our study may assist organizations in lessening de-structive work behavior. Supervisor aggression is a growingproblem for organizations and employees (cf. Tepper et al.,2006). By identifying factors that influence employees’ reac-tions to supervisor aggression, organizations gain an under-standing of how to foster more constructive behaviors andlessen destructive behaviors of their employees. Below, wereview likely responses to supervisor aggression and providethe theoretical foundation for our predictions.

Theoretical Overview

Behavioral Reactions to Perceived SupervisorAggression

Retaliation. Theory and research suggest retaliation is aprimary response to aggression. A basic tenet of the frustration–aggression theory (Dollard et al., 1939) is that aggression provokesretaliation. This proposition is supported by social psychologyresearch (see Anderson & Bushman, 2002, for a review). In theorganizational literature, theorists have argued that retaliation al-lows victims to get back at the transgressor (e.g., Mitchell &Ambrose, 2007; Skarlicki & Folger, 2004). Negative reciprocityprinciples of social exchange theory (Gouldner, 1960) provide thebasis of this claim. Reciprocity acts as a behavioral norm, guidingpatterns of interactions. Negative reciprocity patterns suggest vic-tims of mistreatment will return mistreatment to the harmdoer.Research supports these arguments and has shown that employeesretaliate against supervisors for perceived abuse (e.g., Dupre, In-ness, Connelly, Barling, & Hoption, 2006; Inness et al., 2005;Mitchell & Ambrose, 2007; Tepper et al., 2008, 2009; Thau,Bennett, Mitchell, & Marrs, 2009).

Displaced aggression. Even though retaliation is a commonresponse to aggression, some situations do not allow for it. In theseinstances, the frustration–aggression theory (Dollard et al., 1939)suggests, individuals displace their frustration on other targets.Displaced aggression is explained as a cathartic act—one thatallows individuals to vent frustration when aggressing against thetransgressor is not a viable option. Marcus-Newhall, Pedersen,Carlson, and Miller’s (2000) meta-analysis of psychology exper-iments found displaced aggression is a strong reaction. Further,organizational research has shown that employees displace aggres-sion on targets other than the supervisor when victimized bysupervisor aggression (Hoobler & Brass, 2006; Mitchell & Am-brose, 2007). Our research examines displaced aggression targetedtoward victims’ peers and coworkers.

Problem solving. Both retaliation and displaced aggressionare destructive responses to supervisor aggression. Yet, theory andresearch suggest not all employees react destructively (Bies &Tripp, 1996, 1998). For instance, reactance theory (Wortman &Brehm, 1975) posits that both destructive and constructive reac-tions may occur because people strive to maintain a sense ofpersonal control. Constructive responses, such as problem solving,allow victims to regain control by trying to resolve the situation(Weisz, McCabe, & Denning, 1994). Lazarus and colleagues’(Lazarus, 1966; Lazarus & Folkman, 1984) research on stresscoping strategies also suggests that threatening events (i.e., super-visor aggression) stimulate evaluations on how to best cope withthe harm and make problem solving possible. Indeed, research hasshown that some victims of supervisor aggression react withproblem solving (e.g., reconciliation, Aquino, Tripp, & Bies, 2001,2006; constructive resistance strategies, Tepper et al., 2001, 2007).

In sum, employees may respond to supervisor aggression withretaliation, displaced aggression, and/or problem solving. But whatdetermines the behavior in which an individual engages? Socialcognitive theory (Bandura, 1986) suggests reactions to perceivedaggression are dependent on individual and situational factors (cf.,Anderson & Bushman, 2002; Bandura, 1991; Baron, 2004). Wereview theory and research that suggest locus of control (LOC; an

1149REACTIONS TO SUPERVISOR AGGRESSION

individual factor) and the situational factors fear of retaliation andbehavioral modeling should affect employees’ reactions to super-visor aggression. We expect these factors will moderate the rela-tionship between supervisor aggression and employee destructiveand constructive reactions.

LOC as a Moderator of the SupervisorAggression–Employee Behavior Relationship

Our first moderating factor is LOC—a trait that involves theextent to which individuals believe that events are contingent ontheir own behavioral control or are determined by others, fate, orchance (Rotter, 1966). LOC has a strong influence over individu-als’ behavioral choices because it impacts the degree to whichindividuals believe they have the ability to control events affectingthem. Individuals with a high LOC, called internals, are particu-larly self-determined and motivated to control their social envi-ronment in ways that minimize threats to and maximize benefitsfor themselves (Deci & Ryan, 1980). Individuals with a low LOC,called externals, believe other sources affect what happens tothem; consequently, they do not believe they have the ability toexact change that will effectively alter their environment (Deci &Ryan, 1980; Ryan & Deci, 2006).

Social cognitive theory (Bandura, 1986) suggests personalitytraits that involve individuals’ sense of personal control influencetheir reactions to perceived aggression. Similarly, models of orga-nizational behavior that theorize about individuals reactions toperceived work threats highlight the importance of consideringLOC (e.g., Bennett, 1998; Martinko & Gardner, 1982; Martinko,Gundlach, & Douglas, 2002).

Additionally, existing theory and research on personal controlsuggest internals are better able to create and shape more favorablework experiences (Phares, 1976) and handle challenging andthreatening situations (Gatchel, 1980; Wanberg, 1997). Fiske andTaylor (1984) argued this is because internals see themselves as“causal agents” with the ability to change situations directly. Thus,when faced with threatening situations, internals attempt to reduceperceived threats by engaging in behavior to change the environ-ment (see Ng, Sorensen, & Eby, 2006 and Perrewe & Spector,2002, for reviews). In contrast, externals, guided by futility beliefs,are less able to see opportunities to control threatening situationsand, consequently, tend to act ineffectively and even destructivelyin the face of a threat. Their external orientation combined with thethreat from the environment (i.e., supervisor aggression) creates asense of helplessness, which promotes deviant behavior (Martinko& Gardner, 1982). Theorists contend that externals’ destructivebehavior reduces the strain associated with their perceived inabil-ity to change their current situation (Bennett, 1998; Greenberger &Strasser, 1991).

Given the motivational force of LOC in guiding individuals’behaviors in threatening situations, we believe LOC is an impor-tant trait to consider with regard to reactions to supervisor aggres-sion. Supervisors who aggress against their employees pose asignificant threat to employees’ work lives (Tepper, 2007). Theseare individuals who yell, scream, undermine, intimidate, and ver-bally abuse their subordinates (Keashly, 1998; Keashly et al.,1994). Given the threat that supervisor aggression poses to subor-dinates and the motivational effects of LOC, we predict LOC willimpact employees’ reactions to supervisor aggression.

Research suggests that LOC influences individuals’ reactions tosupervisor aggression. For example, studies have shown that in-ternals are more likely than externals to engage in problem solvingbehavior (Spector, 1982). Research has shown that internals, whenchallenged, exert more effort and persist longer in situations thatpresent barriers to achieving goals than do externals (Phares,1976). Further, internals are more likely than externals to engagein more problem-focused activities (Hahn, 2000). Research hasalso shown that internals are more likely than externals to engagein problem solving behavior when they experience a high degreeof stress (Arslan, Dilmac, & Hamarta, 2009; Parkes, 1984). Last,meta-analytic evidence (Ng et al., 2006) has shown that internalsare better equipped to deal with negative tasks and negative socialinteractions and are more likely than externals to engage in prob-lem solving.

Research has shown that externals are less likely than internalsto engage in behavior that attempts to effectively resolve threat-ening situations and, instead, are more likely to behave destruc-tively (Fox & Spector, 1999; Marcus & Schuler, 2004; Ostermanet al., 1999; Romi & Itskowitz, 1990; Storms & Spector, 1987;Zainuddin & Taluja, 1990). For example, Fox and Spector (1999)found externals are more likely than internals to engage in behav-ior that is counterproductive. Further, Wei and Si (2011) found thatexternals were more likely to respond to abusive supervision (aform of supervisor aggression) with destructive work behavior(i.e., sabotage, production deviance, theft).

Overall, this review suggests LOC will influence both construc-tive and destructive reactions to supervisor aggression. Becauseinternals believe they can exact change in their environment,particularly when dealing with stressful and threatening situations,we contend they are more likely to engage in problem solvingbehavior when they encounter an aggressive supervisor. Con-versely, because externals are more likely to react destructively tostressful and threatening events, we predict they will be morelikely to engage in retaliation and coworker displaced aggressionwhen dealing with an aggressive supervisor. In sum, a higher LOCwill influence problem solving reactions and a lower LOC willinfluence destructive reactions to supervisor aggression.

Hypothesis 1: The positive relationship between supervisoraggression and (a) problem solving will be stronger whenlocus of control is higher than lower, and the positive rela-tionships between supervisor aggression and (b) retaliationand (c) coworker displaced aggression will be stronger whenlocus of control is lower than higher.

Situational Moderators of the SupervisorAggression–Employee Behavior Relationship

We draw from social psychology theory and research to identifysituational factors that moderate the effects of perceived supervisoraggression on employee behavioral reactions. In particular, weconsider two factors that research and theory suggest influencereactions to aggression: fear of retaliation and behavioral model-ing.

Fear of retaliation. Aggression theories suggest fear of re-taliation from a harmdoer influences victims’ responses (e.g.,frustration–aggression theory, Dollard et al., 1939; social learningtheory, Bandura, 1973). Fear of retaliation influences reactions to

1150 MITCHELL AND AMBROSE

perceived aggression because of victims’ learned inhibitions (Ban-dura, 1973; Berkowitz, 1983; Sears, 1948). People understand theconsequences of their behavior based on past experiences or bywatching what happens to others. Victims who fear retaliationfrom a harmdoer believe aggressive responses will likely provokefurther acts of aggression against them, particularly when theharmdoer has greater power (e.g., a supervisor of a subordinate;Bandura, 1973; Baron, 1971; Berkowitz, 1983; Sears, 1948). Incontrast, problem solving will likely deter further harm againstvictims or potentially resolve the situation altogether. Conse-quently, victims tend to refrain from aggressing against harmdoerswhen they fear retaliation and, instead, tend to problem solve to tryto resolve the issue.

Few studies in the organizational sciences have specificallyexamined fear of retaliation, but existing studies provide evidenceconsistent with our predictions. For example, Fox and Spector(1999) found when employees believed their organization causedthem harm and they could engage in deviant behavior withoutpenalty (i.e., no fear of retaliation), they engaged in more deviantwork behavior. Similarly, Tripp and Bies’s (1997) qualitativestudy suggests one of the most consistent deterring factors ofretaliatory behaviors among would-be avengers was fear of retal-iation from the harmdoer. Research has also shown that someindividuals who fear a person in authority engage in problemsolving. Some victims try to talk to the offender to resolve theissue (Bies & Tripp, 1996, 1998). Others try to reconcile with theoffender (Aquino et al., 2001, 2006), and others simply reportthe offender to authorities (Bies & Tripp, 1996).

This review suggests fear of retaliation will moderate the effectsof supervisor aggression on retaliation (i.e., supervisor-directedaggression) and problem solving such that higher fear of retaliationwill strengthen the effects on problem solving, but it will weakenthe effects on retaliation.

Hypothesis 2: The positive relationship between supervisoraggression and (a) problem solving will be stronger when fearof retaliation is higher than lower, and the positive relation-ship between supervisor aggression and (b) retaliation will bestronger when fear of retaliation is lower than higher.

Although research shows a fairly clear influence of fear ofretaliation on behavior directed at the aggressor, two primary, yetopposing, views have been proposed for its moderating effects ondisplaced aggression reactions to aggression: the frustration–aggression view (Dollard et al., 1939) and the learned inhibitionview (Bandura, 1973; Tedeschi & Felson, 1994). The frustration–aggression view (Dollard et al., 1939) suggests when victims fearretaliation from an aggressor, displaced aggression is likely be-cause it allows victims to vent hostilities without fear of recoursefrom the harmdoer (Dollard et al., 1939). Dollard et al. (1939)explained that displaced aggression acts as a form of catharsis,allowing victims to express their frustration behaviorally. Becausefear of retaliation inhibits retaliatory reactions, victims feel theneed to express their frustration by displacing aggression on other,more available targets. As previously noted, social psychologyresearch provides evidence of these effects (see Berkowitz, 1989;Marcus-Newhall et al., 2000).

Yet, social psychologists (Tedeschi & Felson, 1994) have notedthe frustration–aggression view of displaced aggression has not

been consistently supported. The frustration–aggression predictionabout displaced aggression is based on the notion of catharsis(individuals releasing negative and aggressive energy without fearof future recourse). However, both Miller (1941) and Sears (1941)contended that individuals may choose not to displace aggressionif learned inhibitions suggest it would be inappropriate. Similarly,Tedeschi and Felson (1994) argued that displaced aggression doesnot always occur because past experiences may allow individualsto discriminate when to and when not to displace aggression.These arguments are consistent with Bandura’s (1973) view thatindividuals refrain from destructive reactions when learned inhi-bitions suggest they are not appropriate. Thus, according to thisperspective, the target of the displaced aggression is important. Itis, then, possible for victims of supervisor aggression to notdisplace aggression on their peers because they understand thatdoing so may tarnish the relationships these victims hold with theircolleagues. Additionally, employees who act aggressively towardpeers (even if their acts are displaced) may be viewed by super-visors as dysfunctional work unit members, which potentiallyprovokes the supervisor to be even more hostile with that em-ployee. In short, the learned inhibitions view suggests that engag-ing in displaced aggressive behaviors would not prove wise forsomeone who is the target of supervisor aggression (Tedeschi &Felson, 1994).

The frustration–aggression and learned inhibitions views bothprovide plausible explanations for how fear of retaliation impactsthe relationship between supervisor aggression and coworker dis-placed aggression. Consequently, we pose competing hypotheses:

Hypothesis 3(a): According to the frustration–aggressionview, the positive relationship between supervisor aggressionand coworker displaced aggression will be stronger when fearof retaliation is higher than lower.

Hypothesis 3(b): According to the learned inhibitions view,the positive relationship between supervisor aggression andcoworker displaced aggression will be weaker when fear ofretaliation is higher than lower.

