university of colorado boulder | university of colorado boulder -...

THE DARK SIDE OF LEADERSHIP:A THREE-LEVEL INVESTIGATION OF THE CASCADING

EFFECT OF ABUSIVE SUPERVISION ON EMPLOYEECREATIVITY

DONG LIUGeorgia Institute of Technology

HUI LIAOUniversity of Maryland

RAYMOND LOIUniversity of Macau

This research sheds light on the role of the dark side of leadership in employeecreativity by examining how and when department leader abusive supervision mayflow down organizational levels to undermine team member creativity. Analyses ofmultiphase, multisource, and multilevel data show that team leader abusive supervi-sion mediates the negative relationship between department leader abusive supervi-sion and team member creativity. Team leaders’ and members’ attributions for themotives behind their own supervisors’ abusive supervision, which we classify asperformance-promotion and injury-initiation motives, determine the extent to whichteam leader abusive supervision accounts for the effect of department leader abusivesupervision on team member creativity.

Because of the rapidly changing economy andcontinuing globalization of business, employee cre-ativity—referring to the development of novel anduseful ideas about products, practices, services orprocedures—has become increasingly crucial forthe survival and competitiveness of organizationstoday (Shalley, Gilson, & Blum, 2009). A plethoraof research has looked at the link between positiveleader behaviors such as “transformational leader-ship” and employee creativity (e.g., Shin & Zhou,2003). Nevertheless, existing knowledge on the roleof leadership in employee creativity remains in-complete because little is known as to whether thedark side of leadership in general and abusive su-pervision in particular may affect creative perfor-mance of employees. Drawing on a comprehensivereview of psychological studies, Baumeister, Brat-slavsky, Finkenauer, and Vohs (2001) concludedthat a general principle across a broad range of

psychological phenomena is that individuals aremore responsive to negative than to positive as-pects of external context, so the negative contextualaspects thus tend to have stronger influence onindividual attitudes and behaviors than the posi-tive ones. More recently, Zhou (2010) pointed outthat a critical and fruitful future avenue for re-search on creativity would be to explore the rela-tionship between the negative side of leadershipand employee creativity. Given the prevalence andfar-reaching impact of abusive supervision (Tep-per, 2007; Tepper, Carr, Breaux, Geider, Hu, & Hua,2009), it is of both theoretical and empirical impor-tance to illuminate the influence of abusive super-vision on employee creativity as well as the bound-ary conditions for this influence.

Tepper conceptualized abusive supervision asleaders’ engagement in “the sustained display ofhostile, verbal and nonverbal behaviors, excludingphysical contact” (2000: 178). A number of studieshave reported that, owing to their higher organiza-tional positions and stronger decisional power,leaders are inclined to exhibit abusive supervisorybehaviors such as ridiculing, yelling at, and intim-idating subordinates; taking credit for subordi-nates’ achievements; and attributing undesirableoutcomes to subordinates’ personal factors (e.g.,Hoobler & Brass, 2006; Tepper et al., 2009). Tepper,

Editor’s note: The manuscript for this article was ac-cepted for publication during the term of AMJ’s formereditor-in-chief, R. Duane Ireland.

We thank our action editor, Jason Colquitt, three anon-ymous reviewers. and Subrahmaniam Tangirala for help-ful comments and suggestions. We are also grateful toEdward Timothy and Susan Lau for their invaluable as-sistance in data collection.

� Academy of Management Journal2012, Vol. 55, No. 5, 1187–1212.http://dx.doi.org/10.5465/amj.2010.0400

1187

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

Duffy, Henle, and Lambert (2006) estimated thatU.S. companies incur a tremendous annual cost of$23.8 billion as a result of abusive supervision’snegative influence on employees. To date, supervi-sory abuse has been found to result in a variety ofattitudinal outcomes among subordinates, such asdecreased job satisfaction (Tepper, Duffy, Hoobler,& Ensley, 2004) and increased intentions to quit(Tepper, 2000). Nevertheless, only a small numberof studies have focused directly on the associationsbetween abusive supervision and subordinate per-formance outcomes (for exceptions, see Aryee,Chen, Sun, and Debrah [2007] and Zellars, Tepper,and Duffy [2002]), and no research has been con-ducted on the impact of abusive supervision onemployee creative performance. This is an unfortu-nate oversight both practically and theoretically.Practically, individual creativity contributes sub-stantively to organizational effectiveness (Shalley,Zhou, & Oldham, 2004), and leader behaviors havealso been argued to play a significant role in thegrowth and prohibition of creativity (Hennessey &Amabile, 2010). Theoretically, in view of extantabusive supervision and creativity research, we ex-pect abusive supervision to decrease employee cre-ativity. Therefore, the first purpose of our researchis to integrate and extend research on leadershipand creativity to examine whether abusive super-vision undermines employee creativity in teams.

In addition, leadership research has highlighteda cascading effect of positive leader behaviors:leaders at a lower level of a hierarchy, may engagein role modeling in which they mimic and displaythe positive behaviors of leaders at a higher hierar-chical level (Bass, 1990; Bass, Waldman, Avolio, &Webb, 1987). More recently, Mayer, Kuenzi, Green-baum, Bardes, and Salvador (2009) presented con-vincing evidence, based on data from a sample of904 employees and 195 supervisors in 195 depart-ments, for the cascading effect of positive aspects ofleadership by demonstrating that top managementethical leadership triggers supervisory ethical lead-ership. A second purpose of this research is tocontribute to the leadership literature, and espe-cially the research on abusive supervision, throughextending the idea that negative aspects of leader-ship may have a cascading effect and investigatingwhether team leader abusive supervision may be afunction of department leader abusive supervision.More specifically, we examine the indirect effect ofdepartment leader abusive supervision on teammember creativity via team leader abusivesupervision.

Another critical issue of note in theory on theimpact of abusive supervision is the role of attribu-tion in subordinates’ perceptions of abuse. In his

review of abusive supervision research, Tepper(2007) pointed out that an important future re-search direction is to scrutinize how subordinatesmay respond differently to supervisory abuse de-pending on the attributed motives for supervisors’abusive behaviors. Prior research on abusive super-vision has implied that two separate types of mo-tives may be associated with supervisory abuse:performance promotion and injury initiation (Tep-per, 2007). Thus, on the one hand, leaders maymistreat their subordinates to enhance subordinateperformance; on the other hand, leaders may exer-cise abusive supervision to purposely harm subor-dinates. Answering the call to incorporate subordi-nates’ attributions for abusive supervision intoexamination of abusive supervision’s effects onsubordinate outcomes (Tepper, 2007), we draw onthe attribution literature (Heider, 1958; Martinko,Harvey, & Douglas, 2007a) to examine when depart-ment leader supervision cascades down to triggerteam leader abusive supervision and in turn, de-creases team member creativity. Specifically, thisresearch looks at the contingent effects of teamleader–attributed performance promotion motivesand injury initiation motives on the link betweendepartment leader abusive supervision and teamleader abusive supervision. In addition, at two lev-els down the organizational hierarchy from depart-ments, we investigate team member–attributed per-formance promotion motives and injury initiationmotives for team leader abusive supervision as theboundary condition of the negative relationship be-tween team leader abusive supervision and teammember creativity.

The current research tests a three-level theoreti-cal model of the relationship between abusive su-pervision and creativity and is designed to makethree unique contributions. First, we offer a theo-retical rationale for and provide the first empiricaltest of whether abusive supervision exerts a nega-tive effect on employee creativity. An examinationof this effect is likely to advance current under-standing of the consequences of the dark side ofleadership in organizations. Second, we build atrickle-down model to unveil how abusive super-vision manifested at the department level flowsdown through an organizational hierarchy to stim-ulate team leader abusive supervision and conse-quently undermines team member creativity. Thismultilevel investigation of the antecedents and out-comes of team leader abusive supervision is cru-cial, in that abusive supervision scholars have in-creasingly recognized the value of a multilevelperspective on the dynamics and richness of abu-sive behaviors that are likely to traverse differentorganizational levels (e.g., Tepper, 2007). Third, we

1188 OctoberAcademy of Management Journal

extend current thinking about the trickle-down ef-fect of leader behaviors by drawing on research onattribution (Heider, 1958) to clarify the boundaryconditions of such an effect on creativity: that is,the contingent roles of subordinates’ attributionsfor the motives behind leaders’ abusive supervi-sion. This examination contributes to both the lead-ership and attribution literatures by demonstratingthe ways in which subordinate attributional per-spectives can affect leadership processes andconsequences.

LITERATURE REVIEW AND HYPOTHESES

The Effect of Abusive Supervision on EmployeeCreativity in Teams

Leaders have traditionally been conceptualizedas an important contextual factor that cultivates orstifles employee creativity (George, 2008). Al-though extant literature has not examined the effectof abusive supervision on creativity, a limited butgrowing body of abusive supervision research hasdemonstrated that exposure to abusive supervisionresults in subordinates’ unwillingness to “go theextra mile” to perform behaviors that benefit theirorganizations (e.g., Zellars et al., 2002), which mayinvolve advancing creative ideas and solutions thatimprove organizational effectiveness. Drawing onprior creativity research, we argue that team leaderabusive supervision may undermine team membercreativity because it reduces team member intrinsicmotivation, which refers to the degree to which anindividual undertakes an activity for the sake ofhis/her enjoyment of and interest in the activityitself, rather than as a result of external pressuresand rewards (Deci, 1972). Intrinsic motivation isconducive to creativity because the more intrinsi-cally motivated toward their jobs employees are,the more likely they are to challenge the status quo,come up with novel and useful ideas, and adhere toinnovative goals in the face of challenges, therebybecoming more creative (Deci & Ryan, 2008; Zhou& George, 2001).

Comprehensive reviews of the creativity litera-ture suggest that intrinsic motivation as an internalprocess that leads to employee creativity has gar-nered the most significant attention from organiza-tional scholars (e.g., George, 2008; Shalley et al.,2004). In particular, a number of conceptual andempirical studies have generated evidence for in-trinsic motivation as a pivotal psychological mech-anism that underlies creativity (e.g., Amabile, 1996;Grant & Berry, 2011). Shalley and colleaguespointed out that “each contextual characteristic af-fects creativity via its effects on employees’ ‘intrin-

sic motivation’ to perform a work assignment”(2004: 935). We argue that abusive supervision mayalso dampen employees’ intrinsic motivation andhence their creativity.

When team members encounter abuse by leaders,in the form of public criticism, derogating com-ments, loud and angry tantrums, rudeness, incon-siderate actions, and coercion, they are apt to feelbelittled, humiliated, and undermined as to theirreputation in the workplace (Keashly & Harvey,2005). Abusive supervision also leads subordinatesto doubt whether organizations respect their con-tributions and whether their jobs are meaningful totheir own and organizations’ development (Rafferty& Restubog, 2011). Studies have demonstrated thatabusive supervision results in employees being un-satisfied with their jobs and intending to quit (Tep-per, 2000). Accordingly, abusive supervisionshould reduce employees’ enjoyment of their jobs,thereby causing diminished intrinsic motivationtowards their jobs. In addition, abusive supervisionis viewed as a significant source of psychologicaldistress (Restubog, Scott, & Zagenczyk, 2011).Abused employees often suffer from depression,anxiety, and emotional exhaustion, and they tendto alienate themselves from their jobs (Aryee et al.,2007; Hoobler & Brass, 2006; Tepper et al., 2004). Insuch a distressed psychological state, abused em-ployees may have little chance of developing inter-est in their work, so their intrinsic motivationshould decline substantially (Deci & Ryan, 2008).

Research has shown that the supportive relation-ships between leaders and subordinates enhancesubordinate intrinsic motivation (Shin & Zhou,2003), whereas pressuring, controlling, and dispar-aging relationships between leaders and subordi-nates, which typically result from abusive supervi-sion (Tepper, 2007), reduce intrinsic motivation(Deci, Connell, & Ryan, 1989; Vansteenkiste, Si-mons, Lens, Sheldon, & Deci, 2004). Consideringthe above arguments as well as abundant empiricalevidence of the positive relationship between in-trinsic motivation and creativity (e.g., Amabile,Hadley, & Kramer, 2002; Grant & Berry, 2011), wepropose:

Hypothesis 1. Team leader abusive supervisionis negatively related to team member creativity.

