Introduction

Around 75% of workplace mistreatment is perpetrated by leaders against subordinates (Hoel and Cooper 2000). To study leaders’ abusive behavior, researchers employ the term abusive supervision to describe leaders’ behaviors such as public humiliation, threats, and angry outbursts, whereby subordinates perceive that supervisors are engaging “in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact” (Tepper 2000, p. 187). The negative consequences of abusive supervision on victims are straightforward, including diminished job performance, emotional exhaustion, increased psychological stress, job strain, and decreased creativity (see Tepper et al. 2017).

It is, however, worth noting that the impact of abusive supervision goes beyond its targeted victims. As the entire team observes the leader’s abusive behavior, there is a threat of spillover effects of mistreatment behaviors within the team (Qiao et al. 2019; Zhang and Liao 2015; Priesemuth et al. 2014). Third-party employees are those who witness the abusive supervision of a co-worker (Priesemuth 2013). From different angles, scholars endeavor to understand the underlying mechanisms between abusive supervision and third parties’ reactions to witnessing abusive supervision toward peers. These third-party employees, when they witness the supervisor mistreating their peers, may feel anger about the leader’s behavior because it is unjust (Folger and Cropanzano 2001, 1998) or, conversely, they may feel content when they believe that the victim deserves it (Mitchell et al. 2015). An implicit assumption in these studies is that third-party employees’ emotional responses and attitudes are a passive reaction triggered by witnessing peer abusive supervision (Zoghbi-Manrique-de-Lara and Suárez-Acosta 2014).

However, life stories point to another angle about third-party employees’ behaviors: in the fall of 2013, Jonathan Martin, a 6′ 5″ (196 cm), 312 lb (141 kg) talented football player for the Miami Dolphins made the front page of the national newspapers. The New York Times entitled the story “A Classic Case of Bullying” (Shpigel 2014) and The Washington Post reported “Dolphins bullied Jonathan Martin” (Boren 2013). The investigation concluded that Coach Jim Turner and three team members “engaged in a pattern of harassment directed at not only Jonathan Martin, but also another young Dolphins offensive lineman.” The coach was a “person of interest” in the investigation because he as a supervisor did not condemn mistreatment behavior and was involved in harassing team players. He was perceived by the team as a role model in harassing victims (Tullia 2017). In this abusive environment of both the leader and peers, Martin reported a high level of psychological duress that twice made him consider committing suicide.

The Miami Dolphins’ case represents an example of third-party employees learning mistreatment behavior in a context of abusive supervision. Research into the third-party employees’ learning mechanism has both theoretical and practical values to the business ethics domain. Practically, leaders are in a position to encourage employees to create a safe, ethical, and respectful working environment, where all employees can contribute positively to the team with better job performance (Bai et al. 2016). Theoretically, we posit that third-party employees’ abuse of target victims are extensions of abusive supervision that further diminish employees’ performance.

To extend the reaction-assumption of the third-party perspective, this study presents an alternative view and theorizes that third-party employees’ actions, rather than being a passive response to abusive leaders, can be an outcome of witnessing abusive supervision (as in the Miami Dolphins’ example). That is, third-party employees’ mistreatment behavior is a social learning outcome that they learn from the abusive leader. Previous studies have supported our argument that individuals learn aggressive behaviors from a role model in a social setting. For example, Goldstein (1975) revealed that aggressive behaviors of people in high status positions significantly affect similar behaviors of others. Lian et al. (2012) found that high power distance orientation subordinates tend to be more tolerant of supervisory mistreatment and mimic these role models. Through a social learning lens, Liu et al.’s (2012) study showed that abusive supervision passes through organizational hierarchies from the department manager to the team leader. Following a similar line of reasoning, we contend that third-party employees can observe and imitate the team leader’s abusive behaviors. In other words, third-party employees learn the concept of peer mistreatment from abusive supervision, which they later enact on the victim. Because, although third-party employees and the victim officially hold equal positions, abusive supervision toward a certain victim actually leads to differentiations of social status within the team (the victim has a lower social status) (Ogunfowora et al. 2019). Third-party employees might learn the cue of status difference in the team, and this learning makes peer mistreatment more likely to take place. Two types of peer mistreatment may be learning outcomes of abusive supervision: (a) peer harassment, that is, co-workers direct a range of dynamic behaviors at the target, aiming to cause embarrassment or to damage the target psychologically or physically (O'Reilly et al. 2015); (b) peer ostracism, that is, co-workers sideline the target by not giving him/her enough attention and consideration despite it being customary or suitable to do so (Williams 2007).