Behavioral modeling. Our final hypotheses examine behav-ioral modeling. Social learning theory (Bandura, 1973; Mischel,1973) and social information processing theory (Salancik & Pfef-fer, 1978) suggest individuals understand what is acceptable be-havior through vicarious learning and, in particular, by observingrelevant social models (e.g., coworkers, peers). Watching othersengage in certain behaviors conveys that the observed behavior isacceptable and supported in the social environment, and, hence,makes individuals more inclined to engage in similar behaviors.Bandura (1973, 1986) argued that when individuals perceive rel-evant social models engaging in certain behavior (whether it isconstructive or destructive), they view that behavior as sociallyappropriate. Indeed, research suggests employees learn about ex-pected work behavior more by watching what coworkers do thanby following formal rules and procedures (Salancik & Pfeffer,1978).

This line of reasoning suggests individuals imitate their peers’behaviors, in a “monkey-see, monkey-do” pattern (cf. Robinson &O’Leary-Kelly, 1998). That is, employees who perceive that co-workers display aggressive behaviors will be more likely to engage


in aggressive behavior (i.e., retaliation, displaced aggression), andemployees who perceive that coworkers display problem solvingbehaviors will be more likely to engage in problem solving.Research supports these ideas. For example, employees are morelikely to react to aversive work conditions with absenteeism ifbehavioral norms among peers suggest absenteeism is permissive(Biron & Bamberger, 2012). Research also shows that employees’observations of coworkers exhibiting antisocial behavior influencethe employees’ own antisocial behavior (Aquino & Douglas, 2003;Robinson & O’Leary-Kelly, 1998). Likewise, employees who area target of aggression, who also observe aggression by coworkers,are more likely to act aggressively (Glomb & Liao, 2003).

Similar influence is found for constructive behavior. Researchhas shown that employees who perceive group norms embracecooperative and helping behavior are more likely to engage insimilar behaviors (De Cremer, van Dijke, & Mayer, 2010; Ehrhart& Naumann, 2004) and that perceptions of group norms thatembrace open discussion about conflict influence individuals toengage in more open discussion (Jehn, 1995). Research has alsodemonstrated that observations of peers’ innovative and creativebehavior positively influence observers’ innovative and creativebehavior (Adarves-Yorno, Postmes, & Haslam, 2007). Finally,research has also shown that problem solving behavioral modeling(perceptions of others engaging in problem solving or of behav-ioral norms associated with problem solving) influences the per-ceiver’s problem solving behavior (Levine, Higgins, & Choi,2000; Taggar & Ellis, 2007).

We suggest coworkers provide a strong basis for behavioralnorms, and the type of behaviors coworkers exert at work willinfluence victims’ reactions to supervisor aggression. Individualswho perceive their coworkers engaging in aggressive or problemsolving behavior will find each type of behavior to be acceptableand supported. Consequently, when confronted with supervisoraggression, employees who perceive coworkers modeling aggres-sive behavior will be more inclined to engage in similar behavior(i.e., retaliation, displaced aggression) and employees who per-ceive coworkers modeling problem solving behavior will be moreinclined to engage in problem solving. We predict

Hypothesis 4: The positive relationships between supervisoraggression and (a) retaliation and (b) coworker displacedaggression will be stronger when aggressive modeling ishigher than lower, and the positive relationship between su-pervisor aggression and (c) problem solving will be strongerwhen problem solving modeling is higher than lower.

We tested the predictions in three studies. Study 1 is an exper-imental design. The experiment examined LOC and fear of retal-iation as moderators on the effects of perceived aggression from aninstructor (the in-class equivalent to workplace supervisor aggres-sion) on two behavioral reactions: retaliation and displaced aggres-sion. Study 2 is a cross-sectional field survey design that examinesthe moderating effects of LOC, fear of retaliation, and aggressivemodeling on retaliation, coworker displaced aggression, and gen-eral problem solving reactions to supervisor aggression. Study 3 isa time-separated field survey study that provides a test of all of ourpredictions and investigates the effects of supervisor aggressionnot only on retaliation and coworker displaced aggression reac-tions but also on specific forms of problem solving (i.e., seeking

social support from coworkers, reconciling with the supervisor,and reporting the supervisor to authorities).

Study 1 Method

Participants and Study Design

Participants were undergraduate students from a large southeast-ern university. The experiment was masked within the coursecurriculum of eight business classes, and the experimental expe-rience was used as a learning exercise in class. At the beginning ofthe semester, students were given an informed consent and apersonality survey. The researcher explained that the personalitysurvey and other class activities would be used for research, withthe student’s consent. Of the 273 students who initially consented,242 participated in the study; 31 students were either absent or lateto class the day of the experiment. To maintain the integrity of theexperiment, late students could not enter the classroom until thedebriefing. Participants were on average 23.3 years old (SD �5.52); 43.2% were female and 61.2% were White (10.6% wereBlack; 13.9% were Hispanic). A 2 � 2 (aggression high/low; fearof retaliation high/low) design was used. Participants were ran-domly assigned to each condition. Approximately an equal numberof participants were in each condition.

Procedure

The experiment took place the week the course’s first gradedassignment was to be distributed to students. The day of theexperiment, after students were seated for class, a confederateentered the classroom. The confederate explained that the instruc-tor could not make it to class that day. However, the instructorasked the confederate to come to the class to pass out the students’graded assignments and ask them to complete a team evaluationform. In addition, the confederate explained that she or he was alsothere on behalf of the management department, which was con-ducting a midyear check on the department’s services and instruc-tors. The confederate read and distributed a letter by the manage-ment department chair, asking the students to help by completingan instructor evaluation and a short management department ser-vices survey.2

The confederate first distributed the graded assignments. Ag-gression was manipulated with instructor comments written on thegraded assignments. Pretesting provided a pool of eight high-

2 The graded assignment that was distributed to students by the confed-erate was an independent student assignment that was required for thecourse. Students expected to receive their graded assignments that partic-ular day in class. The course also required students to work in a teamthroughout the semester, and, consequently, the instructor had teams rou-tinely complete team evaluations. We used the same team evaluation formin the experiment. Therefore, receiving the feedback on the graded assign-ment and having students complete a team evaluation was not unusual forthe students. Additionally, the letter by the chair of the managementdepartment (explaining the midsemester instructor and departmental eval-uation) was signed by the actual chair of the management department ofthis university. These situational aspects enhanced experimental realism forthe students.


aggression comments and eight low-aggression comments.3 Fourhigh-aggression comments (e.g., “I’m not impressed–maybe it’syour lack of talent”) or four low-aggression comments (e.g., “Av-erage”) were written throughout the assignment, depending on thestudents’ condition (high/low). The four comments were randomlyselected from the set of eight to ensure not every graded assign-ment had the exact same comments, thereby enhancing the ma-nipulation’s realism.

Ten minutes after the assignments were distributed, the confed-erate collected them and passed out the instructor evaluation (i.e.,retaliation assessment) and team peer evaluation forms (i.e., dis-placed aggression assessment). The fear of retaliation manipula-tion was set up at this time.4 Students in the high fear of retaliationcondition were instructed to write their name on the instructorevaluation form. Further, students were given a letter from thedepartment chair; the confederate told them, “This letter is fromthe department chair, explaining that your instructor will receive acopy of your evaluation within the next day or so.” Students in thelow fear of retaliation condition did not write their name on theform and were given a different letter from the department chair.The confederate explained, “This letter is from the departmentchair, assuring you that your instructor will never see your com-pleted evaluations—only administration will see your responses.”Instructor and peer evaluation forms were coded to match partic-ipants’ responses.

Following the instructor and team evaluations, participantsreceived the department evaluation (i.e., manipulation checksassessment). Once completed, the confederate thanked them fortheir time, asked them to wait one minute more, and left theroom. The confederate reentered the room with the instructorand began the debriefing. The confederate followed standardprotocol and convention for debriefing participants in experi-ments using deception: Participants were informed about thetrue nature of the experiment and informed about the use ofdeception and why deception was needed. Continued discussionwas provided to minimize negative feelings and emphasize theapplicability of the activity to a work situation (Greenberg &Folger, 1988).5 In particular, the confederate gently approachedthe topic by thanking the students for completing the evalua-tions and surveys and asked them if they had any concernsregarding the activities for the day or any disbelief about theactivities or the procedures. None of the students indicatedsuspicion.6 The confederate then explained the true nature ofthe experiment and explained that none of the comments writtenon the assignments reflected the instructor’s thoughts or feel-ings. The confederate informed the students that they hadexperienced an active learning exercise about workplace ag-gression and apologized for using deception, explaining thatdoing so was important to invoke true reactions associated withthe phenomenon. As the debriefing continued, the confederateand instructor used the activity to highlight the applicability ofthe course material on workplace aggression. Students engagedin lively discussions and talked about incidents of aggressionthat they had either experienced or witnessed from an authorityfigure. The confederate and instructor facilitated discussions onhow students should deal with these situations effectively andconstructively.

3 Consistent with the literature (e.g., Bushman & Baumeister, 1998; D.Cohen, Nisbett, Bowdle, & Schwarz, 1996; Porath & Erez, 2007; Sinaceur,Van Kleef, Neale, Adam, & Haag, 2011), aggression was manipulated byproviding participants negative and insulting feedback. A pilot study wasconducted to identify high and low aggressive instructor comments for theaggression manipulation. Participants were drawn from the same population asin the main study. Students were asked to imagine they had received the listedcomments from an instructor on a graded assignment. They were asked to ratethe comments on a 7-point scale in terms of their constructiveness (defined asstatements that are helpful, considerate, supportive, and provide useful feed-back on your work) or destructiveness (defined as statements that are discour-aging, threatening, pessimistic, and might be viewed as a personal attack; 1 �very constructive, 4 � neutral [neither constructive nor destructive], 7 �destructive). Eight high aggression comments (“This answer is a joke,” “My3-year-old niece could have written a better answer,” “Could you have writtenabout any more nonsense,” “I can’t believe you think this is college levelwork,” “This is simply moronic,” “Is this your answer?!” “I don’t see a careerin management,” “I’m not impressed—maybe it’s your lack of talent”) andeight low aggression comments (“Ok,” “Suitable,” “Adequate,” “Alright,”“Fine,” “Average,” “Sufficient,” and a check mark [√]) were produced.

4 A second pilot test was conducted to test the perceived aggression and fearof retaliation manipulations. Participants were drawn from the same populationas in the main study. First, to assess the perceived aggressiveness of thecomments, we asked participants to imagine they had received a gradedassignment from an instructor; looking through it, they noticed a variety ofinstructor comments. Four randomly selected comments for each condition(high and low aggression) were listed, and participants were asked to reflect onthe comments and respond to questions that served as a manipulation check forthe aggressiveness of the comments. Second, as a test of the fear of retaliationmanipulation, participants were asked to imagine that after receiving theassignment, they were asked to evaluate the instructor’s performance. The highfear of retaliation condition instructions indicated students were required towrite their name and the instructor’s name on their evaluation form. The lowfear of retaliation condition instructions indicated the evaluation would be foradministrative purposes only and the instructor would never see the student’sresponses. Participants then responded to questions that serve as a manipula-tion check for fear of retaliation. The results indicated participants rated thecomments as less aggressive in the low aggression condition than the high aggres-sion condition (M � 3.43 vs. 5.81; F � 84.09, p � .001), and perceptions of fearof retaliation were lower in the low fear of retaliation condition than the high fearof retaliation condition (M � 2.08 vs. 4.55; F � 69.57, p � .001).

5 Beyond the careful debriefing protocol, we took care to design the exper-iment to deal with the use of deception. For example, the researchers main-tained contact with the course instructors to ensure that the students and theinstructor/student relationships were continuing normally. Additionally, nostudents reported discomfort or anything associated with the experimentalactivity to the university’s institutional review board, administration, the in-structor, or the researchers. Last, as a safeguard for the instructors, the re-searchers placed a letter in each instructor’s personnel file, indicating that anexperimental activity on workplace aggression occurred in the class for aca-demic and learning purposes and that if an instructor had received uncharac-teristically low evaluation scores, it might be the consequence of the activity. Noneof the participating instructors received poor end-of-the-year evaluation scores.

6 To further assess whether students were suspicious about the experiment,we evaluated answers in an open-ended question, embedded in the post-experimental questionnaire (i.e., the departmental services survey). In partic-ular, students were asked to indicate whether they had discussed the question-naires they completed that day with anyone prior to entering the classroom orwhether they had any concerns regarding doing so. None of the studentsindicated that they had heard about the exercise or found the procedures orevaluations to be suspect. Additionally, neither during the experimental activ-ity nor during our debriefing discussions did students indicate suspicion.


Measures

Reactions to perceived aggression. Consistent with the lit-erature (e.g., DeWall, Baumeister, Stillman, & Gailliot, 2007;DeWall, Twenge, Gitter, & Baumeister, 2009; Twenge, Baumeis-ter, Tice, & Stucke, 2001), retaliation and displaced aggressionwere assessed by asking participants to evaluate the performanceof the focal target (i.e., the instructor’s performance for retaliation,team’s performance for displaced aggression), with poor evalua-tion scores reflecting an aggressive reaction. Retaliation was as-sessed with a 20-item instructor evaluation that was used to assessinstructor class performance at this university (e.g., “The instruc-tor’s use of class time,” “The instructor’s organization for thecourse”). Students rated the instructor on a 5-point scale (1 �excellent, 5 � poor). Displaced aggression was assessed with a10-item team evaluation form that was used in the course through-out the semester (e.g., “The team’s work is of high quality,”“Members contribute equally to team assignments”). Studentsrated their team on a 5-point scale (1 � all the time, 5 � never).

LOC. LOC was assessed with the 10-item Interpersonal LOCsubscale of the Spheres of Control measure (Paulhus, 1983).Participants rated their expectations of outcomes in interpersonalsituations (e.g., “I can usually achieve what I want if I work hardfor it,” “I can learn almost anything if I set my mind to it”) on a7-point Likert scale (1 � strongly disagree, 7 � strongly agree).High scores indicate an internal LOC.

Controls. Research suggests outcomes influence reactions tomistreatment (Greenberg, 1993). Thus, grades were not manipu-lated; grades written on the assignment were the grades studentsreceived. Further, grades were controlled for in the analysis usinga 100-point scale. Because different instructors’ classes were used,instructors were coded and controlled for in the analysis.

Manipulation checks. The manipulation checks were em-bedded in the management department services questionnaire. Afive-item perceived aggression manipulation check asked studentsto evaluate their last graded assignment (“I felt that the commentswritten by the instructor were highly insensitive”; � � .91). Afour-item fear of retaliation manipulation check asked respondentsto rate the procedure for the instructor evaluation (“I was fearfulthat my instructor would get back at me in some way in theremainder of the course”; � � .89). Participants rated responses on

a 5-point Likert-like scale (1 � strongly disagree, 5 � stronglyagree).