The Trickle-down Effect of Abusive Supervisionin Organizational Hierarchy

Drawing on social learning theory (Bandura,1977, 1986), we further propose that abusive super-vision on the part of a team leader’s own supervisor(i.e., a department leader) can directly trigger the

2012 1189Liu, Liao, and Loi

team leader’s abusive behaviors, thus indirectly de-creasing team member creativity. Social learningtheory suggests that by observing and emulating thebehavior of significant social contacts (e.g., lead-ers), individuals learn how to interpret externalstimuli and respond in a certain manner (Bandura,1986). In the workplace, leaders are in charge of theallotment of critical organizational resources, guideand evaluate followers on completing job tasks, andmake important personnel decisions (e.g., promo-tion, salary, training opportunities), so employeesare inclined to carefully observe their leaders’ be-haviors and decide how they should behave ac-cordingly (Mayer et al., 2009; Tucker, Turner,Barling, & McEvoy, 2010).

Extending the extant literature on the trickle-down effect of leaders’ behaviors (Mayer et al.,2009), we contend that employees may emulate notonly positive but also negative leader behaviors.Some support for our proposition comes from theaggression literature, which suggests that individ-uals engage in aggressive behaviors in part as aresult of learning from role models in a social set-ting (Bandura, 1973, 1977). Previous research onworkplace aggression (e.g., Glomb & Liao, 2003),child aggression (e.g., Tucker et al., 2010), and fam-ily violence (e.g., Brezina, 1999) has provided evi-dence for this social learning perspective on aggres-sion. In particular, Tucker and colleagues (2010)revealed that team leaders’ behaviors significantlyaffect the extent to which team members are aggres-sive. Goldstein (1986) also found that individualaggressive behaviors toward social contacts are cul-tivated by the similar behaviors of those in higherstatus positions.

In the context of abusive supervision, a role mod-eling process characterized by followers’ observa-tional learning of leaders’ abusive supervision maystimulate abusive supervision to flow down fromthe department level to the team level (Bandura,1986; Bass et al., 1987). Specifically, when teamleaders discern frequent occurrence of abuse bytheir department leaders, they tend to feel it isappropriate and acceptable to display abusive be-haviors toward their own subordinates and thusengage in such behaviors frequently. Conversely, ifthey observe little or no abuse by department lead-ers, team leaders may conclude that abusive super-vision is not legitimate or a norm in their respectivedepartments. As a result, they will follow depart-ment leaders in refraining from abusing their directreports—team members. For example, drawingfrom social learning theory, Lian, Ferris, andBrown (2012: 108) argued that “abusive supervi-sion can be ‘learned’ and mimicked by subordi-nates, potentially perpetuating an organizational

atmosphere of abuse.” Using three separate sam-ples, they demonstrated that subordinates modelbehavior on that of their abusive supervisors whenengaging in interpersonal deviance. We want tonote that subordinates do not necessarily learnleaders’ abusive behaviors purposely. Social learn-ing theory has emphasized that “most of the intri-cate responses people display are learned, eitherdeliberately or inadvertently, through the influenceof example” (Bandura, 1973: 44). Researchers haveindeed shown that individuals may mimic socialcontacts’ behaviors unintentionally and subcon-sciously (Chartrand & Bargh, 1999). Therefore, ateam leader may become increasingly abusive as aresult of the frequent exposure to an abusive de-partment leader, even without the team leader’sfull awareness.

Given that oftentimes, a department leader hasfrequent, direct contacts with team leaders, who inturn, supervise team members, department leaderabusive supervision is expected to weaken teammember creativity indirectly, via its influence onteam leader abusive behavior. Specifically, teamleader abusive supervision may serve as the centralpsychological conduit that links department leaderabusive supervision to team member creativity. Hi-erarchically proximal leaders have been demon-strated to have the strongest impact on subordi-nates’ perceptions and behaviors (Offermann &Malamut, 2002). As immediate followers of depart-ment leaders, team leaders are inclined to mimicsupervisory abuse from department leaders. Com-pared to departments, teams constitute a moreproximate work environment for team members,who thus have more frequent and intensive inter-actions with team leaders than department leaders.Consequently, department leaders generally needto rely on team leaders to influence team members;team leader abusive supervision as a function ofdepartment leader abusive supervision thus has amore direct and dominant impact on team membercreativity. Empirical evidence supporting this ideahas been accumulating. For examples, Zohar andLuria (2005) reported that as a hierarchically prox-imal factor for employees, supervisor safety leader-ship translates top management team safety leader-ship to employee safety outcomes. More recently,top management ethical leadership has been foundto have indirect effects on group citizenship behav-iors and deviance via group-level supervisory eth-ical leadership (Mayer et al., 2009). Therefore, wepropose:

Hypothesis 2. Team leader abusive supervisionmediates the cross-level relationship between


department leader abusive supervision andteam member creativity.

The Contingent Effects of Team LeaderAttributions on the Link between DepartmentLeader Abusive Supervision and Team LeaderAbusive Supervision

A social learning theory contention is that thedegree to which people mimic certain behaviors oftheir role models depends on three factors: reten-tion (remembering what behavior one has ob-served); reproduction (having the ability to repro-duce the behavior); and motivation (having a goodreason to justify the initiation of the behavior) (Ban-dura, 1973, 1977). The motivation component ofthe theory underscores the fact that the likelihoodof individuals choosing to enact the behaviorslearned from role models depends on their percep-tions of the purposes or the objectives of the behav-iors (Bandura, 1986). People generally wish to ob-tain positive learning outcomes and avoid negativelearning consequences (Bandura, 1977, 1986). Ifindividuals expect a positive outcome from a be-havior they observed, or believe the behavior wasenacted to achieve a favorable outcome, then theywill have higher motivation to adopt the behavior.On the contrary, if individuals expect a negativeoutcome or believe the behavior was enacted tocause harm, they will have less motivation to adoptthe behavior.

In a similar vein, the central finding of attribu-tion research is that people make causal explana-tions for the behaviors of the individuals aroundthem to adjust their own behaviors to social envi-ronments (Heider, 1958; Martinko et al., 2007a).Accordingly, when encountering abuse, employeesare inclined to develop causal attributions for thereason or the objectives of the actor’s abuse, whichresult in their behavioral responses. In his seminalreview of abusive supervision research, Tepper(2007) pointed out that a fundamental differencebetween abusive supervision and work aggressionis that subordinates may attribute abusive behav-iors to two distinct causal motives—to cause injuryand to accomplish an objective such as elicitinghigh performance—but they attribute work aggres-sion behaviors only to the motive of causing injury.Therefore, in this research, we examine the contin-gent roles of subordinate-attributed performancepromotion and injury initiation motives for leaders’abuse in qualifying the trickle-down effect of abu-sive supervision on team member creativity.

Informed by social learning theory and attribu-tion studies (Bandura, 1986; Mikula, 2003), we ar-gue that after observing abusive behaviors of de-

partment leaders, team leaders proceed to processrelevant information to discern the intended objec-tive of the behavior. When team leaders interpretthe department leaders’ motive as performance pro-motion, they may believe that department leaderabusive supervision is aligned with their own per-sonal interests and will be beneficial for their long-term career development. Because improved per-formance may bring benefits such as promotionand compensation (Carmeli, Shalom, & Weisberg,2007) as well as a greater sense of job enjoymentand accomplishment, subordinate-attributed per-formance promotion motives for leader abuse maystimulate subordinates to legitimize abusive super-vision. As a result, the subordinates will be morelikely to view their leaders as social role modelsthat they aspire to learn from and thus spread sim-ilar behavior down the organizational hierarchy intheir workplace. Therefore, to the extent that abu-sive supervision is attributed to improving subor-dinate performance, a team leader will be morelikely to model his or her own behavior on suchbehavior by the department leader, thereby abusinghis or her team members more. We thus argue thatthe positive relationship between department abu-sive supervision and team leader abusive supervi-sion may be accentuated by team leader–attributedperformance promotion motives.

In contrast, when team leaders interpret depart-ment leaders’ motives as injury initiation, they mayperceive abusive supervision as unethical andharmful to subordinates’ experiences in their organ-ization. People are discouraged from emulating andenacting unethical and harmful behaviors to avoidnegative consequences and inconsistency withmoral standards and social norms (Harman &Thomson, 1996). Therefore, in this case, team lead-ers should engage in abusive supervision less be-cause of ethical concern that abusive supervisionmay harm subordinates. Nonetheless, the per-ceived injury initiation motives may not lead indi-viduals to completely avoid abusing subordinates.As pointed out early, both social learning theoryand empirical studies have shown that people im-itate others’ behaviors both consciously and uncon-sciously (Bandura, 1973; Chartrand & Bargh, 1999).In other words, team leaders may learn from abu-sive department leaders and in turn, abuse teammembers without being fully aware of this sociallearning effect. Therefore, we expect the positivelink between department abusive supervision andteam leader abusive supervision to be attenuatedrather than canceled by team leader–attributed in-jury initiation motives. We propose:


Hypothesis 3a. Team leader–attributed mo-tives moderate the positive relationship be-tween department leader abusive supervisionand team leader abusive supervision: The re-lationship is stronger when team leaders inter-pret department leader abusive supervision asdriven by higher (vs. lower) levels of perfor-mance promotion motives.

Hypothesis 3b. Team leader–attributed mo-tives moderate the positive relationship be-tween department leader abusive supervisionand team leader abusive supervision: The re-lationship is weaker when team leaders inter-pret department leader abusive supervision asdriven by higher (vs. lower) levels of injuryinitiation motives.

The Contingent Effects of Team MemberAttributions on the Link between Team LeaderAbusive Supervision and Team MemberCreativity

Integrating attribution and abusive supervisionresearch, we further contend that team members’causal attributions for abusive behaviors of teamleaders may affect the ways they interpret and re-spond to team leader abusive supervision. Specifi-cally, team members’ causal attributions and teamleader abusive supervision may interact to influ-ence team members’ creativity.

Our earlier arguments suggest that team membersmay also decode the motives behind team leaderabusive supervision as performance promoting orinjury initiating. When team members perceiveteam leader abusive supervision as triggered by thepositive intent of improving their performance andthus as conducive to their personal growth, theteam members are less likely to form negative feel-ings and lose enjoyment of and interest in theirjobs. As a result, they are less prompted to with-draw from advancing novel and useful ideas intheir workplace. Therefore, team leader abusive su-pervision is less negatively related to team membercreativity in the presence of team member–attrib-uted performance promotion motives. In contrast,perceiving team leaders to be abusing subordinateswith the negative purpose of causing harm morelikely makes team members form various negativeemotions and lose interest in their jobs. As a result,the abused team members who make injury initia-tion attributions should exhibit a bigger decrease intheir desire to propose novel and useful ideas re-garding their jobs. Team member–attributed injuryinitiation motives thus exacerbate the negative roleof team leader abusive supervision in the develop-

ment of team member creative performance. There-fore, we hypothesize:

Hypothesis 4a. Team member–attributed mo-tives moderate the negative relationship be-tween team leader abusive supervision andteam member creativity: The relationship isweaker when team members interpret teamleader abusive supervision as driven by higher(vs. lower) levels of performance promotionmotives.

Hypothesis 4b. Team member–attributed mo-tives moderate the negative relationship be-tween team leader abusive supervision andteam member creativity: The relationship isstronger when team members interpret teamleader abusive supervision as driven by higher(vs. lower) levels of injury initiation motives.

An Integrative Moderated Mediation Model

Thus far, we have developed theoretical under-pinnings for the mediating effect of team leaderabusive supervision as well as for the contingenteffects of the causal motives subordinates attributeto supervisors for abuse at different organizationallevels. That is, team leader abusive supervisionmediates the relationship between departmentleader abusive supervision and team member cre-ativity (Hypothesis 2). Team leader–attributed mo-tives moderate the positive relationship betweendepartment leader abusive supervision and teamleader abusive supervision (Hypotheses 3a and 3b);and team member–attributed motives moderate thenegative relationship between team leader abusivesupervision and team member creativity (Hypoth-eses 4a and 4b). The theoretical rationales behindthe above hypotheses also suggest an integrativemoderated mediation model.