Confronted with abusive supervision and peer mistreatment, not everyone responds in the same way, because people assign a causal structure to interpret negative events and the effect of these events on themselves (White 1959), a process through which people make sense of their surroundings. Depending on their interpretation of the environment, people act to influence the environment. Proactive personality, conceptualized as an individual’s tendency to actively seek out opportunities to improve and shape the environment, is an important resource that helps individuals to cope effectively with stressors, even in constrained situations (Kim et al. 2005; Bateman and Crant 1993). In a context of abusive supervision and peer mistreatment, proactive personality as an attribute of individual differences may mitigate the negative impact. This study explores the extent to which proactive personality helps individuals to cope with the stressors caused by supervisor abuse and peer mistreatment in order to continue their work.

To sum up, our study makes four primary contributions to the literature. First, we offer a theoretical rationale for, and an empirical test of, how abusive supervision exerts a negative effect on third-party employees’ peer mistreatment of a victim. Several recent studies have attested that abusive supervision can pass through hierarchies through a social learning mechanism, whereby team leaders may take their manager’s abusive behaviors as a role model, thus triggering team leader abusive supervision (Mawritz et al. 2012; Liu et al. 2012). Yet, little is known about the learning and imposing of deviant behaviors among teammates on the same level. We reason that third-party employees’ mistreatment behavior is a learning outcome of abusive supervision and that these employees also impose the learned behavior (e.g., peer ostracism and peer harassment) on targets. Thus, our research contributes to the theoretical understanding of negative consequences of abusive supervision from a social learning perspective and answers Mitchell, Vogel, and Folger’s (2012) call to focus more on third-party employees. Second, we extend the peer abusive supervision literature by highlighting the unique role of witnessing abusive supervision in provoking harassment and ostracism behavior toward victim employees. We add to the limited research stream on the antecedents of peer harassment and ostracism (Christensen-Salem et al. 2020). Third, although emerging work has demonstrated that abusive supervision can be directly and indirectly detrimental to victims’ job performance (e.g., Harris et al. 2007; Pan and Lin 2018), our study goes one step further by including both job performance and creative performance in the investigation, implying the generalization of our findings in two separate samples. The results could have important implications for monitoring abusive supervision and peer mistreatment, depending on the outcome sought. Fourth, our work incorporates an individual factor, i.e., proactive personality, that may alter the spillover effects of abusive supervision on both job performance and creative performance through third-party peer mistreatment. In doing so, we refine the workplace mistreatment literature by demonstrating that it is necessary to consider important boundary conditions that affect employees’ reactions to abusive supervision (Fig. 1).

Fig. 1
figure 1

The research model

Full size image

Theoretical Framework and Hypothesis Development

The Effects of Abusive Supervision on Employee Performance

Extensive competition in today’s business environment has impelled organizations to motivate employees to perform better (Schneider et al. 2018) and to stimulate employees’ creative ideas for organizational improvement (Amabile 2018). Employee job performance, broadly defined as the value of employee activities in meeting or exceeding expectations about job requirements (Harrison et al. 2006), is the basis of, and vital to, the overall success of the organization. Employees’ creative performance refers to employees’ novel and useful ideas relating to work procedures, methods, services, or products that are ultimately produced (Zhou and Shalley 2003). It is widely regarded as an important and valuable source of organizational development and competitiveness (Liu et al. 2017).

Leadership is an important element that may nourish or impede employee outcomes (Gong et al. 2009). Abusive supervision, according to a handful of studies, exerts a negative influence on both employees’ job and creative performance (see Y. Zhang and Liao 2015). First, the experience of abusive supervision may cause psychological distress, such as depression, anxiety, emotional exhaustion, even poor-quality sleep (Han et al. 2017). Theoretically and proven empirically, stress caused by an upward supervisor imposes a threat of resource loss (either perceived or actual) and makes unreasonable demands on individuals (Harris et al. 2007). The sense of loss and excessive demands distract employees from performing their work. Rather, employees have to expend time and energy on managing “upwards.” In such a state of stress, they are unlikely to put sufficient effort into completing their work to a high quality, not to mention proposing new ideas. Second, abusive supervision takes the form of negative interactions between the leader and subordinates, leading to employees’ pessimistic evaluation of their relationship. Employees’ negative perception of the leader–member relationship results in negative reciprocation with negative paybacks (Cropanzano and Mitchell 2005). Thus, subordinates reciprocate the abusive leader’s negative treatment by withholding their performance efforts, leading to decreased job performance and less creative performance (Xu et al. 2012).