Study 1 Results

Manipulation Checks

The manipulation checks showed the experimental conditionshad the intended effects. Perceptions of aggression were lower inthe low aggression condition than the high aggression condition(M � 2.25 vs. 3.96; F � 227.69, p � .001); perceptions of fear ofretaliation were lower in the low fear of retaliation condition thanthe high fear of retaliation condition (M � 1.44 vs. 3.08; F �247.36, p � .001).

Hypotheses Testing

Table 1 provides the descriptive statistics, correlations, andreliability coefficients for the study’s variables. Table 2 displaysthe moderated multiple regression results. Study 1’s design al-lowed for only a test of predictions for the moderating effects ofLOC and fear of retaliation on retaliation and displaced aggres-sion; effects for behavioral modeling and those on problem solvingbehavior were not tested. The measured variable, LOC, was meancentered to assist in the interpretation of the interactions, andsignificant interactions were plotted and tested at one standarddeviation above and below the mean (J. Cohen, Cohen, West, &Aiken 2003).

Hypothesis 1 predicts lower LOC will strengthen the effects ofperceived aggression on (H1b) retaliation and (H1c) displacedaggression. The interaction was significant on retaliation but noton displaced aggression. Consistent with expectations, the plot inFigure 1 shows the effects on retaliation were stronger when LOCwas lower (b � .87, p � .001) than higher (b � .45, p � .01). Theresults support the prediction for retaliation (H1b) but not fordisplaced aggression (H1c).

Hypothesis 2 predicts lower fear of retaliation will strengthenthe relationship between aggression and retaliation (H2b). Table 2shows the interaction was significant. Table 3 reports the meansand standard deviations for the aggression and fear of retaliationmanipulations. Figure 2 depicts the effects and shows the mean of

Table 1Study 1 Descriptive Statistics and Correlations

Variable M SD 1 2 3 4 5 6 7 8 9 10

1. Aggression (0 � low) 0.52 0.50 —2. Fear of retaliation (0 � low) 0.47 0.50 �.02 —3. Locus of control 5.71 0.69 �.08 .06 (.79)4. Retaliatory aggression 2.63 1.22 .16 �.19 .01 (.97)5. Displaced aggression 1.27 0.37 .02 �.05 .05 �.27 (.83)6. Exam grade 85.15 13.83 �.13 �.19 .00 �.10 .05 —7. Instructor 1 0.22 0.42 .04 .06 .08 .65 �.39 �.14 —8. Instructor 2 0.11 0.31 �.02 .37 .04 �.39 .03 .11 �.19 —9. Instructor 3 0.35 0.48 .01 �.21 �.02 �.18 �.01 .08 �.40 �.26 —

10. Instructor 4 0.13 0.33 �.04 �.36 �.03 .01 .11 .24 �.20 �.13 �.28 —11. Instructor 5 0.13 0.34 �.03 .41 .00 �.24 .19 �.29 �.21 �.14 �.28 �.15

Note. The phi coefficient is reported for aggression and fear of retaliation. Point-biserial correlations are reported for all dichotomous variables, andzero-order correlations are shown for all continuous variables. Correlations greater than |.18| are significant at p � .01 and those greater than |.13| aresignificant at p � .05, two-tailed. Cronbach’s alpha coefficients are reported on the diagonal for measured variables.


the high aggression/low fear of retaliation condition was signifi-cantly different from all other categories (p � .001) and that noneof the other conditions were significantly different from one an-other, supporting Hypothesis 2b.

Hypothesis 3a and 3b reflect competing predictions for theeffects of fear of retaliation on the positive relationship betweenaggression and displaced aggression. The frustration–aggressionview (H3a) suggests fear of retaliation will strengthen the effects,and the learned inhibitions view (H3b) suggests fear of retaliationwill weaken the effects. The interaction was not significant ondisplaced aggression, and, thus, neither view was supported inStudy 1.

Study 1 Discussion

Our purpose in Study 1 was to examine, within an experimentalsetting, the moderating effects of LOC and fear of retaliation onretaliatory and displaced aggression reactions to perceived aggres-

sion. The results support our predictions for retaliation but not fordisplaced aggression. For retaliation, the effects of perceived ag-gression (i.e., aggressive instructor comments) were stronger whenLOC was lower than higher and when fear of retaliation was lowerthan higher. The findings support the notion that externals (lowLOC individuals), guided by futility beliefs, see less opportunity tocontrol the threat (i.e., instructor), which prompts destructive re-actions (i.e., retaliation). Also consistent with our theorizing, theeffects of perceived aggression were stronger when fear of retal-iation was lower than higher; participants retaliated more whenthey did not believe their instructor would know what they wrotein their evaluations. This suggests learned inhibitions affectedparticipants’ reactions (Bandura, 1973); individuals consider theconsequences of reacting aggressively against an aggressive au-thority figure when they fear retaliation from that person.

In contrast to the results for retaliation, no significant effectswere found for displaced aggression. We argued low LOC wouldstrengthen aggressive reactions to perceived aggression (Deci &Ryan, 1980), and we presented opposing predictions for the effectsof fear of retaliation. Based on the frustration–aggression view(Dollard et al., 1939), fear of retaliation would strengthen dis-placed aggression reactions to perceived aggression, and based onthe learned inhibitions view (Bandura, 1973; Tedeschi & Felson,1994), fear of retaliation would weaken those reactions. However,neither the LOC nor the fear of retaliation interaction was signif-icant. The lack of results may be a consequence of the target of

Table 3Study 1 Means and Standard Deviations of Aggression as aFunction of Fear of Retaliation

Variable

Low fear of retaliation High fear of retaliation

M SD n M SD n

Low aggression 2.52a 1.22 64 2.32a 1.45 53High aggression 3.19b 0.74 64 2.45a 1.26 61

Note. Within each row and column, means sharing a subscript do notdiffer from one another; those with different subscripts are different fromone another.

Table 2Study 1 Moderated Multiple Regressions ofHypothesized Relationships

Variable Retaliation Displaced aggression

ControlsAssignment grade �.03 �.01Instructor 1 .55�� .78��

Instructor 2 �.30�� .34��

Instructor 3 �.05 �.53��

Instructor 4 .05 �.28��

Instructor 5 �.16� �.17Predictors

Aggression .24�� .03Fear of retaliation .06 .00Locus of control .07 .19�

ModeratorsAggression � Fear of Retaliation �.20�� .01Aggression � Locus of Control �.15� �.14

R2 .57 .26Adjusted R2 .55 .22F 27.33�� 5.87��

Note. Standardized beta coefficients are reported.� p � .05, two-tailed. �� p � .01, two-tailed. �� p � .001, two-tailed.

2

2.2

2.4

2.6

2.8

3

3.2

3.4

3.6

Low High

Ret

alia

tion

Aggression

High Locus ofControl

Low Locus ofControl

Figure 1. Study 1 Aggression � Locus of Control interaction on retali-ation.

11.21.41.61.8

22.22.42.62.8

33.23.4

Low High

Ret

alia

tion

Aggression

High Fear of Retaliation Low Fear of Retaliation

Figure 2. Study 1 Aggression � Fear of Retaliation interaction onretaliation.


displaced aggression (team members). Team members are individ-uals with whom students had ongoing and interdependent relationsand on whom they had to rely on for future class activities. Indeed,the learned inhibitions view would suggest students should discernto whom and when to displace aggression (Bandura, 1973; Tede-schi & Felson, 1994). Thus, it is possible students understood thatdisplacing aggression against their peers might tarnish the rela-tionships they held with their team members. Students may alsohave believed their displaced aggression might further provoke theaggressor (i.e., the instructor). Alternatively, the lack of resultsmay be a consequence of our displaced aggression measure, as weused a team evaluation and did not ask about aggression directedagainst specific targets.

Study 1’s experimental design offers a number of strengths,such as the allowance of greater control, enhanced internal valid-ity, and the ability to draw causal conclusions. The design alsoprovides behavioral, rather than self-report, indicators of retalia-tion and displaced aggression. Still, this approach has limitations.First, not all predictions could be tested. Second, although we triedto maximize the realism of the experiment, the examination of thehypothesized relationships in an experimental setting with studentsraises concerns about the generalizability of the results. There aredifferences in the nature of the relationship between an instructorand student, compared to a supervisor and subordinate. For in-stance, students are typically in a relationship with their instructorfor only one class and one semester. In contrast, subordinates haveongoing relationships with their supervisors and typically haveonly one supervisor. Further, the extent to which an instructor canexert power over a student may be different than the extent towhich a supervisor can exert power over a subordinate. Thus, twocomplementary field studies were designed to examine our pre-dictions and test the generalizability of Study 1’s findings.

Study 2 Method

Sample and Procedure

Surveys were distributed to individuals randomly selected andcalled to jury duty by a southeastern U.S. county circuit court.Jurors were addressed before they were called to serve. Theresearchers explained the purpose of the study and emphasized thatthe survey had nothing to do with jury duty or the court system.Interested participants picked up surveys from and returned sur-veys to the researcher. Over four weeks, 321 individuals partici-pated in the study (29% response rate). The final sample of 278observations included participants who worked with coworkersand who had complete data for our study’s variables. On average,participants were 42.8 years old (SD � 12.2) and their companytenure was 8.4 years (SD � 7.9); 49.4% were managers, 47.2%were female, and 70.6% were White (13.2% were Hispanic, 8.6%were Black).

Measures

Supervisor aggression. Supervisor aggression was assessedwith Mitchell and Ambrose’s (2007) five-item version of Tepper’s(2000) measure, which assesses active (versus passive) supervisoraggression (e.g., “My boss ridiculed me,” “My boss put me down

in front of others”). Participants rated the frequency they experi-enced the behavior on a 7-point scale (1 � never, 7 � daily).

Moderator variables. The measure for LOC described inStudy 1 was also used in Study 2. Fear of retaliation was assessedwith Fox and Spector’s (1999) three-item fear of future punish-ments measure. Instructions were adapted to asked participants torate perceptions about their immediate supervisor (vs. their orga-nization) on a 7-point Likert-like scale (1 � strongly disagree, 7 �strongly agree; “I am afraid of reacting against my supervisor forfear of future punishments,” “I would not act out against mysupervisor because s/he would retaliate against me,” and “I wouldnot do something against my supervisor because s/he would getback at me in some way”). Consistent with the literature (e.g.,Aquino & Douglas, 2003), coworker aggressive modeling wasassessed with Robinson and O’Leary-Kelly’s (1998) nine-itemantisocial behavior scale. Respondents rated the frequency ofobserved coworkers being aggressive on a 5-point scale (1 �never, 5 � 10 or more times; e.g., “Damage property belonging tothe organization,” “Started an argument with someone at work”).

Dependent variables. Retaliation was assessed with Mitchelland Ambrose’s (2007) 10-item supervisor-directed deviance mea-sure (e.g., “Swore at my supervisor,” “Refused to talk to mysupervisor”). Coworker displaced aggression was assessed byadapting Bennett and Robinson’s (2000) seven-item interpersonaldeviance measure to focus on behaviors intentionally aimed atcoworkers (e.g., “Made an obscene comment or gesture toward acoworker,” “Publicly embarrassed a coworker”). Problem solvingwas assessed with eight items adapted from Rusbult, Farrell,Rogers, and Mainous’ (1988) Voice and Tepper et al.’s (2001)Constructive Resistance measures (“Asked coworkers for adviceabout what to do,” “Tried to reconcile with the person I was havingtrouble with,” “Discussed the problem with your immediate su-pervisor,” “Tried to change the situation to benefit all partiesinvolved,” “Asked the person to clarify the problem,” “Discussedwith the office manager the situation,” Asked the person for moreexplanation about the problem,” and “Tried to convince the personto reassess the problem”).7 For all dependent variables, respon-dents rated the behaviors on a 7-point scale (1 � never, 7 � daily).

Controls. Research shows aggression promotes aggression(see Anderson & Bushman, 2002, for a review); because weexamine coworker displaced aggression, we controlled for per-ceived coworker aggression in the analysis, with adapted itemsfrom Mitchell and Ambrose’s (2007) five-item abusive supervi-sion measure and three items from the Workplace AggressionQuestionnaire (WAQ; Schat et al., 2005).8 Further, consistent withaggression research (e.g., Mitchell & Ambrose, 2007), we con-trolled for trait anger (with Buss & Perry’s 1992 seven-item anger

7 The adapted problem solving measure was pretested on a separatesample of participants from the same population as in Study 2 (N � 62).Exploratory factor analysis, using principal-axis factoring, shows the itemsloaded on one factor (eigenvalue � 6.03). The measure demonstratedacceptable psychometric standards (� � .95; Nunnally & Bernstein, 1994).

8 Measures of coworker aggression were pretested on a separate samplefrom the same population as in Study 2 (N � 85). Exploratory factoranalysis, using principal-axis factoring, shows the items loaded on onefactor (eigenvalue � 5.60) and the measure demonstrated acceptablepsychometric standards (coworker aggression � � .94; Nunnally & Bern-stein, 1994).


subscale of the Aggression Questionnaire), social desirability (withPaulhus’ 1991 18-item measure), and participants’ tenure withtheir supervisor (in years). Last, Aquino and colleagues’ (e.g.,Aquino & Douglas, 2003; Aquino, Galperin, & Bennett, 2004)demonstrated that social status variables (e.g., gender, race, posi-tion in the company) influence aggressive behavior, and so wecontrolled for respondents’ gender (0 � female, 1 � male), race(0 � White, 1 � non-White), and position (0 � non-management,1 � management).

Study 2 Results

Table 4 provides the descriptive statistics, correlations, andreliability coefficients for Study 2’s variables. The measures dis-played acceptable reliabilities. Table 5 displays the moderatedmultiple regression results. As with Study 1, predictor variableswere mean centered and significant interactions were plotted andtested at one standard deviation above and below the mean (J.Cohen et al., 2003).

Hypothesis 1 predicts higher LOC will strengthen the positiveeffects of supervisor aggression on (H1a) problem solving, andlower LOC will strengthen the effects of supervisor aggression on(H1b) retaliation and (H1c) displaced aggression. The interactionwas significant on retaliation but not on problem solving or co-worker displaced aggression. The plot shows the effects on retal-iation (see Figure 3) were stronger when LOC was lower (b � .30,p � .001) than higher (b � .14, p � .01). The results support H1bbut not H1a or H1c.