Specifically, subordinate-attributed motives forsupervisory abuse may moderate the indirect effectof department leader abusive supervision on teammember creativity through team leader abusive su-pervision. The theorizing behind Hypotheses 2, 3a,and 3b indicates that through augmenting or atten-uating the association between department leaderabusive supervision and team leader abusive super-vision, team leaders’ causal motive attributions af-fect the degree to which department leader abusivesupervision flows down organizational levels todamage team member creativity. Likewise, teammembers’ causal motive attributions, owing to theirmoderating influence on the link between teamleader abusive supervision and team member cre-ativity (Hypotheses 4a and 4b), may also hold thepotential of changing the indirect trickle-down ef-


fect of department leader abusive supervision onteam member creativity through team leader abu-sive supervision (Hypothesis 2). Taking these pre-dictions together, we propose two sets of integra-tive moderated mediation hypotheses:

Hypothesis 5a. Team leader–attributed mo-tives moderate the indirect negative effect ofdepartment leader abusive supervision onteam member creativity via team leader abu-sive supervision: The indirect negative effect isstronger when team leaders interpret depart-ment leader abusive supervision as driven byhigher (vs. lower) levels of performance promo-tion motives.

Hypothesis 5b. Team leader–attributed mo-tives moderate the indirect negative effect ofdepartment leader abusive supervision onteam member creativity via team leader abu-sive supervision: The indirect negative effect isweaker when team leaders interpret depart-ment leader abusive supervision as driven byhigher (vs. lower) levels of injury initiationmotives.

Hypothesis 6a. Team member–attributed mo-tives moderate the indirect negative effect ofdepartment leader abusive supervision onteam member creativity via team leader abu-

sive supervision: The indirect negative effect isweaker when team members interpret teamleader abusive supervision as driven by higher(vs. lower) levels of performance promotionmotives.

Hypothesis 6b. Team member–attributed mo-tives moderate the indirect negative effect ofdepartment leader abusive supervision onteam member creativity via team leader abu-sive supervision: The indirect negative effect isstronger when team members interpret teamleader abusive supervision as driven by higher(vs. lower) levels of injury initiation motives.

Figure 1 depicts the hypothesized theoreticalframework in this research.

METHODS

Scale Development for Subordinate-AttributedMotives for Abusive Supervision

No prior research has looked at victims’ attribu-tions for perpetrator’s intentions when studying therole of attributions in abusive supervision. There-fore, following Hinkin’s (1995, 1998) scale devel-opment procedures, we first adopted an inductivemultistage approach to developing a measure ofsubordinate-attributed motives behind abusive su-

FIGURE 1Theoretical Model

Shaded boxes = Team-level constructs evaluated by team leadersWhite boxes = Individual-level constructs evaluated by team members, except for teammember creativity, which was evaluated by team leaders.

Department Leader Abusive Supervision

(Time 1)

Team Leader Abusive Supervision

(Time 2)

Team Member Creativity

(Time 3)

Team Leader–Attributed

Performance Promotion Motives

(Time 1)

Team Leader–Attributed

Injury Initiation Motives

(Time 1)

Team Member–Attributed

Injury Initiation Motives

(Time 2)

H3a: + H5a: +

H3b: – H5b: –

H4a: – H6a: –

H4b: + H6b: +

H2

H1: –

Team Member–AttributedPerformance Promotion Motives

(Time 2)


pervision using two separate samples. First, to gen-erate items, we administered a questionnaire sur-vey to 128 executive MBA and MBA students in alarge northwest research-oriented university in theUnited States. At the beginning of the survey, thestudents read the definition of abusive supervision(Tepper, 2000) and were then asked to draw fromtheir work experience and describe the possiblemotives stimulating their supervisors to supervisethem abusively. Ninety-three students returned ourquestionnaires, providing 189 statements (e.g.,“They desire to stimulate me to meet my perfor-mance goals”; “They desire to make me feel badabout myself”).

Second, three leadership researchers, who wereoutside our research team and not aware of theresearch purpose, were invited to independentlysort the 189 statements into categories. They sortedthe statements into two categories, and their sort-ings showed high consistency (average r � .90).Through extensive discussion, they resolved theirfew disagreements on the statement sorting andreached a final consensus. Then, to increase vari-ance in the three leadership researchers’ ratingsand thus allow us to select a manageable number ofrepresentative items from the 189 statements, weasked the three outside researchers to rate the rep-resentativeness of the statements under each cate-gory on this scale: 1, “not representative at all”; 6,“neutral”; 11, “fully representative.” We retained20 statements with the highest average ratings ineach category; the dropped items largely repre-sented different ways to frame the same contentcovered in the retained statements. We then askedthe three leadership researchers to explain themeanings of the two categories in terms of the se-lected 40 statements. Their explanations were con-sistent with Tepper’s (2007) categorization of thecausal motives for abusive behaviors: performancepromotion and injury initiation.

Third, after refining the 40 statements to improvetheir clarity and accuracy, we explored their factorstructure by surveying 298 employees in a bank inthe northwestern United States. The employeeswere requested to indicate the extent to which eachof the 40 items represented possible causal motivesfor their leaders’ abusive behaviors toward them (1� “not representative at all,” 4 � “neutral,” 7 �“fully representative”). We received 237 responses.Using exploratory factor analyses, we eliminatedthe items with low loadings and high cross-load-ings. A two-factor structure with five items for eachfactor emerged and explained 73 percent of thetotal variance (see Appendix A for the items). Inline with our conceptualization, the five items forfactor 1 represent performance promotion motives

and have factor loadings larger than .75 (� � .85).Factor 2 indicates injury initiation motives andconsists of five items with factor loadings largerthan .77 (� �.88).

To provide further evidence supporting our de-veloped measures, we then performed a series ofconfirmatory factor analyses (CFAs) using LISREL8.7. An excellent fit was found for the two-factormodel (�2 � 85.36, df � 34, NNFI � .93, CFI � .92,RMSEA � .06), with all items loading strongly ontheir expected factors. Additionally, the two-factormodel fit the data significantly better than a one-factor model in which all items were loaded on asingle factor (�2 � 155.78, df � 35, NNFI � .72, CFI� .75, RMSEA � .18; ��2 � 70.42, �df � 1, p �.001). The above scale development proceduresand empirical results support our conceptualiza-tion of subordinate-attributed causal motives forabusive supervision as involving two distinct com-ponents: performance promotion motives and in-jury initiation motives.

Data Collection for Hypothesis Testing

To test our hypotheses, we collected multilevel,multiphase, and multisource data in a large U.S.midwest automobile parts manufacturing com-pany. The company has traditionally maintainedan organizational structure with teams responsiblefor manufacturing similar automobile parts nestedin the same department. Within each team, mem-bers manufactured the same type of automobileparts (e.g., automobile bearings) and offered tech-nical support to one another (e.g., helping cowork-ers resolve their technical problems and checkingthe quality of automobile parts made by cowork-ers). Team leaders, who reported to departmentleaders, were in charge of monitoring and evaluat-ing the work of members. Yet department leadersmight directly provide technical advice and perfor-mance feedback to team members. Team memberswere encouraged to find creative ways to enhancetheir production processes (e.g., proposing usefulideas about how to enhance coordination withteammates and resolve conflict) as well as the qual-ity of manufactured automobile parts (e.g., devel-oping new software for more accurately identifyingdeficient products) and solve work-related prob-lems (e.g., designing a website for team members toprovide immediate constructive feedback to eachother). To solicit participation, we gave multiplepresentations on the purpose and benefits of ourresearch project to the company’s top managementteam and human resource management depart-ment. With strong support from the surveyed com-pany’s management, we were able to conduct a


three-phase data collection with the intent of as-sessing the causal relationships specified inour model.

In phase 1, questionnaires were sent to 153 teamleaders under 25 departments through the company’sinternal mail system. On the questionnaire, we askedteam leaders to evaluate department leader abusivesupervision as well as the performance promotionand injury initiation motives to which they attributedit. We received 138 responses from team leaders in 22departments, thus obtaining response rates of 91 per-cent at the team level and 88 percent at the depart-ment level. One month later, we engaged in phase 2data collection. Questionnaires were administered toall 1,392 team members supervised by the 138 teamleaders who responded to our phase 1 data collection.On the questionnaire, we requested team members torate team leader abusive supervision as well as theperformance promotion and injury initiation motivesto which they attributed team leader abusive super-vision. Of the team members, 915 responded to oursurvey, a response rate of 66 percent. In phase 3,which took place one month after phase 2, we admin-istered questionnaires to 138 team leaders to collecttheir evaluations of the 915 responding team mem-bers’ creativity, again through the company’s internalmail system. The 762 evaluations we obtained from108 team leaders constituted response rates of 83percent at the team member level and 78 percent atthe team level. From the surveyed company’s produc-tion control department, we also collected data ondepartment leader–team leader relationship length(number of days), team leader-member relationshiplength (number of days) and team member objectivetask performance (number of automobile parts pro-duced in the month prior to our survey).

In sum, our final sample was composed of 762team members from 108 teams under 22 depart-ments. On average, a department had 4.81 teams(s.d. � 1.22), and a team had 7.06 members (s.d. �5.11). The average department leader–team leaderrelationship length was 138.25 days (s.d. � 112.11);the average team leader–member relationshiplength was 104.93 days (s.d. � 100.47); and mem-bers’ average age was 31.53 years (s.d. � 3.87).Among team members, 319 were female and ac-counted for 41 percent of the sample. We did notfind any significant difference between the finalsample (n � 762) and target sample (n � 1,392) interms of demographic characteristics.

Measures

Abusive supervision was rated on the scale 1,“never,” to 5, “very often.” All other measures were

rated either 1, “strongly disagree,” to 7, “stronglyagree” or 1, “never,” to 7, “always.”

Department leader and team leader abusivesupervision. Using a 15-item scale developed byTepper (2000), we asked team leaders to evaluatethe abusive behaviors of their department leadersand asked team members to assess the abusivenessof their team leaders (� � .93, team leader abusivesupervision; � � .90, department leader abusivesupervision).

Team leader– and member–attributed perfor-mance promotion motives. The five-item scale de-veloped for this study was used to measure teamleaders’ and team members’ attributions of perfor-mance promotion motives as behind their supervi-sors’ abusive behaviors (� � .77, team leader–at-tributed performance promotion motives; � � .80,team member–attributed performance promotionmotives).

Team leader– and member–attributed injuryinitiation motives. Again, we measured team lead-ers’ and team members’ attributions of injury initi-ation motives as behind their supervisors’ abusivebehaviors using the five-item scale that we devel-oped for this study ( � � .76, team leader–attributedinjury initiation motives; � � .81, team member–attributed injury initiation motives).

Team member creativity. In accordance withprior creativity studies (e.g., Zhou, 2003; Zhou &George, 2001), team member creativity was as-sessed by team leaders via a 13-item scale devel-oped by Zhou and George (2001) (� � .89).

Control variables. Following prior attribution,creativity, leadership and abusive supervision re-search (e.g., Amabile, 1996; Erdogan & Liden, 2002;Martinko et al., 2007a; Tepper, 2007), we includedcontrol variables at three levels to rule out expla-nations representing alternatives to our model. Atthe individual level (level 1), we controlled forteam member’s age, gender, education, length ofrelationship with team leader, and objective taskperformance. At the team level (level 2), we con-trolled for team size (i.e., number of members) anddepartment leader–team leader relationship length.Number of teams and number of employees withina department were controlled for at the departmentlevel (level 3).

Roberts, Walton, and Viechtbauer (2006) re-ported that age is negatively associated with neu-roticism, or dispositional tendency to experiencenegative emotions. Researchers have shown nega-tive emotions exert a significant effect on creativity(Feist, 1999). Task performance is an effective in-dicator of an individual’s skills in his or her job,which is, according to Amabile’s (1983) compo-nential conceptualization of creativity, a funda-


mental contributor to the individual’s creativity.In addition, subordinates’ age, gender, education,and task performance were significantly associ-ated with supervisors’ liking for them (Tsui &O’Reilly, 1989). The victim-precipitation litera-ture underscores the fact that leaders are prone tomistreat employees who are not likeable (Elias,1986). More recently, Tepper, Moss, and Duffy(2011) verified that those with lower task perfor-mance are more likely to be targets of abusivesupervision. Given that leader-subordinate rela-tionship length, team size, number of teams, anddepartment size may affect the quality of thesocial exchanges between leaders and followers(Cogliser & Schriesheim, 2000; Erdogan & Liden,2002; Maslyn & Uhl-Bien, 2001) and thus theextent to which an employee is exposed to abu-sive behaviors of team and department leaders,we controlled for these variables as well.