While the above reasoning of how abusive supervision negatively impacts employee job performance also applies to the abusive supervision–employee creative performance link, but we additionally suggest that abusive supervision further stymies employees’ creative performance by undermining intrinsic motivation. Intrinsic motivation is a motivational state whereby employees carry out their work because they are interested in it (Zhang et al. 2014), because, with intrinsic motivation, individuals are motivated to look for better or more creative ways to do their tasks. Therefore, intrinsic motivation is conductive to employees’ creative performance. However, when employees are abused by their supervisor, they form negative perceptions of their self-worth at work (Jian et al. 2012), doubt their value and contributions to the team, and question their competence to fulfill their job (Tepper et al. 2011). Such doubt may diminish the enjoyment that employees obtain from proposing new ideas (Tepper 2000). Employees lose their motivation to go the extra mile, hide their knowledge (Jahanzeb et al. 2019), and withhold novel and original ideas for fear of rocking the boat (Grant 2017). Thus, abusive supervision reduces intrinsic motivation, consequently resulting in a decrease in employees’ creative performance. Aligned with the previous research, we expect that:

Hypothesis 1

Abusive supervision relates negatively to the target employee’s performance in terms of (a) creative performance and (b) job performance.

The Spillover Effects of Abusive Supervision: A Social Learning Mechanism

Currently, researchers interested in the relationship between abusive supervision and the two types of employee outcomes (i.e., job performance and creative performance) have focused largely on (1) the target employees, arguing that the target employees’ intrinsic motivation and their perception of the leader–member relationship, both of which are crucial for employee outcomes, are undermined by abusive supervisory (e.g., H. Zhang et al. 2014; Zheng and Liu 2017), or (2) third-party employees, proposing a social comparison process whereby third-party employees compare their own experience of abusive supervision with the witnessed abusive supervision of the target employee. The comparison process in turn impacts the third-party employees’ outcomes (e.g., Jiang et al. 2019). Third-party employees’ reactions to witnessing the abusive supervision of the target employee include feelings of sympathy (Mitchell et al. 2015), schadenfreude (Qiao et al. 2019; Xu et al. 2020), unfairness (Blader et al. 2013), and so on. Both types of research interests take a reactive position to show how either the victim or the third-party employees respond to the directly experienced or the witnessed abusive supervision.

However, the Miami Dolphins’ case in the introduction implies that third-party employees might be a channel between abusive supervision and the target employee. Specifically, Coach Turner was perceived by the team as a role model in harassing the victim, resulting in several team members further abusing the victim. We contend that witnessing abusive supervision, besides the possibly triggering third-party emotional/justice reactions, might contrarily become a learning source for third-party employees. To explain this mechanism, we apply the social learning theory in the hypotheses development.

The tenets of social learning theory argue that individuals can acquire social behavior through observing others and imitating behaviors of significant social contacts, known as role models (Bandura 1977). By observing their role model, individuals make judgements of the observed behavior based on the witnessed consequences and determine behaviors that are deemed to be encouraged or acceptable in a given situation (Trevino et al. 2000). This attention on the role model is one primary condition for social learning. In an organizational context, people in formal positions of authority are usually deemed credible as role models. Their behavior is a key source of social learning in a team (Bai et al. 2019). From observation and interpretation, employees determine how to behave in the team context (Tucker et al. 2010). Employees imitate not only positive but also negative behaviors by the role model. Both Liu et al. (2012) and Mawritz et al. (2012) found a trickle-down model of abusive supervision across hierarchical levels. In their study, team leaders interpreted their manager’s abusive behavior as meaning that workplace mistreatment of peers was acceptable, or to some extent legitimate.

Given that teammates are officially equal to one another, third-party employees cannot behave as if they are in a supervisory role regarding the target. As evidenced in several studies, employees under abusive supervision display aggressive behaviors toward their peers (Tucker et al. 2010; Pradhan et al. 2019; Richard et al. 2018). Sensing that behavioral signals from the abusive leader mean that peer mistreatment is acceptable, third-party employees’ learning outcomes toward the target employee can be explicit and direct, known as peer harassment such as humiliation and assaults, or implicit and indirect, known as ostracism such as ignoring and silence. To present a comprehensive picture of negative learning outcomes, we include both harassment and ostracism in our examination. According to Bowling and Beehr (2006), peer harassment refers to explicit interpersonal behavior, including a range of active verbal and non-verbal behaviors, that are intended to harm another employee in the workplace. Unwanted and aggressive behaviors are performed, including extreme harassment such as homicide and physical assault, as well as the more common harassment in minor instances, such as dirty looks, obscene gestures, yelling, and belittling. These harassment behaviors engage the target in social dynamics that involve social interactions (O'Reilly et al. 2015). This study excludes harassment motivated by sex and race. Unlike harassment, which involves the performance of unwanted behavior toward the target, ostracism involves the absence of a wanted behavior. Robinson, O’Reilly, and Wang (2013) defined ostracism as an individual or a group omitting to take actions that engage another organizational member when it is socially appropriate to do so. Examples of ostracism includes ignoring greetings, exclusion from invitations, falling silent when the target employee seeks to join the conversation.