Hypothesis 2 predicts higher fear of retaliation will strengthenthe positive effects of supervisor aggression on (H2a) problemsolving and lower fear of retaliation will strengthen the positiveeffects of supervisor aggression on (H2b) retaliation. The interac-tion was significant on problem solving and retaliation. SupportingH2a and H2b, the effects were stronger on problem solving (seeFigure 4) when fear of retaliation was higher (b � .23, p � .05)than lower (b � �.03, ns) and on retaliation (see Figure 5) whenfear of retaliation was lower (b � .33, p � .001) than higher (b �.11, p � .01).

Hypothesis 3 presents competing predictions about how fear ofretaliation will impact coworker displaced aggression reactions tosupervisor aggression. The frustration–aggression view (H3a) pre-dicts higher fear of retaliation will strengthen the positive relation-ship; the learned inhibitions view (H3b) predicts higher fear ofretaliation will weaken the positive relationship. The interactionwas significant, and the plot (see Figure 6) supports the learnedinhibitions view (H3b). The positive relationship between super-visor aggression and displaced aggression was weaker when fearof retaliation was higher (b � �.13, p � .05) than lower (b � .13,p � .05). In fact, the relationship was negative when fear ofretaliation was higher than lower.

Hypothesis 4 predicts behavioral modeling will impact the ef-fects of supervisor aggression on our dependent variables. Study 2tests only the aggressive modeling predictions, which suggestaggressive modeling will strengthen the positive effects of super-visor aggression on (H4a) retaliation and (H4b) coworker dis-placed aggression. The interaction was significant on both depen-dent variables, and the plots support our predictions. Bothretaliation (see Figure 7) and coworker displaced aggression (seeFigure 8) were stronger when aggressive modeling was higher(b � .32, p � .001 for retaliation; b � .13, p � .05 for coworkerdisplaced aggression) than lower (b � .12, p � .01 for retaliation;b � �.13, p � .05 for coworker displaced aggression).

Study 2 Discussion

Study 2 complements Study 1 by investigating the predictionswith working adults. We tested the moderating effects of LOC,fear of retaliation, and aggressive modeling on the relationshipsamong supervisor aggression and retaliation, coworker displacedaggression, and problem solving. Study 2’s findings are consistentwith Study 1 with regard to the effects on retaliation. That is, LOCand fear of retaliation moderated the effects, such that lower LOCand lower fear of retaliation strengthened retaliatory reactions tosupervisor aggression.

The other findings from Study 2 were also revealing. For ex-ample, although LOC significantly moderated the effects of super-

Table 4Study 2 Summary of Descriptive Statistics and Correlations

Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. Supervisor aggression 1.47 0.91 (.90)2. Locus of control 5.62 0.76 �.06 (.73)3. Fear of retaliation 2.95 1.99 .17 �.22 (.90)4. Coworker aggressive modeling 2.33 1.03 .40 �.05 .18 (.92)5. Retaliation 1.43 0.66 .61 �.10 .45 .10 (.84)6. Coworker displaced aggression 1.79 0.81 .40 �.06 .44 .08 .73 (.85)7. Problem solving 2.87 1.33 .24 .06 .27 .07 .16 .08 (.93)8. Gender (0 � female) 0.53 0.50 .11 .02 .03 �.13 .14 .21 �.00 —9. Race (0 � White) 0.71 0.46 .00 .03 .11 �.05 .00 .08 �.05 .07 —

10. Position (0 � non-management) 0.50 0.50 �.01 .18 .05 �.11 �.06 �.05 .20 .17 .15 —11. Tenure with the supervisor 3.63 4.44 �.06 .05 .01 �.02 �.07 �.12 .03 �.07 .13 .19 —12. Social desirability 2.63 0.57 .15 �.07 .25 .05 .45 .51 .06 .19 �.07 �.10 �.16 (.73)13. Trait anger 1.99 0.70 .14 �.05 .22 .17 .31 .28 .05 .08 .00 .01 �.04 .30 (.79)14. Coworker aggression 1.74 0.90 .59 �.07 .50 .18 .56 .51 .27 .03 .02 �.00 �.07 .22 .30 (.88)

Note. Correlations greater than |.14| are significant at p � .01 and those greater than |.11| are significant at p � .05, two-tailed. Cronbach’s alphacoefficients are reported on the diagonal for measured variables.


visor aggression on retaliation, it did not influence the relationshipbetween supervisor aggression and problem solving or coworkerdisplaced aggression. Theory about LOC suggests externals do notbelieve they can exact change when they face threats (i.e., super-visor aggression), which motivates them to react destructively.Yet, the results of Study 2 (and Study 1) for LOC suggest indi-viduals may gauge the usefulness of engaging in problem solvingand certain destructive reactions. It is possible that internals maynot engage in problem solving if they do not believe the behaviorsare possible or available to them within their work setting. Further,victims may also believe displacing aggression might ruin rela-tionships with peers, making the situation even more taxing. Thisline of reasoning suggests victims’ behavioral tendencies withregard to LOC may be more nuanced than previous researchsuggests.

The results for the moderating effects of fear of retaliationsupported the learned inhibitions view (Bandura, 1973; Tedeschi& Felson, 1994). A higher fear of retaliation inhibited destructivereactions and enhanced constructive reactions to supervisor ag-gression. The relationships between supervisor aggression andboth forms of aggressive reactions (retaliation and coworker dis-placed aggression) were stronger when fear of retaliation waslower than higher, and the relationship between supervisor aggres-sion and problem solving was stronger when fear of retaliation washigher than lower. The results suggest fearing retaliation from anaggressive supervisor creates a situation that motivates individualsto not aggress outwardly at work and, instead, to try to find aresolution to the problem. Notably, the results did not support thefrustration–aggression view of displaced aggression; the high fearof retaliation slope was significant and negative, meaning victimswere less likely to vent their hostilities on coworkers and were notovercome by catharsis.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion

Supervisor Aggression


Low Locus ofControl

Figure 3. Study 2 Supervisor Aggression � Locus of Control interactionon retaliation.

2

2.2

2.4

2.6

2.8

3

3.2

3.4

Low High

Pro

ble

m S

olvi

ng


High Fear ofRetaliation

Low Fear ofRetaliation

Figure 4. Study 2 Supervisor Aggression � Fear of Retaliation interac-tion on problem solving.

Table 5Study 2 Multiple Regressions for Hypothesized Relationships

Variable Retaliation Coworker displaced aggression Problem solving

ControlsGender (0 � female) .01 .12�� .02Race (0 � White) .01 .05 �.10Position (0 � non-management) �.04 �.06 .21��

Tenure with the supervisor .02 �.02 .03Social desirability .27�� .33�� .01Trait anger .10�� .05 �.05Coworker aggression .13�� .27�� .15��

PredictorsSupervisor aggression .31�� .00 .07Locus of control �.01 .01 .02Fear of retaliation �.03 .00 .02Coworker aggressive modeling .15�� .17�� .18��

ModeratorsSupervisor Aggression � Locus of Control �.11�� .07 .05Supervisor Aggression � Fear of Retaliation �.18�� .17�� .11�

Supervisor Aggression � Coworker Aggressive Modeling .21�� .21��

R2 .63 .54 .63Adjusted R2 .61 .51 .61F 31.48�� 21.88�� 31.48��

Note. For all dependent variables, N � 278. Standardized beta coefficients are reported.� p � .05, one-tailed. �� p � .01, one-tailed. �� p � .001, one-tailed.


Last, Study 2 provides a test of the aggressive modeling pre-dictions and shows that when victims perceived more (rather thanless) aggressive modeling from coworkers, they were more in-clined to respond to supervisor aggression with aggression. Thepositive relationship between supervisor aggression and both re-taliation and coworker displaced aggression was stronger whencoworker aggressive modeling was higher than lower. When pro-voked by an aggressive supervisor, victims imitate the behaviors oftheir coworkers; they believe such aggressive behaviors are ac-ceptable within the work environment.

Study 2 has notable strengthens. It examines the hypotheses ina diverse, randomly selected sample. Additionally, it replicates themoderating effects of LOC and fear of retaliation on the relation-ship between supervisor aggression and retaliation, which werefound in the experiment, addressing issues of generalizability.However, it too has limitations. First, Study 2’s data were collectedfrom a single source at a single point in time, increasing thepotential for common method bias (Podsakoff, MacKenzie, Lee, &Podsakoff, 2003). Second, Study 2 did not allow for a test of all ofour predictions; the effects of problem solving modeling on prob-lem solving reactions to supervisor aggression were not tested.Last, it is possible that not all of our respondents were able to

engage in the problem solving behaviors assessed (e.g., report thesupervisor to a higher authority), as we did not ask participants ifthese paths of recourse were available to them. Study 3 wasdesigned to address these issues. We used a time-separated datacollection design for Study 3, where the predictor and moderatorvariables were collected at Time 1 and dependent variables werecollected at Time 2. Second, we assessed problem solving mod-eling, allowing for an examination of all of our hypotheses. Fi-nally, in Study 3, we examined a range of problem solving behav-iors (seeking social support from coworkers, trying to reconcilewith the supervisor, and reporting the aggressive supervisor toauthorities), and we asked respondents to indicate whether theyhad the opportunity to engage in the listed problem solving be-haviors.

Study 3 Method

Sample and Procedure

Data were collected from individuals working in full-time po-sitions from various organizations located in the southeastern U.S.in industries including finance, insurance, education, health care,information technology, and retail. Surveys were administered via

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion


High Fear of Retaliation

Low Fear of Retaliation

Figure 5. Study 2 Supervisor Aggression � Fear of Retaliation interac-tion on retaliation.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Cow

orke

r D

isp

lace

d A

ggre

ssio

n




Figure 6. Study 2 Supervisor Aggression � Fear of Retaliation interac-tion on coworker displaced aggression.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion


High CoworkerAggressive Modeling

Low CoworkerAggressive Modeling

Figure 7. Study 2 Supervisor Aggression � Coworker Aggressive Mod-eling interaction on retaliation.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Cow

orke

r D

isp

lace

d A

ggre

ssio

n


High Coworker Aggressive Modeling

Low Coworker Aggressive Modeling

Figure 8. Study 2 Supervisor Aggression � Coworker Aggressive Mod-eling interaction on coworker displaced aggression.


the Internet. Business administration students were asked to serveas organizational contacts for course credit. The researchers askedstudents to recruit up to two full-time working adults who workedwith coworkers to complete two surveys, which would be sepa-rated by approximately three weeks. This data collection techniquehas been used successfully by a variety of researchers (e.g., DeCremer et al., 2010; Grant & Mayer, 2009; Greenbaum, Mawritz,& Eissa, 2012).

We received contact information for 674 potential study partic-ipants. From these, 371 individuals participated in the Time 1survey (55% response rate). At Time 1, participants were askedabout supervisor aggression, locus of control, fear of retaliation,aggressive modeling, problem solving modeling, and demograph-ics. At Time 2, participants were asked about their behaviors(retaliation, coworker displaced aggression, seeking social support,and reconciling with and reporting the supervisor). Of the 298individuals who participated in the Time 2 survey, the final samplewas limited to 243 individuals, who had complete data for ourstudy’s variables and were verified to work with coworkers. Onaverage, participants were 41.1 years old (SD � 13.7) and theircompany tenure was 8.7 years (SD � 8.9); 43.1% were managers,57.8% were female, and 86% were White (7.1% were Hispanic,1.6% were Black).

Measures

Supervisor aggression. The measure described in Study 2was also used in Study 3.

Moderator variables. The measures of LOC, fear of retali-ation, and coworker aggressive modeling described in Study 2were used in Study 3. For problem solving modeling, we adaptedthe problem solving behavior measure from Study 2 and askedrespondents to rate how often they observed coworkers engagingin the listed behaviors on a 7-point scale (1 � never, 7 � always;e.g., “Ask for advice about the problem,” “Tried to change thesituation to benefit all parties involved”).

Dependent variables. The measures of retaliation and co-worker displaced aggression described in Study 2 were also usedin Study 3. We assessed three forms of problem solving: seekingcoworker social support, reconciling with the supervisor, andreporting the supervisor to authorities. Seeking coworker socialsupport was measured with three items from Westring and Ryan’s(2010) measure (“Asked for emotional support,” “Sought help onhow to handle the conflict and problem,” and “Got advice on howto sort out the issue”). Reconciling with the supervisor was mea-sured with two items from Wade’s (1989) reconciliation subscale(“I made an effort to be more concerned” and “I tried to reconcilethe situation with the supervisor”). Reporting the supervisor toauthorities was measured with five items (“Reported the issue tothe office manager or Human Resource representative,” “Dis-cussed the situation with someone in a managerial position,”“Reported the issue to someone in authority to resolve the issue,”“Discussed with the office manager how you felt about the situa-tion,” and “Asked someone in authority to directly help youresolve the situation”).9 For all problem solving measures, partic-ipants were asked whether the problem solving behaviors wereavailable to them in their work environment; the analyses includeddata only for those who indicated the problem solving behaviorwas available. Respondents rated the frequency they engaged in

the stated behaviors on a 7-point scale (1 � never, 7 � always) forall dependent variables.

Controls. The control variables used in Study 2 were in-cluded in our Study 3 analyses. We also controlled for the traitemotional stability, using Saucier’s (1997) 10-item measure, asthis is a trait that impacts both constructive and destructive behav-ior (see Perrewe & Spector, 2002).

Study 3 Results

Table 6 provides the descriptive statistics, correlations, andreliability coefficients for the Study 3 variables. The measuresdisplayed acceptable reliabilities. We followed the analytic proce-dures described in Study 2. Table 7 displays the multiple moder-ated regression results.

Hypothesis 1 predicts higher LOC will strengthen the positiveeffects of supervisor aggression on (H1a) problem solving andlower LOC will strengthen the positive effects of supervisor ag-gression on (H1b) retaliation and (H1c) displaced aggression. Theinteraction was significant on two problem solving behaviors(reconciling and reporting to authorities), retaliation, and coworkerdisplaced aggression but not on seeking coworker social support.Thus, H1a was primarily supported. The plots support H1a forreconciling (see Figure 9) and reporting to authorities (see Figure10), H1b for retaliation (see Figure 11), and H1c for coworkerdisplaced aggression (see Figure 12). The effects of supervisoraggression were stronger on both reconciling and reporting toauthorities when LOC was higher (b � .22, p � .05 for reconcil-ing; b � .23, p � .05 for reporting to authorities) than lower (b ��.16, ns for reconciling; b � �.05, ns for reporting to authorities).Further, the effects of supervisor aggression were stronger on bothretaliation and coworker displaced aggression when LOC waslower (b � .23, p � .001 for retaliation; b � .20, p � .01 forcoworker displaced aggression) than higher (b � .03, ns forretaliation; b � .01, ns for coworker displaced aggression).