Analytical Strategy

We first conducted CFAs to confirm the dimen-sionality and the discriminant validity of ourmulti-item measures. Next, to partition the vari-ance at the individual (team member), team, anddepartment levels in hypothesis testing, we calcu-lated hierarchical linear models with HLM 6.08(Raudenbush, Bryk, Cheong, & Congdon, 2004) totest our hypotheses. These HLM3 models couldestimate the individual-level effects (level 1) aswell as the separate effects of team-level and de-partment-level predictors (levels 2 and 3) on theintercepts and slopes at the individual level. Toseparate the cross-level from between-group inter-action (Hofmann & Gavin, 1998), we used thegroup-mean-centering technique when testing thecross-level interactive effects of team leader abu-sive supervision and team leader attributions fordepartment leader abusive supervision on teammember creativity (the second-stage moderation ef-fects in Hypotheses 5a and 5b) as well as the cross-level interactive effects of department leader abu-sive supervision and team member attributions forteam leader abusive supervision on team leaderabusive supervision (the first-stage moderation ef-fects in Hypotheses 6a and 6b). We also added thegroup means back at the higher level to properlycontrol for the main effects of the lower-level fac-tors when assessing these cross-level interactioneffects (Hofmann & Gavin, 1998). For the rest of theanalyses, we applied grand mean centering to re-duce potential collinearity between level 2 inter-cept and slope terms and to model the potentialinfluences of both within- and between-team vari-

ances (Hofmann & Gavin, 1998; Mathieu & Tay-lor, 2007).

We tested our moderated mediation Hypothe-ses 5a, 5b, 6a, and 6b taking the moderated pathanalysis approach (Edwards & Lambert, 2007),which integrates moderated regression proce-dures into the path analytic method for testingmediation. In testing moderated mediation, themoderated path analysis approach (Edwards &Lambert, 2007) can address the shortcomings ofBaron and Kenny’s (1986) moderated causal stepsapproach and more clearly delineate the moder-ated and mediated nature of the relationshipsamong variables.

RESULTS

The variables’ descriptive statistics, reliabilities,and correlations, as well as data sources and col-lection schedule, are displayed in Table 1. Thereliabilities of our variables are above .75, and theircorrelations are as expected.

Preliminary Analyses

The CFA results demonstrate that our hypothe-sized four-factor model (i.e., abusive supervision,performance promotion motives, injury initiationmotives, and creativity) was a better fit to the data(�2

(659) � 1,850.05, RMSEA � .07, CFI � .92, NNFI� .91) than these more parsimonious models: athree-factor one collapsing performance promotionmotives and injury initiation motives (��2

(3) �1,123.36, p � .001, RMSEA � .12, CFI � .81, NNFI� .80); a two-factor one with all independent vari-ables loaded on one factor (��2

(5) � 1,327.55, p �.001, RMSEA � .14, CFI � .78, NNFI � .76); and aone-factor model with all variables loaded on asingle factor (��2

(6) � 1,823.93, p �. 001, RMSEA �.18, CFI � .73, NNFI � .70). To justify that HLM3was appropriate for analyzing our three-level data,we first ran null models with no predictors butcreativity as the dependent variable (Raudenbushet al., 2004). The results show significant between-team variance (�2

(86) � 215.68, p � .001; ICC1 � .28[indicating 28 percent of variance resides betweenteams]) and between-department variance (�2

(21) �112.43, p � .001; ICC1 � .18 [indicating 18 percentof variance resides between departments]) in cre-ativity. The above results substantiate that HLM3could be applied to test our multilevel hypotheses.

The Main and Mediating Effects of Team LeaderAbusive Supervision

Hypothesis 1 proposes a negative relationshipbetween team leader abusive supervision (level 1)


TA

BL

E1

Ind

ivid

ual

-Lev

elD

escr

ipti

veS

tati

stic

s,R

elia

bili

ties

,an

dC

orre

lati

ons

amon

gM

easu

res,

and

Dat

aS

ourc

esan

dC

olle

ctio

nS

ched

ule

a

Var

iabl

esM

ean

s.d

.1

23

45

67

89

1011

1213

1415

16

1.A

ge,t

ime

131

.53

3.87

2.G

end

er,t

ime

11.

410.

39.0

23.

Ed

uca

tion

,tim

e1

4.07

1.01

�.1

0**

.05

4.T

eam

lead

er-m

embe

rre

lati

onsh

iple

ngt

h,t

ime

110

4.93

100.

47.0

7�

.05

�.0

55.

Obj

ecti

vep

erfo

rman

ce,t

ime

142

.60

13.1

2�

.01

�.0

2.0

2.0

46.

Tea

msi

ze,t

ime

17.

065.

11�

.07*

.00

�.0

5.0

6�

.01

7.D

epar

tmen

tle

ader

–tea

mle

ader

rela

tion

ship

len

gth

,tim

e1

138.

2511

2.11

.01

�.0

3�

.02

.00

.05

.06

8.N

um

ber

ofte

ams

ina

dep

artm

ent,

tim

e1

4.91

1.22

�.0

4�

.02

.06

.03

�.0

1�

.36*

*.0

19.

Nu

mbe

rof

emp

loye

esin

ad

epar

tmen

t,ti

me

134

.63

21.0

9.0

9*�

.06

�.0

5.0

6.0

0.7

7**

.04

.03

10.

Dep

artm

ent

lead

erab

usi

vesu

per

visi

on,t

ime

12.

370.

87.0

0.0

4�

.00

�.0

8*�

.07

.06

.06

�.1

2**

�.0

3(.9

0)11

.T

eam

lead

er–a

ttri

bute

dp

erfo

rman

cep

rom

otio

nm

otiv

es,t

ime

14.

231.

61.0

4�

.05

�.0

8*�

.11*

*�

.03

.09*

.01

.04

.17*

*.0

5(.7

7)

12.

Tea

mle

ader

–att

ribu

ted

inju

ryin

itia

tion

mot

ives

,tim

e1

4.03

0.96

�.0

1.0

2�

.06

.10*

*.0

0�

.04

�.0

0�

.09*

�.0

6�

.08*

�.1

5**

(.76)

13.

Tea

mle

ader

abu

sive

sup

ervi

sion

,tim

e2

2.20

1.09

.06

�.0

4.0

6�

.01

.03

.08*

�.0

2.0

5.0

2.1

3**

.09*

�.0

8*(.9

3)14

.T

eam

mem

ber–

attr

ibu

ted

per

form

ance

pro

mot

ion

mot

ives

,tim

e2

3.86

0.91

.10*

*�

.02

�.0

0.1

1**

.08*

.08*

.01

.00

.08

.01

.01

�.0

7.0

9*(.8

0)

15.

Tea

mm

embe

r–at

trib

ute

din

jury

init

iati

onm

otiv

es,t

ime

24.

241.

01.0

8*�

.01

�.0

1.0

6.1

2**

.02

.00

�.0

7.0

6�

.09*

.06

.08*

�.1

0*�

.15*

*(.8

1)

16.

Cre

ativ

ity,

tim

e3

4.20

1.21

�.0

7*�

.04

�.0

4�

.01

.03

.04

�.0

1.0

8*.0

3�

.09*

.03

�.0

6�

.12*

*.0

3�

.08*

(.89)

an

�76

2at

the

ind

ivid

ual

leve

l,n

�10

8at

the

team

leve

l,an

dn

�22

atth

ed

epar

tmen

tle

vel;

reli

abil

itie

sof

the

scal

esar

ebo

ldfa

ced

and

not

edin

the

dia

gon

als.

Dat

aon

vari

able

s1–

3an

d13

–15

wer

ere

por

ted

byin

div

idu

alte

amm

embe

rs;d

ata

onva

riab

les

4–9

wer

eco

llec

ted

from

the

com

pan

y’s

man

ufa

ctu

rin

gco

ntr

old

epar

tmen

tor

hu

man

reso

urc

ed

epar

tmen

t;va

riab

les

10–1

2an

d16

wer

eev

alu

ated

byte

amle

ader

s.*

p�

.05

**p

�.0

1T

wo-

tail

edte

sts.

and team member creativity (level 1). As shown bythe results of model 7 in Table 2, team leader abu-sive supervision was negatively related to teammember creativity (� � �.53, p � .01). Thus, Hy-pothesis 1 is supported. Hypothesis 2 suggests teamleader abusive supervision (level 1) mediates thecross-level relationship between department leaderabusive supervision (level 2) and team member cre-ativity (level 1). We used the PRODCLIN program(MacKinnon, Fritz, Williams, & Lockwood, 2007) toconduct a product of coefficients test for this hy-pothesis; these results indicate that the indirecteffect of department leader abusive supervision onteam member creativity via team leader abusivesupervision is significant (95% confidence interval[CI] � �.25, �.02 [not containing 0]). Therefore,Hypothesis 2 receives support.

The Contingent Effects of Team LeaderAttributions

With respect to Hypothesis 3a, the interactiveeffect of department leader abusive supervision(level 2) and team leader–attributed performancepromotion motives (level 2) on team leader abusivesupervision (level 2) was also significant (� � .12, p� .05). Figure 2 and slope tests show that with highteam leader–attributed performance promotion mo-tives (1 s.d. above the mean), department leaderabusive supervision was more positively related toteam leader abusive supervision (� � .30, p � .05)than with low team leader–attributed performancepromotion motives (1 s.d. below the mean; � � .15,p � .05). Thus, Hypothesis 3a is supported.

Regarding Hypothesis 3b, the results of model 3in Table 2 reveal that the interactive effect of de-partment leader abusive supervision (level 2) andteam leader–attributed injury initiation motives(level 2) on team leader abusive supervision (level2) was significant (� � �.10, p � .01). Slope tests,as shown in Figure 3, demonstrate that when teamleader–attributed injury initiation motives werehigh (1 s.d. above the mean), department leaderabusive supervision was less positively related toteam leader abusive supervision (� � .13, p � .05)than when team leader–attributed injury initiationmotives were low (1 s.d. below the mean; � � .31,p � .01). Therefore, Hypothesis 3b is verified.

The Contingent Effects of Team MemberAttributions

Per Hypothesis 4a, the results of model 8 in Table2 indicate that the interaction between team leaderabusive supervision (level 1) and team member–attributed performance promotion motives (level 1)

was positively related to team member creativity(level 1) (� � .15, p � .05). Figure 4 and slope testsreveal that when team member–attributed perfor-mance promotion motives were high (1 s.d. abovethe mean), team leader abusive supervision wasless negatively related to team member creativity (�� .37, p � .01) than when team member–attrib-uted performance promotion motives were low(1 s.d. below the mean; � � �.66, p � .01). Hence,Hypothesis 4a receives support.

With regard to Hypothesis 4b, as shown by theresults of model 8 in Table 2, the interactive effectof team leader abusive supervision (level 1) andteam member–attributed injury initiation motives(level 1) on team member creativity (level 1) wassignificant (� � �.20, p � .01). Figure 5 and slopetests demonstrate that with high team member–attributed injury initiation motives (1 s.d. above themean), team leader abusive supervision was morenegatively related to team member creativity (� ��.72, p � .01) than with low team member–attrib-uted injury initiation motives (1 s.d. below themean; � � �.31, p � .01). Thus, Hypothesis 4b isverified.

An Integrative Moderated Mediation Model

To test moderated mediation Hypotheses 5a, 5b,6a, and 6b, we applied Edwards and Lambert’s(2007) moderated path analysis approach to esti-mate four sets of effects at the high and low levelsof the moderating variables (i.e., team leader attri-butions and team member attributions): a first-stageeffect (i.e., the effect of department leader abusivesupervision on team leader abusive supervision);second-stage effect (i.e., the effect of team leaderabusive supervision on team member creativity);direct effect (i.e., the effect of department leaderabusive supervision on team member creativity);and overall indirect effect (i.e., the effect of depart-ment leader abusive supervision on team membercreativity through team leader abusivesupervision).