Hypothesis 2 predicts higher fear of retaliation will strengthenthe positive effects of supervisor aggression on (H2a) problemsolving and lower fear of retaliation will strengthen the positiveeffects of supervisor aggression on (H2b) retaliation. The interac-tion was significant on retaliation but not on the forms of problemsolving. The plot for retaliation (see Figure 13) shows the rela-tionship between supervisor aggression and retaliation was stron-ger when fear of retaliation was lower (b � .21, p � .05) thanhigher (b � .05, ns). The results support H2b but not H2a.

Hypothesis 3 presents competing predictions about how fear ofretaliation will impact coworker displaced aggression reactions tosupervisor aggression. The frustration–aggression view (H3a) pre-dicts higher fear of retaliation will strengthen the positive effects;the learned inhibitions view (H3b) predicts higher fear of retalia-tion will weaken the effects. The interaction was significant, andthe plot (see Figure 14) supports the learned inhibitions view(H3b); the effects were weaker when fear of retaliation was higher(b � �.02, ns) than lower (b � .24, p � .001).

Hypothesis 4 predicts aggressive modeling will strengthen thepositive effects of supervisor aggression on (H4a) retaliation and

9 Items were constructed based on qualitative evidence presented in priorresearch (cf. Bies & Tripp, 1996).


(H4b) coworker displaced aggression and problem solving mod-eling will strengthen the effects of supervisor aggression on (H4c)problem solving. The interactions were significant for retaliation,coworker displaced aggression, and seeking coworker social sup-port but not for reconciling or reporting to authorities. As pre-dicted, the plots show the effects for retaliation (see Figure 15) andcoworker displaced aggression (see Figure 16) were stronger whenaggressive modeling was higher (b � .22, p � .001 for retaliation;b � .17, p � .01 for coworker displaced aggression) than lower(b � .04, ns for retaliation; b � .04, ns for coworker displacedaggression), and the effects on seeking social support (see Figure17) were stronger when problem solving modeling was higher(b � .22, p � .05) than lower (b � �.05, ns). The results supportH4a and H4b and provide partial support for H4c.

Study 3 Discussion

Our purpose in Study 3 was to provide a test of all predictions,examine the generalizability of Study 1, and complement Study 2’sdesign. All of the predicted effects for retaliation and coworkerdisplaced aggression were supported in Study 3, but only some ofthe predictions for problem solving were supported. The differen-tial effects for our moderators on problem solving may be a factorof the distinct forms of problem solving we assessed. We expandon these ideas below and discuss each of our predictions in turn.

The results support our LOC predictions for destructive reac-tions and for two forms of problem solving behavior. A lower LOCstrengthened the positive effects of supervisor aggression on re-taliation and displaced aggression, and a higher LOC strengthenedthe effects on reconciling with and reporting the supervisor toauthorities. Interestingly, LOC did not moderate the effects onseeking social support problem solving. This result may indicateinternals are more discriminating in how they problem solve. Bothreconciling with and reporting the supervisor allow internals to tryto directly resolve the problem. In contrast, seeking coworkers’social support involves victims leaning on others for consolation.This behavior is a less direct approach to addressing the problem,relies on others rather than oneself, and may not result in actiondirected at the problem. The correlations between LOC and theproblem solving behaviors support these arguments. LOC waspositively related to reconciling with and reporting the supervisor,but it was negatively correlated with seeking coworker socialsupport. These correlations suggest internals are more likely toengage in more proactive, independent problem solving behaviors(i.e., reconciling, reporting the supervisor) and less likely to en-gage in dependent problem solving behavior (i.e., seeking co-worker social support).

The results also support the learned inhibition view of fear ofretaliation (Bandura, 1973; Tedeschi & Felson, 1994). For thedestructive reactions, higher fear of retaliation weakened the rela-tionship between supervisor aggression and both retaliation andcoworker displaced aggression. Hence, victims’ learned inhibi-tions allowed them to understand that acting aggressively—whether against the supervisor or coworkers—would likely maketheir situation worse.

Last, the results for the behavioral modeling predictions aresupported for aggressive modeling but are only partially supportedfor problem solving modeling. The effects of supervisor aggres-sion on retaliation and coworker displaced aggression were stron-T

able

6St

udy

3Su

mm

ary

ofD

escr

ipti

veSt

atis

tics

and

Cor

rela

tion

s

Var

iabl

eM

SD1

23

45

67

89

1011

1213

1415

1617

18

1.Su

perv

isor

aggr

essi

on1.

650.

89(.

89)

2.L

ocus

ofco

ntro

l4.

000.

40�

.15

(.70

)3.

Fear

ofre

talia

tion

2.33

1.06

.39

�.2

1(.

93)

4.C

owor

ker

aggr

essi

vem

odel

ing

2.68

1.06

.34

�.0

8.2

3(.

91)

5.Pr

oble

mso

lvin

gm

odel

ing

4.44

0.99

�.1

3.1

1�

.13

�.2

0(.

89)

6.R

etal

iatio

n1.

460.

58.3

2�

.14

.22

.46

�.0

5(.

86)

7.C

owor

ker

disp

lace

dag

gres

sion

1.77

0.67

.24

�.1

5.1

0.4

8�

.04

.61

(.81

)8.

Seek

ing

soci

alsu

ppor

t3.

641.

11.0

3�

.16

.01

.06

.28

.17

.18

(.80

)9.

Rec

onci

ling

4.42

1.44

�.0

5.2

2�

.14

�.0

4.3

1�

.03

�.0

3.3

3(.

82)

10.

Rep

ortin

gto

anau

thor

ity3.

010.

95�

.04

.20

�.1

7�

.06

.30

�.0

5�

.07

.34

.46

(.74

)11

.G

ende

r(0

�fe

mal

e)1.

580.

50.0

1�

.10

.06

�.0

5.0

8�

.10

�.1

2.3

0.1

2.1

9—

12.

Rac

e(0

�W

hite

)1.

400.

49�

.08

.02

.05

�.0

6�

.03

�.0

2�

.06

.03

�.0

2.0

7.0

8—

13.

Posi

tion

(0�

non-

man

agem

ent)

1.43

0.50

�.0

5.0

9�

.18

�.1

0.1

2�

.11

�.0

7�

.03

.15

.08

�.2

3�

.12

—14

.T

enur

ew

ithth

esu

perv

isor

3.69

4.38

�.0

8.0

0�

.04

�.0

7.0

5�

.03

�.0

9.0

1.1

0.1

5.0

8�

.07

.22

—15

.So

cial

desi

rabi

lity

2.67

0.55

.11

�.1

9.1

4.3

5.0

0.4

2.5

2.1

0�

.05

�.0

7�

.07

�.0

6�

.17

�.1

7(.

80)

16.

Tra

itan

ger

2.04

0.71

.16

�.1

4.1

4.1

3�

.07

.22

.21

.10

.00

.10

.06

.06

�.1

3�

.02

.21

(.80

)17

.E

mot

iona

lst

abili

ty3.

790.

56�

.22

.44

�.2

2�

.23

.02

�.2

2�

.17

�.2

6.0

4.0

5�

.09

�.0

0.1

5�

.03

�.2

9�

.37

(.83

)18

.C

owor

ker

aggr

essi

on1.

770.

90.5

6�

.15

.32

.61

�.2

0.4

0.4

4.0

6�

.03

�.0

7�

.01

�.0

3�

.07

�.0

8.2

7.2

9�

.28

(.93

)

Not

e.C

orre

latio

nsgr

eate

rth

an|.1

4|ar

esi

gnif

ican

tatp

�.0

1an

dth

ose

grea

ter

than

|.12|

are

sign

ific

anta

tp�

.05,

two-

taile

d.C

ronb

ach’

sal

pha

coef

fici

ents

are

repo

rted

onth

edi

agon

alfo

rm

easu

red

vari

able

s.


ger when coworker aggressive modeling was higher than lower,and the effects of supervisor aggression on seeking coworkersocial support were stronger when problem solving modeling washigher than lower. The pattern of these interactions shows co-worker modeling impacts employees’ reactions to supervisor ag-gression in a “monkey-see, monkey-do” pattern (cf. Robinson &O’Leary-Kelly, 1998). Yet, the influence of coworker modeling isnot found for reconciling with and reporting the supervisor prob-lem solving. Thus, it appears aggressive modeling led to moregeneralized behavior (aggression targeted at supervisors and ag-gression targeted at coworkers) than did problem solving. For

problem solving modeling, victims target problem solving reac-tions toward the source of the behavioral modeling (i.e., cowork-ers). It appears employees engage in problem solving (i.e. supportseeking) that is likely to be favorably received by the model.

General Discussion

Together, the results of our three studies provide useful insightsabout employees’ reactions to supervisor aggression (see the Ap-pendix for a summary of the findings across the three studies).Consistent patterns emerged for the moderating effects of LOC,fear of retaliation, and aggressive modeling on retaliatory reactions

3.8

4

4.2

4.4

4.6

4.8

5

5.2

Low High

Rec

onci

le



Low Locus ofControl

Figure 9. Study 3 Supervisor Aggression � Locus of Control interactionon reconciling.

2.4

2.6

2.8

3

3.2

3.4

3.6

Low High

Rep

orti

ng

to A

uth

orit

ies



Low Locus ofControl

Figure 10. Study 3 Supervisor Aggression � Locus of Control interac-tion on reporting to authorities.

Table 7Study 3 Multiple Regressions for Hypothesized Relationships

Variable RetaliationCoworker displaced

aggressionSeeking coworker

social support ReconcilingReporting to an

authority

ControlsGender (0 � female) �.06 �.10�� .28�� .18�� .14��

Race (0 � White) .05 .02 .04 �.04 .09Position (0 � non-management) �.02 .02 .06 .11 .02Tenure with the supervisor �.00 �.03 �.07 .03 .18��

Social desirability .27�� .39�� .01 �.06 �.02Trait anger .14�� .10� �.03 �.04 .14��

Emotional stability .04 .08 �.22�� .11 .05Coworker aggression .02 .17� .00 .05 �.05

PredictorsSupervisor aggression .20�� .14�� .07 .02 .08Locus of control �.07 �.11�� .03 .25�� .21��

Fear of retaliation .05 �.03 .00 �.07 �.09Coworker aggressive modeling .26�� .20�� .07 �.04 .07Problem solving modeling .08 .08 .30�� .22�� .27��

ModeratorsSupervisor Aggression � Locus of Control �.14�� .11�� .04 .11� .12�

Supervisor Aggression � Fear of Retaliation �.15�� .22�� .11 �.04 .01Supervisor Aggression � Coworker Aggressive Modeling .16�� .10�

Supervisor Aggression � Problem Solving Modeling .13�� .07 �.04R2 .41 .47 .26 .19 .22Adjusted R2 .37 .44 .21 .13 .16F 9.74�� 12.63�� 5.00�� 3.31�� 3.57��

Note. For retaliation, N � 238; for coworker displaced aggression, N � 240; for support seeking, N � 243; for reconciling, N � 239; for reporting toan authority, N � 219. Standardized beta coefficients are reported.� p � .05, one-tailed. �� p � .01, one-tailed. �� p � .001, one-tailed.


to supervisor aggression; there were also consistent moderatingeffects for fear of retaliation and aggressive modeling on coworkerdisplaced aggression reactions. Inconsistencies occurred for theeffects of the moderators on problem solving reactions to super-visor aggression. We believe these inconsistencies can be ex-plained by considering the type of problem solving. Below, weprovide a more detailed discussion. For each prediction, we dis-cuss the similarities and differences across the studies.

Hypothesis 1 predicted LOC would moderate the effects ofsupervisor aggression on victims’ reactions such that lower LOC(external orientation) would influence retaliation and displacedaggression and higher LOC (internal orientation) would influenceproblem solving. We were able to test this prediction in all threestudies for retaliation and displaced aggression but only in Studies2 and 3 for problem solving. The findings were robust for theeffects on the destructive reactions. All three studies show lowLOC strengthened the relationship between supervisor aggressionand retaliation, and the same effect was shown for coworkerdisplaced aggression in Study 3. Consistent with our theorizing,this general pattern suggests externals respond to aggression de-structively, whereas internals were more self-controlled.

The results also suggest internals were more likely to take actionthat could change the situation. In Study 3, we limited our analyses

to victims who felt they had the opportunity to engage in problemsolving behaviors. Study 3’s results show internals were morelikely to engage in problem solving behavior but only when thebehavior reflected a high level of independence and control (i.e.,reconciling with and reporting the supervisor). LOC did not influ-ence individuals to seek coworker social support—a behavior thatinvolves seeking help and consolation from peers but does notfocus on directly changing the situation or resolving the problem.The results from Study 1 are consistent with these findings, eventhough we did not assess problem solving explicitly. The coverstory given to students was that the chair of the department wouldbe using the instructor evaluations as a basis of assessing thedepartment and the instructors’ effectiveness. Implied is that theevaluations would be used as a basis for change. Consistent withthe stronger effects of supervisor aggression on problem solvingreporting behavior found in Study 3, in Study 1 there was apositive relationship between instructor aggression and negativeinstructor evaluation. It appears students viewed the instructorevaluation as a mechanism for enacting change.10

In contrast to Study 1 and Study 3, there was not a significantinteraction for supervisor aggression and LOC for problem solvingin Study 2. For Study 2, we relied on a general measure of problemsolving, which assessed the three types of problem solving behav-ior captured in Study 3 in an overall measure. The results of Study3 show LOC differentially impacted the problem solving behaviorsand only for behaviors that emphasized internals’ control. Thus, itis possible that the lack of results for LOC in Study 2 highlights theneed for LOC researchers to consider the type of behavior thatallows victims to directly change the situation. Alternatively, thelack of effects of LOC on problem solving in Study 2 might stemfrom whether victims felt they were able to engage in the problemsolving activities in their workplace. Nevertheless, the resultssuggest, overall, that internals are more likely to engage in behav-ior they believe can resolve threats they face.

Hypotheses 2 and 3 predicted fear of retaliation would moderatethe relationships between supervisor aggression and victims’ re-actions. We were able to test the predictions in Study 1 for

10 We thank an anonymous reviewer for this idea.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion



Low Locus ofControl

Figure 11. Study 3 Supervisor Aggression � Locus of Control interac-tion on retaliation.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Cow

orke

r D

isp

lace

d

Agg

ress

ion



Low Locus ofControl

Figure 12. Study 3 Supervisor Aggression � Locus of Control interac-tion on coworker displaced aggression.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion




Figure 13. Study 3 Supervisor Aggression � Fear of Retaliation inter-action on retaliation.


retaliation and coworker displaced aggression and in Studies 2 and3 on all of the victims’ reactions. Altogether, the pattern of theresults support the learned inhibitions view (Bandura, 1973; Te-deschi & Felson, 1994), which suggests higher fear of retaliationstrengthens the effects of supervisor aggression on problem solv-ing and lower fear of retaliation strengthens the effects on retali-ation and displaced aggression. We found consistent effects forfear of retaliation on the destructive reactions and some inconsis-tencies with problem solving. Fear of retaliation weakened thepositive effects of supervisor aggression on retaliation in all threestudies and on coworker displaced aggression in Studies 2 and 3.Overall, this suggests employees consider their destructive behav-ior in context and discern when to (and not to) react destructively.In a situation where employees are fearful of retaliatory acts by anaggressive supervisor, they refrain from acting out aggressively.Though the effects for problem solving could only be tested inStudies 2 and 3, only in Study 2 (where problem solving wasmeasured as an overall construct) did significant results emerge.Consistent with our theorizing, fear of retaliation heightenedawareness about the threatening situation, making victims morelikely to problem solve. As a whole, the results across the threestudies suggest individuals consider the situational context andreact according to their learned inhibitions.

Last, Hypothesis 4 predicted behavioral modeling would mod-erate the effects of supervisor aggression on victims’ reactions;coworker aggressive modeling was predicted to strengthen retali-ation and displaced aggression reactions, and problem solvingmodeling was predicted to strengthen problem solving reactions.In Study 1, we were unable to test these predictions, and in Study2, we tested only the effects of aggressive modeling. However, inStudy 3, we examined both forms of behavioral modeling onvictims’ reactions. The findings support the predictions for aggres-sive modeling. In both studies, aggressive modeling strengthenedthe positive effects of supervisor aggression on the destructivereactions—employees reacted to supervisor aggression more de-structively (i.e., retaliating, displacing aggression) when cowork-ers modeled aggressive behavior. However, the effects of problemsolving modeling were more complex. Problem solving modelingmoderated and strengthened the effects of supervisor aggressiononly on problem solving behavior that was targeted at the sourceof modeling (i.e., coworkers). Problem solving modeling strength-ened the relationship between supervisor aggression and seekingcoworker support. In short, coworker aggressive modeling affectsaggression targeted at both supervisors and coworkers, whereascoworker problem solving modeling seems to affect onlycoworker-targeted problem solving behaviors.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Cow

orke

r D

isp

lace

d A

ggre

ssio

n


High Fear of Retaliation

Low Fear of Retaliation

Figure 14. Study 3 Supervisor Aggression � Fear of Retaliation inter-action on coworker displaced aggression.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Ret

alia

tion




Figure 15. Study 3 Supervisor Aggression � Coworker Aggressive Mod-eling interaction on retaliation.

1

1.2

1.4

1.6

1.8

2

2.2

Low High

Cow

orke

r D

isp

lace

d A

ggre

ssio

n




Figure 16. Study 3 Supervisor Aggression � Coworker Aggressive Mod-eling interaction on coworker displaced aggression.

2.6

2.8

3.0

3.2

3.4

3.6

3.8

4.0

4.2

4.4

Low High

Seek

ing

Cow

orke

r So

cial

Su

pp

ort


High Problem SolvingModeling

Low Problem SolvingModeling

Figure 17. Study 3 Supervisor Aggression � Problem Solving Modelinginteraction on seeking coworker social support.


Implications for Theory and Future Research

The results of our studies have implications for theory and futureresearch. First, a great deal of research has emerged to understanddestructive reactions to supervisor aggression (e.g., Lian, Ferris, &Brown, 2012; Mitchell & Ambrose, 2007; Tepper et al., 2008; Thau& Mitchell, 2010). Less empirical attention has been given to under-standing constructive reactions (e.g., Tepper et al., 2001) and evenless to why victims might react one way versus another. Our studyaddresses these issues. The findings across our three studies suggestLOC, fear of retaliation from the aggressive supervisor, and behav-ioral modeling of coworkers impact the degree to which victimsengage in destructive or constructive reactions. Consequently, ourresearch provides new insight for the workplace aggression and abu-sive supervision literatures and for theorizing about employees’ senseof control (i.e., LOC, fear of retaliation) and the impact of behaviorallearning (i.e., behavioral modeling, fear of retaliation). We elaborateon these ideas below.

Our primary goal in this research was to understand the factorsthat influence how employees respond to supervisor aggression.The findings suggest LOC, fear of retaliation, and behavioralmodeling are influential. Notably, the findings show supervisoraggression significantly and directly impacted retaliation, but onlyin Study 2 did it significantly relate to displaced aggression, andacross all studies it did not significantly impact problem solving.This pattern suggests retaliation may be a primary reaction tosupervisor aggression, and that, more or less, the other reactionsdepend upon individual and situational factors. That is, whethervictims of supervisor aggression react destructively versus con-structively depends on victims’ personality (e.g., LOC) and thesituational context under which such abuse occurs (e.g., fearingretaliation from the aggressive supervisor, perceiving coworkersbehave aggressively, perceiving coworkers problem solving).More research is needed to identify other factors that motivatevictims’ responses. Additionally, more research is needed to un-derstand the process by which employees choose to react one wayversus the other. For example, it is entirely possible for employeesto react destructively if they attribute the supervisor’s behavior tothe supervisor, but victims might act constructively if they attributethe supervisor’s behavior to something they did. Again, moretheorizing and research is needed to fully understand reactions tosupervisor aggression.

The findings also have implications for theory and research onemployees’ sense of control. LOC impacted destructive reactionsto supervisor aggression; an external orientation motivates victimsto react aggressively to supervisory threats. This finding is con-sistent with reactance theory (Wortman & Brehm, 1975), whichsuggests that threatening situations invoke destructive behaviorand such behavior allows victims a way to express their hostilitiesabout their helpless state. The results across the three studiessuggest an external orientation heightens reactance for victims ofsupervisor aggression. Consistent with expectations, in Studies 2and 3, internals demonstrated greater control and refrained fromacting destructively. In Study 1 (the experiment), the relationshipbetween supervisor aggression and retaliation was positive forinternals. We argued this is because internals might have felt theirreaction (evaluating the instructor poorly) would exact change inthe aggressor’s behavior. Hence, an external orientation may en-hance reactance behaviors that are destructive, but an internal

orientation may also influence what is seemingly destructive be-havior if victims believe doing so may reduce the aggression in thefuture. Research is needed to test these ideas.

The results show important differential effects of LOC on con-structive reactions to supervisor aggression as well. By and large,prior research suggests LOC has an enduring and positive impacton different forms of employees’ behaviors that reflect high LOCindividuals (internals) acting proactively (Ng et al., 2006, andSpector, 1982, for reviews). Our studies, however, show thatinternals are discerning in their behavioral choices, particularly intheir reactions to supervisor aggression. Their proclivity to actconstructively appears to be bounded. Internals seem to engage inbehaviors that allow them to exert control tendencies but onlywhen the behavior offers a direct way to resolve the problem(reconciling with or reporting the supervisor), as opposed to actsthat may not allow for resolution (seeking coworker social sup-port). Thus, some forms of problem solving more naturally ap-pease control tendencies of internals. Future research might extendthese ideas and examine the effects of LOC on behaviors that canmore or less affect change.

Our findings also extend the aggression literature, specificallywith regard to fear of retaliation and displaced aggression. Wetested two theoretical views about the impact of fear of retaliationon displaced aggression reactions to supervisor aggression. Con-sistent findings across our field studies showed fear of retaliationweakened the positive effects of supervisor aggression on co-worker displaced aggression. This pattern supports the learnedinhibition view (Bandura, 1973; Tedeschi & Felson, 1994) ratherthan the frustration–aggression view (Dollard et al., 1939). Al-though the frustration–aggression view has received empiricalsupport within the social psychology literature (Marcus-Newhall etal., 2000), the learned inhibitions view emerged because the ca-thartic effect of displaced aggression was not consistently found(Tedeschi & Felson, 1994). When aggressive behavior (e.g.,coworker-directed aggression) would likely prove disruptive, thelikelihood of catharsis decreases. Our results support this view; thecontext influenced individuals’ reactions to aggression. Displacingaggression on coworkers—individuals on whom victims have torely—would likely destroy relationships and may further provokethe aggressive supervisor.

Additionally, our findings suggest fear of retaliation from aharmdoer is a powerful deterrent to different forms of aggressivereactions (i.e., retaliation, displaced aggression). Again, this con-trasts with prior social psychology research on displaced aggres-sion, which has shown aggression provokes displaced aggressionwhen individuals fear retaliation from the source of aggression(Marcus-Newhall et al., 2000). Our work suggests researchersshould consider the social context of organizations and how thiscontext impacts behavior (Heath & Sitkin, 2001), specifically interms of the sources and targets of aggressive reactions. That is,fear of retaliation may act to heighten one’s awareness of the socialcontext. Acting out against one’s peers is not wise because theseindividuals may console the victim and/or may respond withhostilities as a consequence of being the target of displaced ag-gression. The social context of our studies (individuals interdepen-dent on others in school and at work) impacted the nature of theaggression � fear of retaliation reaction, wherein fear of retaliationacted as a form of behavioral control. Of course, we are notproposing that supervisors and organizational decision makers go


out and spark fear in their employees. Indeed, research in thebehavioral ethics literature suggests these types of behavioralcontrol tactics, while inhibiting aggressive reactions, may promotemore clandestine deviant behaviors—acts conducted intentionallyand surreptitiously (Trevino, Weaver, Gibson, & Toffler, 1999;Weaver, Trevino, & Cochran, 1999a, 1999b). Instead, we suggestthat our tests differ from experimental research in social psychol-ogy (where individuals displace aggression on strangers) and high-light the importance of considering the social context of aggres-sion. More research is needed to investigate more effectivebehavioral controls that deter supervisor aggression and victims’destructive reactions and those that foster problem solving.

We expected that behavioral modeling would be one factor thatfacilitated destructive and constructive reactions to supervisor ag-gression. We argued that employees learn what is socially accept-able by watching what their peers do and then engage in similarbehavior. Although the results support these arguments for aggres-sive reactions, the influence of problem solving modeling wasmore limited. Problem solving modeling only influenced seekingsupport from coworkers. It may be a consequence of measurement(i.e., more items about seeking support), but it might also bedecision biases at play. Decision research has shown individualshave a tendency to focus on negative rather than positive infor-mation (see Rozin & Royzman, 2001, for a review). Research onleader behavior provides some suggestive evidence to support thisidea. For instance, Detert et al. (2007) found that aggressive leaderbehavior (via perceptions of abusive supervision) provided moreof a managerial mode of influence on employees’ behavior thandid ethical leader behavior (via perceptions of ethical leadership).Thus, it may be that aggressive modeling holds more influence toaffect behavioral learning than does problem solving modelingbecause aggressive acts (being harmful and negative) are moresalient to employees. More research is needed to fully understandthe impact of modeling on constructive work behavior.

Implications for Practice

The findings suggest supervisor aggression influences both ag-gressive and constructive responses from employees. The costs ofsupervisor aggression are considerable (Tepper et al., 2006). Ingeneral, supervisor aggression research suggests organizationaldecision makers should take notice and monitor and train super-visors’ effective interactions with their subordinates. Still, weunderstand this may not fully address the issue. Employees decidewhether a supervisor is acting aggressively, regardless of whetheror not the supervisor is actually intending to harm the subordinate.This reality suggests organizations must also focus on employeesand try to promote constructive rather than destructive behavior. Afirst step in this direction is assuring that employees feel they havethe access to problem solving approaches.

Organizations should strive to inform employees about con-structive problem solving behaviors and present avenues for as-sisting employees with interpersonal conflicts at work (e.g., opendoor policies that may enhance reconciling behavior, safe report-ing mechanisms). The results suggest a range of such options maybe most helpful. For example, internals like to take charge andactively pursue change, particularly with regard to threats in thework environment. The results show that internals were morelikely to engage in problem solving behaviors that allowed them to

try to resolve the problem directly when these options are availableto them.

Additionally, the results of our study suggest behavioral controlsmay assist organizations in inhibiting destructive and enhancingconstructive reactions to supervisor aggression. For example,higher fear of retaliation lessened victims’ destructive reactions.We, again, do not suggest decision makers exact fear in this way.Prior research demonstrates positive approaches for controllingbehavior are a more effective means of promoting constructivebehavior than fostering fear and punishing employees (Trevino etal., 1999; Weaver et al., 1999a, 1999b). Organizations wouldbenefit from policies and programs that promote positive values(e.g., ethics, problem solving) and make these values apparent toemployees (Trevino et al., 1999; Weaver et al., 1999a, 1999b).Thus, creating organizational structures that protect employeesfrom aggressive behavior (i.e., punishments against those whoengage in workplace aggression, monitoring systems that ensureindividuals are treating each other respectfully) can safeguardemployees from supervisor aggression and may lessen the likeli-hood of employees engaging in destructive behavior themselves.Establishing a work environment that emphasizes respectful be-havior and protection of all employees can assist organizations inharnessing constructive rather than destructive interpersonal be-haviors.

Limitations

Like all studies, ours is not without limitations. One limita-tion is that our research did not examine all potential reactionsto perceived aggression. For example, other types of displacedaggression could be investigated in the future (e.g., aggressionagainst strangers, friends, and/or family members). Indeed, itwould be interesting to explore whether different situationaland individual factors make non-work targets of displacedaggression more likely. Additionally, our study examined threemoderator variables, and surely there are others that influenceemployees’ responses. We hope future researchers explorethese ideas further.

Further, Study 2 and 3 used self-reported data, which mayinfluence method bias (Podsakoff et al., 2003). We took severalsteps to reduce such concerns. In Study 2, data were collectedanonymously from respondents to reduce inhibitions about re-sponding to sensitive questions (e.g., supervisor aggression, ag-gressive modeling, retaliation, displaced aggression). In Study 3,participants were recruited via students. Past research has shownthat individuals underreport norm-violating behaviors if data arecollected with the assistance of organizational decision makers—neither design used this method. Research has shown that, with theprecautions we took, respondents are fairly candid in their re-sponses (Bennett & Robinson, 2000). Indeed, research has dem-onstrated self-report data are most appropriate when evaluatingvictims’ perceptions of and reactions to aggression (Aquino &Thau, 2009; Dalal, 2005; Tepper, 2007), and that such data aremore accurate than other-source reports (Dalal, 2005; Sackett,Berry, Wiemann, & Laczo, 2006). These efforts, in conjunctionwith the experimental design in Study 1, provide greater confi-dence in the results.