Hypothesis 5a predicts that team leader–attrib-uted performance promotion motives moderate theindirect negative effect of department leader abu-sive supervision on team member creativitythrough team leader abusive supervision. The re-sults reported in Table 3 suggest that the indirecteffect of department leader abusive supervision onteam member creativity via team leader abusivesupervision was significantly moderated by teamleader–attributed performance promotion motives(�� .08, p � .05). Specifically, as shown byFigure 6, the indirect effect was stronger with highteam leader–attributed performance promotion


TA

BL

E2

HL

MR

esu

lts:

Th

eM

ain

and

Inte

ract

ive

Eff

ects

ofA

busi

veS

up

ervi

sion

and

Att

ribu

ted

Mot

ives

onT

eam

Mem

ber

Cre

ativ

itya

Var

iabl

es

Tea

mL

ead

erA

busi

veS

up

ervi

sion

:L

evel

1,T

ime

2T

eam

Mem

ber

Cre

ativ

ity:

Lev

el1,

Tim

e3

Mod

el1

Mod

el2

Mod

el3

Mod

el4b

Mod

el5b

Mod

el6

Mod

el7

Mod

el8

Mod

el9b

Mod

el10

b

1In

terc

ept

2.36

**2.

41**

2.33

**2.

39**

2.37

**4.

13**

4.09

**4.

15**

4.14

**4.

15**

2Le

vel

1co

ntr

olT

eam

mem

ber

age

.04

.02

.08

.08

.08

�.1

3**

�.1

3*�

.09*

�.1

0*�

.09*

Tea

mm

embe

rge

nd

er�

.01

.01

�.0

1�

.02

�.0

1�

.00

�.0

1.0

1�

.00

.00

Tea

mm

embe

red

uca

tion

.06*

.06*

.06

.06

.06

�.0

1.0

1.0

1�

.00

.01

Tea

mle

ader

-mem

ber

rela

tion

ship

len

gth

�.0

0.0

2.0

3.0

3.0

3�

.09

�.0

5�

.05

�.0

5�

.07

Tea

mm

embe

rob

ject

ive

per

form

ance

.05

.04

�.0

1�

.01

�.0

1�

.01

�.0

3.0

5.0

2.0

43

Leve

l1

ind

epen

den

tT

eam

lead

erab

usi

vesu

per

visi

on�

.53*

*�

.51*

*�

.50*

*�

.53*

*T

eam

mem

ber–

attr

ibu

ted

per

form

ance

pro

mot

ion

mot

ives

.04

.05

Tea

mm

embe

r–at

trib

ute

din

jury

init

iati

onm

otiv

es�

.32*

*�

.32*

*

Tea

mle

ader

abu

sive

sup

ervi

sion

�te

amm

embe

r–at

trib

ute

dp

erfo

rman

cep

rom

otio

nm

otiv

es.1

5*.1

3*

Tea

mle

ader

abu

sive

sup

ervi

sion

�te

amm

embe

r–at

trib

ute

din

jury

init

iati

onm

otiv

es�

.20*

*�

.19*

*

4Le

vel

2co

ntr

olan

din

dep

end

ent

Tea

msi

ze.0

8.0

9.1

7*.1

2*.1

4*.0

4.0

8.1

3*.1

2*.1

3*D

epar

tmen

tle

ader

–tea

mle

ader

rela

tion

ship

len

gth

�.0

2�

.03

�.0

1�

.01

�.0

2�

.03

�.0

4�

.02

�.0

3�

.03

Dep

artm

ent

lead

erab

usi

vesu

per

visi

on.2

2*.2

4*.2

5*.2

3*�

.24*

�.1

8�

.07

�.0

6�

.07

Tea

mle

ader

-att

ribu

ted

per

form

ance

pro

mot

ion

mot

ives

.01

.07

.05

.07

.06

Tea

mle

ader

–att

ribu

ted

inju

ryin

itia

tion

mot

ives

�.1

7**

�.2

0**

�.0

6�

.06

�.0

6D

epar

tmen

tle

ader

abu

sive

sup

ervi

sion

�te

amle

ader

-att

ribu

ted

per

form

ance

pro

mot

ion

mot

ives

.12*

.12*

.07

.08

.07

Dep

artm

ent

lead

erab

usi

vesu

per

visi

on�

team

lead

er–a

ttri

bute

din

jury

init

iati

onm

otiv

es�

.10*

*�

.13*

*�

.05

�.0

5�

.05

5Le

vel

3co

ntr

olN

um

ber

ofte

ams

ina

dep

artm

ent

.14

.09

.09

.12

.06

.39*

.34

.36*

.37*

.37*

Nu

mbe

rof

emp

loye

esin

ad

epar

tmen

t.0

4.0

3�

.01

.01

.01

�.0

2.0

3.0

5.0

5.0

6R

2c

.07*

.13*

.22*

.17*

.18*

.09*

.14*

.25*

.20*

.22*

an

�76

2at

the

ind

ivid

ual

leve

l,n

�10

8at

the

team

leve

l,an

dn

�22

atth

ed

epar

tmen

tle

vel.

To

faci

lita

teco

mp

aris

ons

ofth

em

agn

itu

des

ofef

fect

sat

each

leve

lst

emm

ing

from

dif

fere

ntl

ysc

aled

vari

able

s,w

een

tere

dst

and

ard

ized

pre

dic

tors

toth

ees

tim

ated

mod

els.

bM

odel

s4–

5an

d9–

10ar

esu

pp

lem

enta

ryan

alys

esto

dem

onst

rate

that

the

det

ecte

dm

oder

atin

gef

fect

sar

en

otca

use

dby

the

mu

ltic

olli

nea

rity

amon

gth

etw

oty

pes

ofm

otiv

es.

cR

2is

calc

ula

ted

base

don

pro

por

tion

alre

du

ctio

nof

erro

rva

rian

ce,

du

eto

pre

dic

tors

inth

em

odel

sof

Tab

le2

(Sn

ijd

ers

&B

oske

r,19

99).

*p

�.0

5**

p�

.01

Tw

o-ta

iled

test

s.

motives (� � �.15, p � .05) than with low teamleader–attributed performance promotion motives(� � �.07, p � .05). Additionally, Table 3 showsthat team leader–attributed performance promotionmotives moderated the indirect effect of depart-ment leader abusive supervision on team membercreativity via team leader abusive supervision ow-ing to its moderating effect on the relationship be-tween department leader abusive supervision andteam leader abusive supervision (i.e., the first-stageeffect; �� .15, p � .05). Team leader–attributedperformance promotion did not moderate the rela-tionship between team leader abusive supervisionand team member creativity (i.e., the second-stage

effect; �� .04, n.s.). Consequently, Hypothesis 5ais supported.

Hypothesis 5b predicts that team leader–attrib-uted injury initiation motives moderate the indirectnegative effect of department leader abusive super-vision on team member creativity through teamleader abusive supervision. The results of Table 3reveal a significant overall moderating effect ofteam leader–attributed injury initiation motives onthe indirect relationship of department leader abu-sive supervision with team member creativity viateam leader abusive supervision (�� .06, p �.05). In full support of Hypothesis 5b, Figure 7suggests the indirect effect was stronger with low

FIGURE 2The Interactive Effect of Department Leader Abusive Supervision and Team Leader–Attributed

Performance Promotion Motives on Team Leader Abusive Supervision

1.70

1.80

1.90

2.00

2.10

2.20

2.30

2.40

2.50

2.60

Low DepartmentLeader Abusive

Supervision

High DepartmentLeader Abusive

Supervision

Team LeaderAbusive

Supervision

Low team leader–attributed performancepromotion motives

High team leader–attributed performancepromotion motives

FIGURE 3The Interactive Effect of Department Leader Abusive Supervision and Team Leader–Attributed Injury

Initiation Motives on Team Leader Abusive Supervision

1.80

1.90

2.00

2.10

2.20

2.30

2.40

2.50

2.60

2.70


Supervision


Supervision

Team LeaderAbusive

Supervision

Low team leader–attributed injuryinitiation motives

High team leader–attributed injuryinitiation motives


team leader–attributed injury initiation motives (�� .10, p � .05) than with high team leader–attributed injury initiation motives (� � �.04, p �.05). Furthermore, the results of Table 3 show thatteam leader–attributed injury initiation motivesmoderated the indirect effect of department leaderabusive supervision on team member creativity,because of its moderating role in the first-stage ef-fect (�� .18, p � .01), namely, the relationshipbetween department leader abusive supervisionand team leader abusive supervision. In addition,team leader–attributed injury initiation motivesdid not moderate the second-stage effect, or therelationship between team leader abusive supervi-

sion and team member creativity (�� .01, n.s.).Therefore, Hypothesis 5b receives support.

The results of Table 4 support Hypothesis 6a; theindirect negative effect of department leader abu-sive supervision on team member creativity viateam leader abusive supervision was moderated byteam member-attributed performance promotionmotives (�� .09, p � .01). Specifically, as indi-cated by Figure 8, the indirect negative effect wasstronger in the presence of low team member-attrib-uted performance promotion motives (� � �.20,p � .01) than in the presence of high team member-attributed performance promotion motives (� ��.11, p � .01). The results of Table 4 further show

FIGURE 4The Interactive Effect of Team Leader Abusive Supervision and Team Member–Attributed Performance

Promotion Motives on Team Member Creativity

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

5.00

Low Team LeaderAbusive Supervision

High Team LeaderAbusive Supervision

TeamMember

Creativity

Low team member–attributed performancepromotion motives

High team member–attributed performancepromotion motives

FIGURE 5The Interactive Effect of Team Leader Abusive Supervision and Team Member–Attributed Injury Initiation

Motives on Team Member Creativity

3.00

3.20

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

5.00

Low Team LeaderAbusive Supervision

High Team LeaderAbusive Supervision

TeamMember

Creativity

Low team member–attributed injuryinitiation motives

High team member–attributed injuryinitiation motives


that the overall significant moderating effect ofteam member–attributed performance promotionmotives on the indirect effect was caused by itssignificant moderating role in the second-stage ef-fect (�� .29, p � .05), namely, the relationshipbetween team leader abusive supervision and teammember creativity. In addition, team member–at-tributed performance promotion motives did notmoderate the first-stage effect (�� .01, n.s.),namely, the relationship between departmentleader abusive supervision and team leader abusivesupervision. Hence, Hypothesis 6a is verified.

The results reported in Table 4 also substantiateHypothesis 6b; team member–attributed injury ini-tiation motives moderated the indirect effect of de-partment leader abusive supervision on team mem-ber creativity via team leader abusive supervision(�� .10, p � .01). Figure 9 suggests that theindirect negative effect was stronger with high teammember–attributed injury initiation motives (� ��.19, p � .01) than with low team member–attrib-uted injury initiation motives (� � �.09, p � .01).The overall significant moderating effect of teammember–attributed injury initiation resulted from

TABLE 3Results of the Moderated Path Analysisa

Moderator Variables

Department Leader Abusive Supervision (X1) ¡

Team Leader Abusive Supervision (M) ¡ Creativity (Y)

First StagePMX1

Second StagePYM

Direct EffectPYX1

Indirect EffectPMX1 � PYM

Team leader–attributed performance promotion motivesb

Low (�1 s.d.) .15* �.46** �.06 �.07*High (�1 s.d.) .30* �.50** �.03 �.15*Differences between low and high �.15* .04 �.03 .08*

First StagePMX2

Second StagePYM

Direct EffectPYX2


Team leader–attributed injury initiation motivesb

Low (�1 s.d.) .31** �.33** �.01 �.10*High (�1 s.d.) .13* �.32** �.03 �.04*Differences between low and high .18** �.01 .02 �.06*

a n � 762 at the individual level, n � 108 at the team level, and n � 22 at the department level.b “Low” moderator variable refers to one standard deviation below the mean of the moderator; “high” moderator variable refers to one

standard deviation above the mean of the moderator.* p � .05

** p � .01Two-tailed tests.

FIGURE 6The Moderating Effect of Team Leader–Attributed Performance Promotion Motives on the Indirect

Relationship of Department Leader Abusive Supervision with Team Member Creativity throughTeam Leader Abusive Supervision

3.70

3.80

3.90

4.00

4.10

4.20

4.30

4.40

4.50

4.60


Supervision


Supervision

Low team leader–attributed performancepromotion motives

High team leader–attributed performancepromotion motives

TeamMember

Creativity


its significant moderating role in the second-stageeffect that is the relationship between team leaderabusive supervision and team member creativity(�� .41, p � .01). Additionally, the first-stageeffect (i.e., the relationship between departmentleader abusive supervision and team leader abusivesupervision) was not moderated by team member–attributed injury initiation motives (�� .02, n.s.).Thus, Hypothesis 6b receives support.