Conclusion

Workplace aggression researchers have spent considerable ef-fort investigating the nature and costs of aggression (see Neuman& Baron, 1998, for a review). An unfortunate reality for organi-zational decision makers is that supervisor aggression is on therise, and such acts are detrimental. Although research has shownthat some victims respond destructively to supervisor aggression,research has also shown that some victims react constructively(Bies & Tripp, 1996, 1998). Our research investigates the factorsthat influence victims’ responses and demonstrates that both indi-vidual and situational factors influence employee responses tosupervisor aggression. The results of our studies suggest organi-zations can lessen destructive reactions and enhance constructivereactions through the selection of their employees (having employ-ees with a strong LOC) and by enhancing constructive ways foremployees to respond to supervisor aggression, integrating behav-ioral controls that hold employees accountable for and punishdestructive behavior, and fostering effective (rather than aggres-sive) behavior among organizational members.

References

Adarves-Yorno, I., Postmes, T., & Haslam, S. A. (2007). Creative inno-vation or crazy irrelevance? The contribution of group norms and socialidentity to creative behavior. Journal of Experimental Social Psychol-ogy, 43, 410–416. doi:10.1016/j.jesp.2006.02.013

Anderson, C. A., & Bushman, B. J. (2002). Human aggression. AnnualReview of Psychology, 53, 27–51. doi:10.1146/annurev.psych.53.100901.135231

Aquino, K., & Douglas, S. (2003). Identity threat and antisocial behaviorin organizations: The moderating effects of individual differences, ag-gressive modeling, and hierarchical status. Organizational Behavior andHuman Decision Processes, 90, 195–208. doi:10.1016/S0749-5978(02)00517-4

Aquino, K., Galperin, B. L., & Bennett, R. J. (2004). Social status andaggressiveness as moderators of the relationship between interactionaljustice and workplace deviance. Journal of Applied Social Psychology,34, 1001–1029. doi:10.1111/j.1559-1816.2004.tb02581.x

Aquino, K., & Thau, S. (2009). Workplace victimization: Aggression fromthe target’s perspective. Annual Review of Psychology, 60, 717–741.doi:10.1146/annurev.psych.60.110707.163703

Aquino, K., Tripp, T. M., & Bies, R. J. (2001). How employees respond topersonal offense: The effects of blame attribution, victim status, andoffender status on revenge and reconciliation in the workplace. Journalof Applied Psychology, 86, 52–59. doi:10.1037/0021-9010.86.1.52

Aquino, K., Tripp, T. M., & Bies, R. J. (2006). Getting even or moving on?Status variables and procedural justice as predictors of revenge, forgive-ness, and reconciliation in organizations. Journal of Applied Psychology,91, 653–668. doi:10.1037/0021-9010.91.3.653

Arslan, C., Dilmac, B., & Hamarta, E. (2009). Coping with stress and traitanxiety in terms of locus of control: A study with Turkish universitystudents. Social Behavior and Personality, 37, 791–800. doi:10.2224/sbp.2009.37.6.791

Bandura, A. (1973). Aggression: A social learning analysis. EnglewoodCliffs, NJ: Prentice Hall.

Bandura, A. (1986). Social foundations of thought and action: A socialcognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1991). Social cognitive theory of moral thought and action. InW. Kurtines & J. Gewirtz (Eds.), Handbook of moral behavior anddevelopment: Theory, research and applications (pp. 45–103). Hillsdale,NJ: Erlbaum.

Baron, R. (1971). Exposure to an aggressive model and apparent proba-

bility of retaliation from the victim as determinants of adult aggressivebehavior. Journal of Experimental Social Psychology, 7, 343–355. doi:10.1016/0022-1031(71)90033-3

Baron, R. (2004). Workplace aggression and violence: Insights from basicresearch. In R. Griffin & A. O’Leary-Kelly (Eds.), The dark side oforganizational behavior (pp. 23–61). New York, NY: Jossey-Bass.

Bennett, R. J. (1998). Perceived powerlessness as a cause of deviance. InR. Griffin, A. O’Leary-Kelly, & J. Collins (Eds.), Dysfunctional behav-ior in organizations (pp. 221–339). Stamford, CT: JAI Press.

Bennett, R. J., & Robinson, S. L. (2000). The development of a measure ofworkplace deviance. Journal of Applied Psychology, 85, 349–360. doi:10.1037/0021-9010.85.3.349

Berkowitz, L. (1983). The experience of anger as a parallel process in thedisplay of impulsive, “angry” aggression. In R. G. Geen & E. I. Don-nerstein (Eds.), Aggression: Theoretical and empirical reviews (Vol. 1,pp. 103–134). New York, NY: Academic Press.

Berkowitz, L. (1989). Frustration–aggression hypothesis: Examination andreformulation. Psychological Bulletin, 106, 59–73. doi:10.1037/0033-2909.106.1.59

Bies, R. J., & Tripp, T. M. (1996). Beyond distrust: “Getting even” and theneed for revenge. In R. Kramer & T. Tyler (Eds.), Trust in organizations(pp. 246–260). Thousand Oaks, CA: Sage.

Bies, R. J., & Tripp, T. M. (1998). Two faces of the powerless: Copingwith tyranny in organizations. In R. Kramer & M. Neal (Eds.), Powerand influence in organizations (pp. 203–220). Thousand Oaks, CA:Sage.

Biron, M., & Bamberger, P. (2012). Aversive workplace conditions andabsenteeism: Taking referent group norms and supervisor support intoaccount. Journal of Applied Psychology, 97, 901–912. doi:10.1037/a0027437

Bureau of Justice Statistics. (1998). About 2 million people attacked orthreatened in the workplace every year [Press release]. Washington, DC:Author.

Bureau of Justice Statistics. (2005). Survey of workplace violence preven-tion [Press release]. Washington, DC: Author.

Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcis-sism, self-esteem, and direct and displaced aggression: Does self-love orself-hate lead to violence? Journal of Personality and Social Psychol-ogy, 75, 219–229. doi:10.1037/0022-3514.75.1.219

Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journalof Personality and Social Psychology, 63, 452–459. doi:10.1037/0022-3514.63.3.452

Carlson, D., Ferguson, M., Perrewe, P., & Whitten, D. (2011). The falloutfrom abusive supervision: An examination of subordinates and theirpartners. Personnel Psychology, 64, 937–961. doi:10.1111/j.1744-6570.2011.01232.x

Cohen, D., Nisbett, R. E., Bowdle, B. F., & Schwarz, N. (1996). Insult,aggression, and the southern culture of honor: An “experimental eth-nography.” Journal of Personality and Social Psychology, 70, 945–960.doi:10.1037/0022-3514.70.5.945

Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied multipleregression/correlation analysis for the behavioral sciences (3rd ed.).Hillsdale, NJ: Erlbaum.

Cortina, L. M., Magley, V. J., Williams, J. H., & Langhout, R. D. (2001).Incivility in the workplace: Incidence and impact. Journal of Occupa-tional Health Psychology, 6, 64–80. doi:10.1037/1076-8998.6.1.64

Dalal, R. S. (2005). A meta-analysis of the relationship between organi-zational citizenship behavior and counterproductive work behavior.Journal of Applied Psychology, 90, 1241–1255. doi:10.1037/0021-9010.90.6.1241

Deci, E. L., & Ryan, R. M. (1980). The empirical exploration of intrinsicmotivation processes. In L. Berkowitz (Ed.), Advances in experimentalsocial psychology (Vol. 13, pp. 39–80). New York, NY: AcademicPress.


De Cremer, D., van Dijke, M., & Mayer, D. M. (2010). Cooperating when“you” and “I” are treated fairly: The moderating role of leader proto-typicality. Journal of Applied Psychology, 95, 1121–1133. doi:10.1037/a0020419

Detert, J. R., Trevino, L. K., Burris, E. R., & Andiappan, M. (2007).Managerial modes of influence and counterproductivity in organiza-tions: A longitudinal business-unit-level investigation. Journal of Ap-plied Psychology, 92, 993–1005. doi:10.1037/0021-9010.92.4.993

DeWall, C. N., Baumeister, R. F., Stillman, T. F., & Gailliot, M. T. (2007).Violence restrained: Effects of self-regulation and its depletion on ag-gression. Journal of Experimental Social Psychology, 43, 62–76. doi:10.1016/j.jesp.2005.12.005

DeWall, C. N., Twenge, J. M., Gitter, S. A., & Baumeister, R. F. (2009).It’s the thought that counts: The role of hostile cognition in shapingaggressive responses to social exclusion. Journal of Personality andSocial Psychology, 96, 45–59. doi:10.1037/a0013196

Dollard, J., Doob, L. W., Miller, N. E., Mowrer, O. H., & Sears, R. R.(1939). Frustration and aggression. New Haven, CT: Yale UniversityPress.

Dunlop, P. D., & Lee, K. (2004). Workplace deviance, organizationalcitizenship behavior, and business unit performance: The bad apples dospoil the whole barrel. Journal of Organizational Behavior, 25, 67–80.doi:10.1002/job.243

Dupre, K. E., Inness, M., Connelly, C. E., Barling, J., & Hoption, C.(2006). Workplace aggression in teenage part-time employees. Journalof Applied Psychology, 91, 987–997. doi:10.1037/0021-9010.91.5.987

Ehrhart, M. G., & Naumann, S. E. (2004). Organizational citizenshipbehavior in work groups: A group norms approach. Journal of AppliedPsychology, 89, 960–974. doi:10.1037/0021-9010.89.6.960

Eron, L. D. (1987). The development of aggressive behavior from theperspective of a developing behaviorism. American Psychologist, 42,435–442. doi:10.1037/0003-066X.42.5.435

Fisher, A. (2005). How to prevent violence at work. Fortune, 151, 42.Fiske, S. T., & Taylor, S. E. (1984). Social cognition. New York, NY:

Random House.Fox, S., & Spector, P. E. (1999). A model of work frustration–aggression.

Journal of Organizational Behavior, 20, 915–931. doi:10.1002/(SICI)1099-1379(199911)20:6�915::AID-JOB918�3.0.CO;2-6

Gatchel, R. J. (1980). Perceived control: A review and evaluation oftherapeutic implications. In A. Baum & J. Singer (Eds.), Advances inenvironmental psychology (pp. 1–22). Hillsdale, NJ: Erlbaum.

Glomb, T. M., & Liao, H. (2003). Interpersonal aggression in work groups:Social influence, reciprocal, and individual effects. Academy of Man-agement Journal, 46, 486–496. doi:10.2307/30040640

Gouldner, A. (1960). The norm of reciprocity. American SociologicalReview, 25, 161–178. doi:10.2307/2092623

Grant, A. M., & Mayer, D. M. (2009). Good soldiers and good actors:Prosocial and impression management motives as interactive predictorsof affiliative citizenship behaviors. Journal of Applied Psychology, 94,900–912. doi:10.1037/a0013770

Greenbaum, R. L., Mawritz, M. B., & Eissa, G. (2012). Bottom-linementality as an antecedent of social undermining and the moderatingroles of core self-evaluations and conscientiousness. Journal of AppliedPsychology, 97, 343–359. doi:10.1037/a0025217

Greenberg, J. (1993). Stealing in the name of justice: Informational andinterpersonal moderators of theft reactions to underpayment inequity.Organizational Behavior and Human Decision Processes, 54, 81–103.doi:10.1006/obhd.1993.1004

Greenberg, J., & Folger, R. (1988).Controversial issues in social researchmethods. New York, NY: Springer.

Greenberger, D., & Strasser, S. (1991). The role of situational and dispo-sitional factors in the enhancement of personal control in organizations.Research in Organizational Behavior, 13, 111–145.

Gurchiek, K. (2005). Workplace violence on the upswing. HR Magazine,50, 27–28.

Hahn, S. E. (2000). The effects of locus of control on daily exposure,coping and reactivity to work interpersonal stressors: A diary study.Personality and Individual Differences, 29, 729–748. doi:10.1016/S0191-8869(99)00228-7

Harris, K. J., Kacmar, K. M., & Zivnuska, S. (2007). An investigation ofabusive supervision as a predictor of performance and the meaning ofwork as a moderator of the relationship. Leadership Quarterly, 18,252–263. doi:10.1016/j.leaqua.2007.03.007

Heath, C., & Sitkin, S. B. (2001). Big-B versus Big-O: What is organiza-tional about organizational behavior? Journal of Organizational Behav-ior, 22, 43–58. doi:10.1002/job.77

Hershcovis, M. S., & Barling, J. (2010). Toward a multi-foci approach toworkplace aggression: A meta-analytic review of outcomes from differ-ent perpetrators. Journal of Organizational Behavior, 31, 24–44. doi:10.1002/job.621

Hershcovis, M. S., Turner, N., Barling, J., Arnold, K. A., Dupre, K. E.,Innes, M., . . . Sivanathan, N. (2007). Predicting workplace aggression:A meta-analysis. Journal of Applied Psychology, 92, 228–238. doi:10.1037/0021-9010.92.1.228

Hoobler, J. M., & Brass, D. J. (2006). Abusive supervision and familyundermining as displaced aggression. Journal of Applied Psychology,91, 1125–1133. doi:10.1037/0021-9010.91.5.1125

Inness, M., Barling, J., & Turner, N. (2005). Understanding supervisor-targeted aggression: A within-person between-jobs design. Journal ofApplied Psychology, 90, 731–739. doi:10.1037/0021-9010.90.4.731

Jehn, K. A. (1995). A multimethod examination of the benefits anddetriments of intragroup conflict. Administrative Science Quarterly, 40,256–282. doi:10.2307/2393638

Keashly, L. (1998). Emotional abuse in the workplace: Conceptual andempirical issues. Journal of Emotional Abuse, 1, 85–117. doi:10.1300/J135v01n01_05

Keashly, L., Trott, V., & MacLean, L. M. (1994). Abusive behavior in theworkplace: A preliminary investigation. Violence and Victims, 9, 341–357.

Lazarus, R. S. (1966). Psychological stress and the coping process. NewYork, NY: McGraw-Hill.

Lazarus, R. S., & Folkman, S. (1984). Coping and adaptation. In W. D.Gentry (Ed.), The handbook of behavioral medicine (pp. 282–325). NewYork, NY: Guilford Press.

Levine, J. M., Higgins, E. T., & Choi, H.-S. (2000). Development ofstrategic norms in groups. Organizational Behavior and Human Deci-sion Processes, 82, 88–101. doi:10.1006/obhd.2000.2889

Lian, H., Ferris, D. L., & Brown, D. (2012). Does power distance exac-erbate or mitigate the effects of abusive supervision? It depends on theoutcome. Journal of Applied Psychology, 97, 107–123. doi:10.1037/a0024610

Marcus, B., & Schuler, H. (2004). Antecedents of counterproductive be-havior at work: A general perspective. Journal of Applied Psychology,89, 647–660. doi:10.1037/0021-9010.89.4.647

Marcus-Newhall, A., Pedersen, W. C., Carlson, M., & Miller, N. (2000).Displaced aggression is alive and well: A meta-analytic review. Journalof Personality and Social Psychology, 78, 670–689. doi:10.1037/0022-3514.78.4.670

Martinko, M. J., & Gardner, W. L. (1982). Learned helplessness: Analternative explanation for performance deficits. Academy of Manage-ment Review, 7, 195–204.

Martinko, M. J., Gundlach, M. J., & Douglas, S. C. (2002). Toward anintegrative theory of counterproductive workplace behavior: A casualreasoning perspective. International Journal of Selection and Assess-ment, 10, 36–50. doi:10.1111/1468-2389.00192

Miller, N. (1941). The frustration–aggression hypothesis. PsychologicalReview, 48, 337–342. doi:10.1037/h0055861


Mischel, W. (1973). Toward a cognitive social learning reconceptualiza-tion of personality. Psychological Review, 80, 252–283. doi:10.1037/h0035002

Mitchell, M. S., & Ambrose, M. (2007). Abusive supervision and work-place deviance and the moderating effects of negative reciprocity beliefs.Journal of Applied Psychology, 92, 1159–1168. doi:10.1037/0021-9010.92.4.1159

Neuman, J., & Baron, R. (1998). Workplace violence and workplaceaggression: Evidence concerning specific forms, potential causes, andpreferred targets. Journal of Management, 24, 391–419.

Neuman, J., & Keashly, L. (2003, April). Workplace bullying: Persistentpatterns of workplace aggression. Paper presented at the meeting of theSociety for Industrial and Organizational Psychology, Orlando, FL.

Ng, T. W. H., Sorensen, K. L., & Eby, L. T. (2006). Locus of control atwork: A meta-analysis. Journal of Organizational Behavior, 27, 1057–1087. doi:10.1002/job.416

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.).New York, NY: McGraw-Hill.

Osterman, K., Bjorkqvist, K., Lagerspetz, K. M. J., Charetier, S., Caprara,G. V., & Pastorelli, C. (1999). Locus of control and three types ofaggression. Aggressive Behavior, 25, 61–65. doi:10.1002/(SICI)1098-2337(1999)25:1�61::AID-AB6�3.0.CO;2-G

Pachter, B. (2003). Bosses behaving badly. Harvard Business Review, 81,14–15.

Parkes, K. R. (1984). Locus of control, cognitive appraisal, and coping instressful episodes. Journal of Personality and Social Psychology, 46,655–668. doi:10.1037/0022-3514.46.3.655

Paulhus, D. (1983). Sphere-specific measures of perceived control. Journalof Personality and Social Psychology, 44, 1253–1265. doi:10.1037/0022-3514.44.6.1253

Paulhus, D. (1991). Measurement and control of response bias. In J. P.Robinson, P. Shaver, & L. Wrightsman (Eds.), Measures of psychologyand social psychological attitudes (pp. 17–59). San Diego, CA: Aca-demic Press.

Perrewe, P., & Spector, P. E. (2002). Personality research in the organi-zational sciences. Research in Personnel and Human Resources Man-agement, 21, 1–63. doi:10.1016/S0742-7301(02)21001-4

Phares, E. J. (1976). Locus of control personality. Morristown, NJ: GeneralLearning Press.

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003).Common method bias in behavioral research: A critical review of theliterature and recommended remedies. Journal of Applied Psychology,88, 879–903. doi:10.1037/0021-9010.88.5.879

Porath, C. L., & Erez, A. (2007). Does rudeness really matter? The effectsof rudeness on task performance and helpfulness. Academy of Manage-ment Journal, 50, 1181–1197. doi:10.2307/20159919

Robinson, S., & O’Leary-Kelly, A. (1998). Monkey see, monkey do: Theinfluence of work groups on the antisocial behavior of employees.Academy of Management Journal, 41, 658–672. doi:10.2307/256963

Romi, S., & Itskowitz, R. (1990). The relationship between locus of controland type of aggression in middle-class and culturally deprived children.Personality and Individual Differences, 11, 327–333. doi:10.1016/0191-8869(90)90214-C

Rotter, J. B. (1966). Generalized expectancies for internal versus externalcontrol of reinforcement. Psychological Monographs, 80 (Whole No.609), 1–28.

Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity domi-nance, and contagion. Personality and Social Psychology Review, 5,296–320. doi:10.1207/S15327957PSPR0504_2

Rusbult, C. E., Farrell, D., Rogers, G., & Mainous, A. G., III. (1988).Impact of exchange variables on exit, voice, loyalty, and neglect: Anintegrative model of responses to declining job satisfaction. Academy ofManagement Journal, 31, 599–627. doi:10.2307/256461

Ryan, R. M., & Deci, E. L. (2006). Self-regulation and the problem of

human autonomy: Does psychology need choice, self-determination, andwill? Journal of Personality, 74, 1557–1586. doi:10.1111/j.1467-6494.2006.00420.x

Sackett, P. R., Berry, C. M., Wiemann, S. A., & Laczo, R. M. (2006).Citizenship and counterproductive behavior: Clarifying relations be-tween the two domains. Human Performance, 19, 441– 464. doi:10.1207/s15327043hup1904_7

Salancik, G. R., & Pfeffer, J. (1978). A social information processingapproach to job attitudes and task design. Administrative Science Quar-terly, 23, 224–253. doi:10.2307/2392563

Saucier, G. (1997). Effects of variable selection on the factor structure ofperson descriptors. Journal of Personality and Social Psychology, 73,1296–1312. doi:10.1037/0022-3514.73.6.1296

Schat, A. C. H., Desmarais, S., & Kelloway, E. K. (2005, August). TheWorkplace Aggression Questionnaire (WAQ): Validation of a measureof exposure to aggression at work. Paper presented at the meeting of theAcademy of Management, Honolulu, HI.

Sears, R. R. (1941). Non-aggressive reactions to frustration. PsychologicalReview, 48, 343–346. doi:10.1037/h0059717

Sears, R. R. (1948). Non-aggressive reactions to frustration. PsychologicalReview, 48, 337–342.

Sinaceur, M., Van Kleef, G., Neale, M., Adam, H., & Haag, C. (2011). Hotor cold: Is communicating anger or threats more effective in negotiation?Journal of Applied Psychology, 96, 1018–1032. doi:10.1037/a0023896

Skarlicki, D. P., & Folger, R. (2004). Broadening our understanding oforganizational retaliatory behavior. In R. W. Griffin & A. M. O’Leary-Kelly (Eds.), The dark side of organizational behavior (pp. 373–402).San Francisco, CA: Jossey-Bass.

Spector, P. E. (1982). Behavior in organizations as a function of employ-ee’s locus of control. Psychological Bulletin, 91, 482–497. doi:10.1037/0033-2909.91.3.482

Storms, P. L., & Spector, P. E. (1987). Relationships of organizationalfrustration with reported behavioral reactions: The moderating effect oflocus of control. Journal of Occupational Psychology, 60, 227–234.doi:10.1111/j.2044-8325.1987.tb00255.x

Taggar, S., & Ellis, R. (2007). The role of leaders in shaping formal teamnorms. Leadership Quarterly, 18, 105–120. doi:10.1016/j.leaqua.2007.01.002

Tedeschi, J., & Felson, R. (1994). Violence, aggression, and coerciveactions. Washington, DC: American Psychological Association.

Tepper, B. J. (2000). Consequences of abusive supervision. Academy ofManagement Journal, 43, 178–190. doi:10.2307/1556375

Tepper, B. J. (2007). Abusive supervision in work organizations: Review,synthesis, and research agenda. Journal of Management, 33, 261–289.doi:10.1177/0149206307300812

Tepper, B. J., Carr, J. C., Breaux, D. M., Geider, S., Hu, C., & Hua, W.(2009). Abusive supervision, intentions to quit, and employees’ work-place deviance: A power/dependence analysis. Organizational Behaviorand Human Decision Processes, 109, 156 –167. doi:10.1016/j.obhdp.2009.03.004

Tepper, B. J., Duffy, M. K., Henle, C. A., & Lambert, L. S. (2006).Procedural injustice, victim precipitation, and abusive supervision. Per-sonnel Psychology, 59, 101–123. doi:10.1111/j.1744-6570.2006.00725.x

Tepper, B. J., Duffy, M. K., & Shaw, J. D. (2001). Personality moderatorsof the relationship between abusive supervision and subordinates’ resis-tance. Journal of Applied Psychology, 86, 974–983. doi:10.1037/0021-9010.86.5.974

Tepper, B. J., Henle, C., Lambert, L., Giacalone, R., & Duffy, M. K.(2008). Abusive supervision and subordinates’ organizational deviance.Journal of Applied Psychology, 93, 721–732. doi:10.1037/0021-9010.93.4.721

Tepper, B. J., Moss, S. E., Lockhart, D. E., & Carr, J. C. (2007). Abusivesupervision, upward maintenance communications, and subordinates’


psychological distress. Academy of Management Journal, 50, 1169–1180. doi:10.2307/20159918

Thau, S., Bennett, R. J., Mitchell, M. S., & Marrs, M. (2009). Howmanagement style moderates the relationship between abusive supervi-sion and workplace deviance: An uncertainty management theory per-spective. Organizational Behavior and Human Decision Processes, 108,79–92. doi:10.1016/j.obhdp.2008.06.003

Thau, S., & Mitchell, M. S. (2010). Self-gain or self-regulation impair-ment? Tests of competing explanations of the supervisor abuse andemployee deviance relationship through perceptions of distributive jus-tice. Journal of Applied Psychology, 95, 1009–1031. doi:10.1037/a0020540

Thelen, J. B. (2009). Is that a threat? An employer’s response to threats byemployees can prevent workplace violence. HR Magazine, 54, 61–63.

Trevino, L. K., Weaver, G. R., Gibson, D. G., & Toffler, B. L. (1999).Managing ethics and legal compliance: What works and what hurts.California Management Review, 41, 131–151.

Tripp, T., & Bies, R. (1997). What’s good about revenge. In R. Lewicki,R. Bies, & B. Sheppard (Eds.), Research on negotiation in organizations(Vol. 6, pp. 145–160). Greenwich, CT: JAI Press.

Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001). Ifyou can’t join them, beat them: Effects of social exclusion on aggressivebehavior. Journal of Personality and Social Psychology, 81, 1058–1069.doi:10.1037/0022-3514.81.6.1058

Wade, S. H. (1989). The development of a scale to measure forgiveness(Unpublished doctoral dissertation). Fuller Theological Seminary, Pas-adena, CA.

Wanberg, C. R. (1997). Antecedents and outcomes of coping behaviors

among unemployed and reemployed individuals. Journal of AppliedPsychology, 82, 731–744. doi:10.1037/0021-9010.82.5.731

Weaver, G. R., Trevino, L. K., & Cochran, P. L. (1999a). Corporate ethicspractices in the mid-1990’s: An empirical study of the Fortune 1000.Journal of Business Ethics, 18, 283–294. doi:10.1023/A:1005726901050

Weaver, G. R., Trevino, L. K., & Cochran, P. L. (1999b). Corporate ethicsprograms as control systems: Influences of executive commitment andenvironmental factors. Academy of Management Journal, 42, 41–57.doi:10.2307/256873

Wei, F., & Si, S. (2011). Tit for tat? Abusive supervision and counterpro-ductive behaviors: The moderating effects of locus of control andperceived mobility. Asia Pacific Journal of Management. Advanceonline publication. doi:10.1007/s10490-011-9251-y

Weisz, J. R., McCabe, M. A., & Denning, M. D. (1994). Primary andsecondary control among children undergoing medical procedures: Ad-justment as a function of coping style. Journal of Consulting andClinical Psychology, 62, 324–332. doi:10.1037/0022-006X.62.2.324

Westring, A. F., & Ryan, A. M. (2010). Personality and inter-role conflictand enrichment: Investigating the mediating role of support. HumanRelations, 63, 1815–1834. doi:10.1177/0018726710371236

Wortman, C. B., & Brehm, J. W. (1975). Responses to uncontrollableoutcomes: An integration of reactance theory and the learned helpless-ness model. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 8, pp. 277–336). New York, NY: Academic Press.

Zainuddin, R., & Taluja, H. (1990). Aggression and locus of control amongundergraduate students. Journal of Personality and Clinical Studies, 6,211–215.

Appendix

Summary of the Findings Across the Three Studies

Prediction Study 1 Study 2 Study 3

H1: Moderating effects of LOC(a) on problem solving — ns Mixed support(b) on retaliation Supported Supported Supported(c) on displaced aggression ns ns Supported

H2: Moderating effects of fear of retaliation(a) on problem solving — Supported ns(b) on retaliation Supported Supported Supported

H3: Moderating effects of fear of retaliation on displaced aggression(a) Frustration–aggression view — — —(b) Learned inhibitions view ns Supported Supported

H4: Moderating effects of behavioral modeling(a) Aggressive modeling on retaliation — Supported Supported(b) Aggressive modeling on displaced aggression — Supported Supported(c) Problem solving modeling on problem solving — ns Mixed support

Note. H1 � Hypothesis 1; H2 � Hypothesis 2; H3 � Hypothesis 3; H4 � Hypothesis 4; LOC � locus of control; ns �not significant; a dash indicates that the prediction was not tested or, for H3 (opposing predictions about the moderatingeffects of fear of retaliation on coworker displaced aggression reactions to supervisor aggression), that the opposing viewwas not supported.

Received July 17, 2010Revision received June 18, 2012

Accepted June 25, 2012 �


Employees Behavioral Reactions to Supervisor Aggression ... · Employees Behavioral Reactions to...

Documents

Transcript of Employees Behavioral Reactions to Supervisor Aggression ... · Employees Behavioral Reactions to...