Supplementary Analysis

In this research, we examine how followers’attributions of leader abusive supervision mo-tives may moderate the trickle-down process ofabusive supervision and the ultimate effect onemployee creativity. One might argue that foremployees who did not experience any abusivesupervision, it would be irrelevant to examine

FIGURE 7The Moderating Effect of Team Leader–Attributed Injury Initiation Motives on the Indirect Relationship

of Department Leader Abusive Supervision with Team Member Creativity through Team LeaderAbusive Supervision

3.90

4.00

4.10

4.20

4.30

4.40

4.50


Supervision


Supervision

TeamMember

Creativity

Low team leader–attributed injuryinitiation motives

High team leader–attributed injuryinitiation motives

TABLE 4Results of the Moderated Path Analysisa

Moderator Variables

Department Leader Abusive Supervision(X1) ¡Team Leader Abusive Supervision

(M) ¡ Creativity (Y)

First StagePMX1

Second StagePYM

Direct EffectPYX1


Team member–attributed performance promotion motivesb

Low (�1 s.d.) .30** �.66** �.09 �.20**High (�1 s.d.) .31** �.37** �.08 �.11**Differences between low and high �.01 �.29* �.01 �.09**

First StagePMX2

Second StagePYM

Direct EffectPYX2


Team member–-attributed injury initiation motivesb

Low (�1 s.d.) .29** �.31** �.06 �.09**High (�1 s.d.) .27** �.72** �.07 �.19**Differences between low and high .02 .41** .01 .10**

a n � 762 at the individual level, n � 108 at the team level, and n � 22 at the department level.b “Low” moderator variable refers to one standard deviation below the mean of the moderator; “high” moderator variable refers to one

standard deviation above the mean of the moderator.* p � .05

** p � .01Two-tailed tests.


their attributions of abusive supervision. In oursample, on the scale 1, “never,” to 5, “very often,”the mean of department leader abusive supervi-sion was 2.37, and the mean of team leader abu-sive supervision was 2.20. Therefore, on average,the respondents in our sample experienced somelevel of abusive supervision. However, 2 teamleaders and 10 team members indicated 1(“never”) for all of the abusive supervision items,thus reporting no abusive supervision. Although

this represented a very small portion of our sam-ple of 108 team leaders and 762 team members, toalleviate concern as to how these respondentsmight have influenced our results, we reper-formed all the analyses excluding them andfound that the statistical significance of the hy-pothesis tests remained unchanged and that themagnitude of the coefficient estimates remainedessentially the same. Considering these resultsand the fact that examining the trickle-down pro-

FIGURE 8The Moderating Effect of Team Member–Attributed Performance Promotion Motives on the Indirect

Relationship of Department Leader Abusive Supervision with Team Member Creativity throughTeam Leader Abusive Supervision

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

5.00

Low Department Leader Abusive

Supervision

High Department Leader Abusive

Supervision

TeamMember

Creativity

Low team member–attributed performancepromotion motives

High team member–attributed performancepromotion motives

FIGURE 9The Moderating Effect of Team Member–Attributed Injury Initiation Motives on the Indirect Relationship

of Department Leader Abusive Supervision with Team Member Creativity throughTeam Leader Abusive Supervision

3.20

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

5.00

Low Department Leader Abusive

Supervision

High Department Leader Abusive

Supervision

TeamMember

Creativity

Low team member–attributed injuryinitiation motives

High team member–attributed injuryinitiation motives


cess of abusive supervision and the effect onemployee creativity would still be relevant forthose who experienced no abusive supervision,we retained the full sample for all analyses. Re-sults of these supplementary analyses are avail-able upon request.

DISCUSSION

This research examined how and when abusivesupervision may cascade down to harm employeecreativity. In support of our conceptual analysis,we found that department leader abusive supervi-sion exerted a negative indirect effect on teammember creativity through team leader abusivesupervision and that this cascading effect was mod-erated by team leader and team member attribu-tions. Team leader–attributed performance promo-tion motives augmented the positive influence ofdepartment leader abusive supervision on teamleader abusive supervision, whereas team leader–attributed injury initiation motives weakened suchinfluence. The negative relationship between teamleader abusive supervision and team member cre-ativity was accentuated by team member–attrib-uted injury initiation motives but attenuated byteam member–attributed performance promotionmotives. The findings of this research generatesome interesting theoretical and managerialimplications.

Theoretical Implications

Our findings contribute to the leadership, cre-ativity, and attribution literatures in four primaryways. First, this research enhances understandingof the role of negative aspects of leadership in thedevelopment of employee creativity. Past researchconcerning the link between leadership and cre-ativity has exclusively concentrated on identifyingpositive leader behaviors that may facilitate subor-dinate creativity; identified behaviors includetransformational leadership (e.g., Shin & Zhou,2003) and a supportive supervisory style (Oldham& Cummings, 1996). Consequently, the influenceon creativity of negative leader behaviors such asabusive supervision, which has been found to oc-cur frequently in organizational contexts (Tepper,2000; Tepper et al., 2009 ), has generally been leftunexplored. We addressed this research gap by us-ing a temporally lagged field research design toprovide empirical evidence about the detrimentaleffect of abusive supervision on subordinate cre-ativity. Our research also responds to creativityscholars’ call for a fine-grained, multilevel ap-proach to the precursors of individual creativity

(Zhou & Shalley, 2008). Using three-level hierarchi-cal data, we demonstrate that department leaderabusive supervision and team leader abusive super-vision exerted unique, independent effects on em-ployee creativity. Hence, our research testifies tothe promise of extending creativity theory by inves-tigating socially undesirable behaviors by supervi-sors (e.g., abusive supervision) from a multilevelperspective. In addition, we add to the abusivesupervision literature by broadening the range ofindividual outcomes resulting from abusive behav-iors to follower performance consequences (e.g.,creativity). This addition is important because theextant literature has primarily examined the attitu-dinal outcomes of abusive supervision for the sub-ordinates involved, such as their job satisfactionand psychological distress (for exceptions, seeAryee et al. [2007] and Harris, Kacmar, and Ziv-nuska [2007]).

Second, whereas recent years have witnessed agrowing interest in studying abusive supervision,antecedents of abusive supervision are little stud-ied (Breaux, 2010; Tepper et al., 2011). Respondingto Tepper’s (2007) call for exploring the predictorsof abusive supervision from a multilevel perspec-tive, we investigate the trickle-down impact of de-partment leader abusive supervision on teamleader abusive supervision and, in turn, team mem-ber creativity, with data collected in three phasesbearing out the specified causal relationshipsamong the study variables. Team leader abusivesupervision was found to account for the negativeinfluence of department leader abusive supervisionon team member creativity. This finding supportssocial learning theory and sheds light on the so-cially learned nature of workplace abusive behav-iors (Bandura, 1986). That is, middle-level manag-ers are likely to adopt their supervisors as rolemodels and thus mimic top management’s abuse byabusing their own subordinates. Our results alsoreinforce the veracity of the cascading effect ofleader behaviors. Adding to the literature on thecascading effect of positive leader behaviors (e.g.,Bass et al., 1987; Mayer et al., 2009), we show thatleaders lower in an organizational hierarchy mayalso emulate negative behaviors of higher-levelleaders (e.g., abusive behaviors). This suggests thatleadership research may advance by shifting itsemphasis from positive leader behaviors to com-bined leader behaviors of different valences andstudying how they may jointly trickle down organ-izational levels to affect followers’ attitudes andbehaviors.

A third theoretical implication of this research isthat it extends the trickle-down model of leader-ship by integrating attribution theory to unveil the


contingent roles of subordinate attributions in stim-ulating or preventing the spreading of abusive su-pervision down an organizational hierarchy.Guided by Green and Mitchell’s (1979) theoreticalframework, researchers have generally looked athow leaders’ causal attributions for followers’ be-haviors may impinge on leaders’ responses to fol-lowers (e.g., Campbell & Swift, 2006; Eastman,1994; Lam, Huang, & Snape, 2007). For example,Lam and colleagues (2007) documented that super-visors respond differently to employee feedback-seeking behavior depending on whether they as-cribe the behavior to employee performance-enhancement motives or impression managementmotives. Employees whose extra-role behaviors arelabeled citizenship behaviors by supervisors re-ceive greater rewards than employees whose extra-role behaviors are interpreted as ingratiation bysupervisors (Eastman, 1994). Surprisingly, empiri-cal research that examines followers’ attributionsfor leader behaviors is still scarce. Directly address-ing this research gap, we found that followers notonly develop two causal attributions for leaders’abuse, but also that such attributions significantlyaffect the impact of abusive supervision. Our re-sults thus suggest that examining follower attribu-tions as the boundary conditions of the cascadingeffect of leader behaviors may generate crucial in-sights into the research on leadership processesand consequences.

Finally, in support of attribution theory, our testof the overall integrative moderated mediationmodel provides solid evidence that the extent towhich team leader abusive supervision mediatesthe relationship between department leader abu-sive supervision and team member creativity de-pends on team leaders’ own and followers’ attribu-tions. Past findings on cascading effects ofleadership corroborate that middle-level managersare the pivotal psychological link between topmanagement and frontline workers (e.g., Zohar &Luria, 2005). However, past research is silent as tothe conditions under which the mediating effect ofmiddle-level managers is amplified or attenuated.This may be a partial consequence of methodolog-ical defects. We employed a unified moderatedpath analysis method (Edwards & Lambert, 2007) toovercome the methodological limitations in previ-ous organizational studies and conduct a holistictest of the model. As such, our research offers valu-able insights into how to simultaneously examinecascading behavioral contagion effects as well astheir boundary conditions through conceptualizingand testing a moderated mediation model.

Practical Implications

Our research brings significant implications forpractice. We have shown that abusive supervisionby top management renders middle-level managersmore likely to display abusive behaviors and harmemployee creativity. This result ought to serve as awarning to organizations that abusive supervisionshould be avoided. An overall zero-tolerance pol-icy with regard to abusive supervision should beadopted and stated to organizational members at allhierarchical levels (Sutton, 2007). In addition, or-ganizations should implement training programs toteach management how to prevent the occurrenceof abusive supervision. Our results imply that asemployees’ immediate supervisors, middle-levelmanagers (e.g., team leaders) hold the key tospreading abusive behaviors from top managementto employees. Thus, organizations may want to in-vest more in training programs geared toward mid-dle-level managers. Furthermore, an organizationalculture of fairness, combined with leaders’ en-hanced interpersonal skills, may help cultivate awork environment that sets employees free fromabusive supervision (Aryee et al., 2007).

Another implication of our research relates di-rectly to organization members’ attributions regard-ing leaders’ abusive behaviors. We found that sub-ordinates’ judgments about why leaders abusethem either exacerbate or mitigate the effects ofabusive supervision. Accordingly, organizationsshould take into account the role of subordinateattributions about abusive supervision when theytake steps to enhance the managerial awareness ofabusive supervision’s detrimental effects and dis-courage middle managers from modeling and fur-ther spreading such behavior to mistreat their ownsubordinates. In particular, middle managersshould be aware that they are more likely to engagein abusive supervision and harm subordinate cre-ativity when they attribute abuse from higher-levelmanagers to performance promotion motives. Ad-ditionally, middle managers need to be informedthat abusive supervision will have a more detri-mental effect on their subordinates’ creativity iftheir subordinates perceive its motive as initiatinginjury rather than promoting performance. There-fore, middle managers should especially avoidmaking subordinates feel that they are subject toabusive supervision triggered by injury initiationmotives, and they should endeavor to convey theirintent to enhance subordinate performance to mit-igate the negative impact of abusive supervision onsubordinate creativity. Nevertheless, our researchindicates that even coupled with performance-en-hancing motives, abusive supervision still prevents


employee creativity. Hence, to encourage employeecreativity, organizational leaders should exercisetransformational leadership, which has beenshown to be positively associated with employeecreative performance (e.g., Shin & Zhou, 2003).

Limitations and Directions for Future Research

As with any empirical study, ours has several lim-itations that point to avenues for future research.First, we did not empirically test the possible psycho-logical mechanisms between team leader abusive su-pervision and team member creativity, because ourtheoretical model centers on the conditional, cascad-ing effect of abusive supervision from departmentleader to team leader to ultimately undermine teammember creativity. In this research, we draw on thetheoretical arguments and empirical evidence fromthe extant creativity research (e.g., Bies & Tripp, 1998;Deci et al., 1989; Hoobler & Brass, 2006) to theorizethat team leader abusive supervision will undermineteam member intrinsic motivation and subsequently,creativity. Accordingly, a valuable extension of thisresearch is to empirically test if subordinate intrinsicmotivation may indeed serve as the mediating pro-cess that links abusive supervision to subordinatecreativity.

Second, although we draw on social learning the-ory (Bandura, 1986) to explain the trickle-down effectof abusive supervision, additional theory can perhapsbe invoked to explain it. For example, from a justiceperspective, abused subordinates’ sense of injusticemay arise when they hold their supervisors account-able for their abusive behavior (Breaux, Tepper, Carr,& Folger, 2010). Consequently, they may be promptedto engage in abuse to vent their negative feelings suchas frustration and anger (Aryee et al., 2007; Tepper etal., 2006). In addition, Miller, Pederson, Earleywine,and Pollock’s (2003) displaced aggression model sug-gests that a triggering provocation (e.g., abuse) stimu-lates individuals to develop cognitive rumination andultimately aggression displacement (e.g., abusinglower-status individuals). More recently, Breaux(2010) adopted Mullen’s (2007) affective model ofjustice reasoning to explain why abusive supervisionoccurs. She verified an underlying emotional conduitthat links supervisors’ experiences of interpersonaljustice from their own superiors to their engagementin abusing followers. Thus, to further the understand-ing of the cascading effect of abusive supervision,scholars can extend our research by testing whetherrole modeling, justice perception, abuse displace-ment, and emotions (e.g., anger) may be the mediatingmechanisms in the relationship between departmentleader abusive supervision and team leader abusivesupervision.

Third, given the focus of our research on subor-dinates’ reactions to abusive supervision, we ex-plored subordinate attributions only. Yet Martinkoand Gardner (1987) contended that both leader andsubordinate attributions for subordinate successesand failures might explain variance in leader andsubordinate behaviors. Therefore, a valuable exten-sion of our research is to investigate how leader andsubordinate attributions for leader abusive supervi-sion may simultaneously affect the ways subordi-nates respond to abusive supervision and the ex-tent to which leaders engage in further abuse. Wespeculate that when leaders interpret their abusivebehaviors as driven by performance promotion mo-tives rather than injury initiation motives, they willtend to abuse subordinates more. In contrast, sub-ordinates will respond less negatively to abusivesupervision if they attribute leader abusive behav-ior to performance promotion motives instead ofinjury initiation motives.

Fourth, our research could also be extended byconsidering leader and subordinate attributionstyles regarding abusive supervision. Attributionstyle has been conceptualized as a stable individualtrait that reflects the propensity to “explain differ-ent events in the same way (i.e., make similar attri-butions) over time” (Martinko, Moss, Douglas, &Borkowski, 2007b: 159). Martinko and colleagues(2007b) demonstrated that leaders’ optimistic attri-bution styles and subordinates’ pessimistic attribu-tion styles interact to shape subordinates’ percep-tions of leader-subordinate relationship quality.Accordingly, future research may scrutinize thejoint effects of leaders’ optimistic attribution stylesand subordinates’ pessimistic attribution styles onleader and subordinate interpretations of the mo-tives behind abusive behaviors and ultimately, con-sequences of abusive supervision.

Fifth, another interesting avenue for future re-search is to explore how individual differencessuch as degree of negative affectivity may affect anemployee’s experience with abusive supervisionand creativity.1 High negative affectivity may leadto more victimization as a target of abusive super-vision as supervisors might find such employees tobe provocative, weak, and obedient (Aquino,Lewis, & Bradfield, 1999; Aryee et al., 2007). Neg-ative affectivity has also been found to be nega-tively associated with creativity (Ng & Feldman,2009). Therefore, ideally, it needs to be controlledfor to examine the unique influence of abusivesupervision on creativity. Although we do not have

1 We thank an anonymous reviewer for offering thisinsight.


negative affectivity in our current data, we con-trolled for team members’ task performance (i.e.,number of automobile parts manufactured in themonth prior to our survey, a key metric the com-pany uses to evaluate employees’ job performanceand determine pay and bonus). Prior research hasshown that individual differences such as affectiv-ity and personality affect job performance (formeta-analytical reviews, see Barrick and Mount[1991]. Hurtz and Donovan [2000], and Kaplan,Bradley, Luchman, and Haynes [2009]). Accord-ingly, we argue that to the extent negative affectiv-ity and other individual difference variables affect/manifest their influences in job performance andcreativity as well as encounters of abusive supervi-sion, controlling for job performance would at leastto some extent account for the effects of negativeaffectivity and other individual differences vari-ables. Nonetheless, we acknowledge it would beideal to measure individual differences variablesdirectly in future research.

Finally, future research may control for em-ployee job characteristics and adopt objective mea-sures of employee creativity to replicate our results.Job characteristics may not only determine the ex-tent to which employees can be creative but alsobear on the relationship between leaders and em-ployees. Additionally, following past creativity re-search, we asked team leaders to evaluate teammember creativity. Nevertheless, supervisory eval-uations may be subject to numerous contextual andpersonal biases (Landy & Farr, 1980). Hence, a re-cent trend in creativity research is to use objectivemeasures of creativity, such as creative perfor-mance bonuses, new ideas proposed, invention dis-closure forms, research reports, and patent an-nouncements (e.g., Liao, Liu, & Loi, 2010; Oldham& Cummings, 1996; Taggar, 2001; Tierney, Farmer,& Graen, 1999). Therefore, it would be worth inves-tigating in the future if subjective and objectivemeasures of creativity may generate convergent re-sults for our proposed model.

Conclusions

Our examination of the contingent roles of sub-ordinate attributions in the ways department leaderabusive supervision flows down organizational hi-erarchy to undermine employee creativity makesvaluable contributions to leadership, attribution,and creativity literatures. We call upon scholars tocontinue to look into the multilevel causation andeffectuation of abusive supervision from an attribu-tion perspective, considering the abundance of re-search opportunities in this area.

REFERENCES

Amabile, T. M. 1983. The social psychology of creativ-ity. New York: Springer-Verlag.

Amabile, T. M. 1996. Creativity in context. Boulder, CO:Westview.

Amabile, T. M., Hadley, C. N., & Kramer, S. J. 2002.Creativity under the gun. Harvard Business Review,80(8): 52–61.

Aquino, K., Lewis, M. U., & Bradfield, M. 1999. Justiceconstructs, negative affectivity, and employee devi-ance: A proposed model and empirical test. Journalof Organizational Behavior, 20: 1073–1091.

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

Bandura, A. 1973. Aggression: A social learning anal-ysis. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. 1977. Social learning theory. New York:General Learning.

Bandura, A. 1986. Social foundations of thought andaction: A social cognitive theory. Englewood Cliffs,NJ: Prentice-Hall.

Baron, R. M., & Kenny, D. A. 1986. The moderator-medi-ator variable distinction in social psychologicalresearch: Conceptual, strategic and statistical con-siderations. Journal of Personality and Social Psy-chology, 51: 1173–1182.

Barrick, M. R., & Mount, M. K. 1991. The Big Five per-sonality dimensions and job performance: A meta-analysis. Personnel Psychology, 44: 1–26.

Bass, B. M. 1990. From transactional to transformationalleadership: Learning to share the vision. Organiza-tional Dynamics, 18(3): 19–31.

Bass, B. M., Waldman, D. A., Avolio, B. J., & Webb, M.1987. Transformational leadership and the fallingdominoes effect. Group and Organization Studies,12: 73–87.

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., &Vohs, K. D. 2001. Bad is stronger than good. Reviewof General Psychology, 5: 323–370.

Bies, R. J., & Tripp, T. M. 1998. Revenge in organizations:The good, the bad, and the ugly. In R. W., Griffin, A.O’Leary-Kelly, & J. Collins (Eds.), Dysfunctional be-havior in organizations: Non-violent dysfunctionalbehavior: 49–67. Greenwich, CT: JAI Press.

Breaux, D. M. 2010. An experimental investigation ofabusive supervision as an emotional reaction toinjustice. Unpublished doctoral dissertation, theFlorida State University.

Breaux, D. M., Tepper, B. J., Carr, J. C., & Folger, R. G.2010. An attributional analysis of employees’ re-


sponses to abusive supervision. In L. L. Neider &C. A. Schriesheim (Eds.), The dark side of manage-ment: 69–92. Charlotte, NC: Information Age.

Brezina, T. 1999. Teenage violence toward parents as anadaptation to family strain: From a national surveyof male adolescents. Youth & Society, 30: 416–444.

Campbell, C. R., & Swift, C. O. 2006. Attributional com-parisons across biases and leader-member exchangestatus. Journal of Managerial Issues, 18: 393–408.

Carmeli, A., Shalom, R., & Weisberg, J. 2007. Consider-ations in organizational career advancement: Whatreally matters. Personnel Review, 36: 190–205.

Chartrand, T. L., & Bargh, J. A. 1999. The chameleoneffect: The perception-behavior link and social inter-action. Journal of Personality and Social Psychol-ogy, 76: 893–910.

Cogliser, C. C., & Schriesheim, C. A. 2000. Exploringwork unit context and leader–member exchange: Amulti-level perspective. Journal of OrganizationalBehavior, 21: 487–511.

Deci, E. L., Connell, J. P., & Ryan, R. M. 1989. Self-determination in a work organization. Journal ofApplied Psychology, 74: 580–590.

Deci, E. L., & Ryan, R. M. 2008. Facilitating optimalmotivation and psychological well-being acrosslife’s domains. Canadian Psychology, 49: 14–23.

Deci, E. 1972. Intrinsic motivation, extrinsic reinforce-ment, and inequity. Journal of Personality and So-cial Psychology, 22: 113–120.

Eastman, K. K. 1994. In the eyes of the beholder: Anattributional approach to ingratiation and organiza-tional citizenship behavior. Academy of Manage-ment Journal, 37: 1379–1391.

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

Elias, R. 1986. The politics of victimization. New York:Oxford Press.

Erdogan, B., & Liden, R. C. 2002. Social exchanges in theworkplace: A review of recent developments and fu-ture research directions in leader-member exchangetheory. In L. L. Neider & C. A. Schriesheim (Eds.), Lead-ership: 65–114. Greenwich, CT: Information Age.

Feist, G. J. 1999. Affect in artistic and scientific creativ-ity. In S. W. Russ (Ed.), Affect, creative experience,and psychological adjustment: 93–108. Philadel-phia: Brunner/Mazel.

George, J. M. 2008. Creativity in organizations. In J. P.Walsh & A. P. Brief (Eds.), Academy of Manage-ment annals, vol. 1: 439–477. London: Routledge/Taylor & Francis.

Glomb, T., & Liao, H. 2003. Interpersonal aggression inwork groups: Social influence, reciprocal, and indi-vidual effects. Academy of Management Journal,46: 486–496.

Goldstein, J. 1986. Aggression and crimes of violence.New York: Oxford University Press.

Grant, A. M., & Berry, J. 2011. The necessity of others isthe mother of invention: Intrinsic and prosocial mo-tivations, perspective-taking, and creativity. Acad-emy of Management Journal, 54: 73–96.

Green, S. G., & Mitchell, T. R. 1979. Attributional pro-cesses of leaders in leader-member interactions. Or-ganizational Behavior and Human Performance,23: 429–458.

Harman, G., & Thomson, J. J. 1996. Moral relativism andmoral objectivity. Oxford, U.K.: Blackwell.

Harris, K. J., Kacmar, K. M., & Zivnuska, S. 2007. Aninvestigation of abusive supervision as a predictor ofperformance and the meaning of work as a modera-tor of the relationship. Leadership Quarterly, 18:252–263.

Heider, F. 1958. The psychology of interpersonal rela-tions. New York: Wiley.

Hennessey, B. A., & Amabile, T. M. 2010. Creativity. InS. T. Fiske (Ed.), Annual review of psychology, vol.61: 569–598. Palo Alto, CA: Annual Reviews.

Hinkin, T. R. 1995. A review of scale development prac-tices in the study of organizations. Journal of Man-agement, 21: 967–988.

Hinkin, T. R. 1998. A brief tutorial on the development ofmeasures for use in survey questionnaires. Organi-zational Research Methods, 1: 104–121.

Hofmann, D. A., & Gavin, M. B. 1998. Centering decisionsin hierarchical linear models: Implications for re-search in organizations. Journal of Management,23: 723–744.

Hoobler, J. M., & Brass, D. J. 2006. Abusive supervisionand family undermining as displaced aggression.Journal of Applied Psychology, 91: 1125–1133.

Hurtz, G. M., & Donovan, J. J. 2000. Personality and jobperformance: The big five revisited. Journal of Ap-plied Psychology, 85: 869–879.

Kaplan, S., Bradley, J. C., Luchman, J. N., & Haynes, D.2009. On the role of positive and negative affectivityin job performance: A meta-analytic investigation.Journal of Applied Psychology, 94: 162–176.

Keashly, L., & Harvey, S. 2005. Emotional abuse in theworkplace. In S. Fox & P. Spector (Eds.), Counter-productive work behaviors: 201–236. Washington,DC: American Psychological Association.

Lam, W., Huang, X., & Snape, E. 2007. Feedback-seekingbehavior and leader-member exchange: Do supervi-


sor-attributed motives matter? Academy of Manage-ment Journal, 50: 348–363.

Landy, F. J., & Farr, J. L. 1980. Performance rating. Psy-chological Bulletin, 87: 72–107.

Lian, H., Ferris, D. L., & Brown, D. J. 2012. Does powerdistance exacerbate or mitigate the effects of abusivesupervision? It depends on the outcome. Journal ofApplied Psychology, 97: 107–123.

Liao, H., Liu, D., & Loi, R. 2010. Looking at both sides ofthe social exchange coin: A social cognitive perspec-tive on the joint effects of relationship quality anddifferentiation on creativity. Academy of Manage-ment Journal, 53: 1090–1109.

MacKinnon, D. P., Fritz, M. S., Williams, J., & Lockwood,C. M. 2007. Distribution of the product confidencelimits for the indirect effect: Program PRODCLIN.Behavior Research Methods, 39: 384–389.

Martinko, M. J., & Gardner, W. L. 1987. The leader-mem-ber attribution process. Academy of ManagementReview, 12: 235–249.

Martinko, M. J., Harvey, P., & Douglas, S. C. 2007a. Therole, function, and contributions of attribution the-ory to leadership: A review. Leadership Quarterly,18: 561–585.

Martinko, M. J., Moss, S. E., Douglas, S. C., & Borkowski,N. 2007b. Anticipating the inevitable: When the at-tribution styles of leaders and members clash. Orga-nizational Behavior and Human Decision Pro-cesses, 104: 158–174.

Maslyn, J. M., & Uhl-Bien, M. 2001. Leader-member ex-change and its dimensions: Effects of self-effort andother’s effort on relationship quality. Journal of Ap-plied Psychology, 86: 697–708.

Mathieu, J. E., & Taylor, S. R. 2007. A framework fortesting meso-mediational relationships in organiza-tional behavior. Journal of Organizational Behav-ior, 28: 141–172.

Mayer, D., Kuenzi, M., Greenbaum, R., Bardes, M., & Sal-vador, R. 2009. How low does ethical leadership flow?Test of a trickle-down model. Organizational Behav-ior and Human Decision Processes, 108: 1–13.

Mikula, G. 2003. Testing an attribution-of-blame modelof judgments of injustice. European Journal of So-cial Psychology, 33: 793–811.

Miller, N., Pederson, W. C., Earleywine, M., & Pollock,V. E. 2003. A theoretical model of triggered dis-placed aggression. Personality and Social Psychol-ogy Review, 7: 75–97.

Mullen, E. 2007. The reciprocal relationship betweenaffect and perceptions of fairness. In K. Törnblom &R. Vermunt (Eds.), Distributive and procedural jus-tice: Research and social applications: 15–37. Bur-lington, VT: Ashgate.

Ng, T. W. H., & Feldman, D. C. 2009. Occupational em-beddedness and job performance. Journal of Orga-nizational Behavior, 30: 863–891.

Offermann, L. R., & Malamut, A. B. 2002. When leadersharass: The impact of target perceptions of organiza-tional leadership and climate on harassment report-ing and outcomes. Journal of Applied Psychology,87: 885–893.

Oldham, G. R., & Cummings, A. 1996. Employee creativ-ity: Personal and contextual factors at work. Acad-emy of Management Journal, 39: 607–634.

Rafferty, A. E., & Restubog, S. L. D. 2011. The influence ofabusive supervisors on followers’ organizational cit-izenship behaviors: The hidden costs of abusive su-pervision. British Journal of Management, 22: 270–285.

Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., & Cong-don, R. T., Jr.. 2004. HLM 6: Hierarchical linear andnonlinear modeling. Chicago: Scientific SoftwareInternational.

Restubog, S. L. D., Scott, K. D., & Zagenczyk, T. J. 2011.When distress hits home: The role of contextualfactors and psychological distress in predicting em-ployees’ responses to abusive supervision. Journalof Applied Psychology, 96: 713–729.

Roberts, B. W., Walton, K. E., & Viechtbauer, W. 2006.Patterns of mean-level change in personality traitsacross the life course: A meta-analysis of longitudi-nal studies. Psychological Bulletin, 132: 1–25.

Schat, A. C. H., Frone, M. R., & Kelloway, E. K. 2006.Prevalence of workplace aggression in the U.S. work-force: Findings from a national study. In E. K. Kel-loway, J. Barling, & J. J. Hurrell (Eds.), Handbook ofworkplace violence: 47–89. Thousand Oaks, CA:Sage.

Shalley, C. E., Gilson, L. L., & Blum, T. C. 2009. Interac-tive effects of growth need strength, work context,and job complexity on self-reported creative perfor-mance. Academy of Management Journal, 52: 489–505.

Shalley, C. E., Zhou, J., & Oldham, G. R. 2004. The effectsof personal and contextual characteristics on creativ-ity: Where should we go from here? Journal of Man-agement, 30: 933–958.

Shin, S. J., & Zhou, J. 2003. Transformational leadership,conservation, and creativity: Evidence from Korea.Academy of Management Journal, 46: 703–714.

Snijders, T., & Bosker, R. 1999. Multilevel analysis: Anintroduction to basic and advanced multilevelmodeling. London: Sage.

Sutton, R. I. 2007. The no asshole rule: Building a civ-ilized workplace and surviving one that isn’t. NewYork: Warner Business Books.

Taggar, S. 2001. Group composition, creative synergy,


and group performance. Journal of Creative Behav-ior, 35: 261–282.

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

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

Tepper, B. J., Carr, J., Breaux, D. M., Geider, S., Hu, C., &Hua, W. 2009. Abusive supervision, intentions toquit, and employees’ workplace deviance. Organi-zational Behavior and Human Decision Processes,109: 156–167.

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

Tepper, B. J., Duffy, M. K., Hoobler, J. M., & Ensley, M. D.2004. Moderators of the relationships between co-workers’ organizational citizenship behavior and fel-low employees’ attitudes. Journal of Applied Psy-chology, 89: 455–465.

Tepper, B. J., Moss, S. E., & Duffy, M. K. 2011. Predictorsof abusive supervision: Supervisor perceptions ofdeep-level dissimilarity, relationship conflict andsubordinate performance. Academy of Manage-ment Journal, 54: 279–294.

Tierney, P., Farmer, S. M., & Graen, G. B. 1999. Anexamination of leadership and employee creativity:The relevance of traits and relationships. PersonnelPsychology, 52: 591–620.

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

Tucker, S., Turner, N., Barling, J., & McEvoy, M. 2010.Transformational leadership and childrens’ aggres-sion in team settings: A short-term longitudinalstudy. Leadership Quarterly, 21: 389–399.

Vansteenkiste, M., Simons, J., Lens, W., Sheldon, K. M.,& Deci, E. L. 2004. Motivating learning, performance,and persistence: The synergistic effects of intrinsicgoal contents and autonomy-supportive contexts.Journal of Personality and Social Psychology, 87:246–260.

Zellars, K. L., Tepper, B. J., & Duffy, M. K. 2002. Abusivesupervision and subordinates’ organizational citi-zenship behavior. Journal of Applied Psychology,86: 1068–1076.

Zhou, J. 2003. When the presence of creative coworkersis related to creativity: Role of supervisor close mon-itoring, developmental feedback, and creative per-sonality. Journal of Applied Psychology, 88: 413–422.

Zhou, J. 2010. Opportunities and challenges in re-search on creativity. Keynote speech at the 4th Bi-annual Conference of the International Associationfor Chinese Management Research, Shanghai.

Zhou, J., & George, J. M. 2001. When job dissatisfactionleads to creativity: Encouraging the expression ofvoice. Academy of Management Journal, 44: 682–696.

Zhou, J., & Shalley, C. E. 2008. Expanding the scope andimpact of organizational creativity research. In J.Zhou & C. E. Shalley (Eds.), Handbook of organiza-tional creativity: 347–368. New York: Erlbaum.

Zohar, D., & Luria, G. 2005. A multilevel model of safetyclimate: Cross-level relationships between organiza-tion and group-level climates. Journal of AppliedPsychology, 90: 616–628.

Dong Liu ([email protected]) is an assistantprofessor of organizational behavior at the ErnestScheller Jr. College of Business, Georgia Institute of Tech-nology. He received his Ph.D. in business administrationfrom the Michael G. Foster School of Business, Univer-sity of Washington. His research interests include cre-ativity, turnover, leadership, teams, and international en-trepreneurship, with particular focus on exploring themultilevel interface between individuals and organiza-tional context.

Hui Liao ([email protected]) is a professor of man-agement and organization at the University of Maryland’sRobert H. Smith School of Business. She received herPh.D. from the University of Minnesota’s Carlson Schoolof Management. Her current research interests includeservice quality, leadership, creativity, and workforcediversity.

Raymond Loi ([email protected]) is an associate professor ofmanagement at the University of Macau. He received hisPh.D. in management from The Chinese University ofHong Kong. His research interests include social ex-change relationships, organizational justice, interna-tional human resource management, leadership, and ca-reer satisfaction.

APPENDIX A

Measurement Items

The survey read as follows: “Abusive supervisionmeans the sustained display of hostile, verbal and non-verbal behaviors, excluding physical contact. To whatextent, do you agree that the following may be the reasonfor or cause of your supervisor’s behaviors toward you,which could be considered abusive?”


Performance Promotion Motives1. Desire to elicit high performance from me. (.90)2. Desire to send me messages that mistakes will not be

tolerated. (.86)3. Desire to alert me of my mistakes and problems. (.82)4. Desire to push me to work harder. (.80)5. Desire to stimulate me to meet my performance

goals. (.75)Injury Initiation Motives1. Desire to cause injury on me. (.92)2. Desire to hurt my feelings. (.91)3. Desire to harm my reputation. (.87)4. Desire to make me feel bad about myself. (.81)5. Desire to sabotage me at work. (.77)

Abusive Supervision (Tepper, 2000)My supervisor1. Ridicules me.2. Tells me my thoughts or feelings are stupid.3. Gives me the silent treatment.4. Puts me down in front of others.5. Invades my privacy.6. Reminds me of my past mistakes and failures.7. Doesn’t give me credit for jobs requiring a lot of

effort.8. Blames me to save himself/herself embarrassment.9. Breaks promises he/she makes.

10. Expresses anger at me when he/she is mad for an-other reason.

11. Makes negative comments about me to others.12. Is rude to me.13. Does not allow me to interact with my coworkers.14. Tells me I’m incompetent.15. Lies to me.Creativity (Zhou & George, 2001)This team member1. Suggests new ways to achieve goals or objectives.2. Comes up with new and practical ideas to improve

performance.3. Searches out new technologies, processes, tech-

niques, and/or product ideas.4. Suggests new ways to increase quality.5. Is a good source of creative ideas.6. Is not afraid to take risks.7. Promotes and champions ideas to others.8. Exhibits creativity on the job when given the oppor-

tunity to.9. Develops adequate plans and schedules for the im-

plementation of new ideas.10. Often has new and innovative ideas.11. Comes up with creative solutions to problems.12. Often has a fresh approach to problems.13. Suggests new ways of performing work tasks.


university of colorado boulder | university of colorado boulder -...

Documents