Being exposed to peer abusive supervision, third-party employees learn negative, mistreating behaviors from the leader and in turn mistreat the target employee. Because of position similarities among team members, third-party employees may choose to emulate mistreatment behaviors in the form of peer harassment and peer ostracism, as their learning outcomes. The leader, meanwhile, may feel pleased that the rest of the team agrees with his/her behaviors toward the target employee, giving the leader a felt authority in the team. For the third-party employees, the abusive leader’s inaction or even acquiescence to peer mistreatment further reinforces the strong learning norm that the victim deserves mistreatments by others. From a social learning perspective, we propose the following:

Hypothesis 2

Abusive supervision relates positively to peer harassment.

Hypothesis 3

Abusive supervision relates positively to peer ostracism.

Peer Harassment and Peer Ostracism as Mediators

Both peer harassment and peer ostracism violate workplace norms of respect (Ferris et al. 2017). Ideally, a leader enacting positive leadership (e.g., ethical leaders or transformational leadership) would condemn harassment and ostracism behavior because they are unethical and toxic to employee performance (Christensen-Salem et al. 2020). However, when the leader him/herself is abusive to employees, it is unlikely that he/she would explicitly criticize peer employees’ mistreatments. Furthermore, team members perceive authority figures in teams; for example, the team leader has implicit information about everyone’s status within a team. Regarding abusive supervision, evidence has shown that abusive supervision results in employees’ perceptions of status differences in the team (Ogunfowora et al. 2019). In particular, the abusive leader signals to third-party employees that the target employee has a lower status and a lower value to the team. Third-party employees formulate an impression that workplace mistreatment of the target employee of lower status is the norm or even encouraged within their team (Mawritz et al. 2012). Consequently, third-party employees join the abusive supervisor by imposing peer mistreatments, such as peer harassment and peer ostracism, on the victim.

For the target employee, direct and explicit peer harassment becomes another notable source of social concern in the workplace. Social concern relates to a person’s self-concept or perceptions of social status in the team (Frone 2000). Emotional reactions follow this social concern, leading to burn out, nervousness at work, dread, and depression (Nielsen and Einarsen 2012). The experience of emotional fatigue can directly affect how victims behave (Malik et al. 2019). For instance, employees cannot concentrate on their work, leading to reduced performance of both routine and creative tasks (Hur et al. 2016). Apart from emotional reactions, harassment undermines victims’ confidence in their own ability to control events that affect their lives. When employees are confronted with persistent harassment from their co-workers, they may believe that their ideas or work are not respected and considered. Gradually, they will develop a self-concept of not being sufficiently competent to finish their job, handle difficult tasks, or challenge the status quo. Employees’ self-worth declines substantially, impeding their job performance and creativity. Following the above line of reasoning, we predict that third-party employees’ harassment and ostracism behavior is a linking pin between abusive supervision and the target employee’s creative performance and job performance:

Hypothesis 4

Peer harassment mediates the negative relationship between abusive supervision and the target employee’s performance in terms of (a) creative performance and (b) job performance.

Third-party employees’ actions may also take the form of peer ostracism as the learning outcome of abusive supervision. Abusive supervision of the victim indicates that the victim has less value to the role model, causing teammates to ostracize the victim. The impact of ostracism focuses on its threat to a fundamental human need—the need to belong (Walasek et al. 2015; Peng and Zeng 2017; O'Reilly et al. ). When employees are ostracized by their co-workers, such as their greetings not receiving a reply, not being invited to participate with others when it is customary to do so, they will get the feeling that they are marginalized and doubt whether they are considered part of the group (Jetten et al. 2001). Thus, being ostracized by peers may threaten their sense of belonging in the team. Consequently, this lack of identification with the group will have a profound effect on their interest in their normal work (W. Liu et al. 2016), hindering their commitment and involvement in team work (Creasy and Carnes 2017). Target employees develop a sense of being an outsider, associated with exclusion, feelings of worthlessness, lack of value, and inferiority. Therefore, employees who experience ostracism from third-party employees may be less motivated to perform tasks and propose creative solutions to improve team performance:

Hypothesis 5

Peer ostracism mediates the negative relationship between abusive supervision and the target employee’s performance in terms of (a) creative performance and (b) job performance.

The Contingent Effect of Proactive Personality

The last interest in our study is to investigate how individual difference alters the negative mediation path. In a constrained working environment, a person’s ability to remain positive and then identify and act against the constraining forces is crucial for his/her performance. The concept of proactive personality captures these characteristics. Bateman and Crant (1993) defined proactive personality as “one who is relatively unconstrained by situational forces and who effects environmental change” (p. 105). A proactive personality is an important resource for individuals to cope effectively with stressors encountered (Bolger and Zuckerman 1995). It shows the degree to which a person reaches out to change things in the workplace and to take actions against an unfavorable environment. Research on proactive personality has suggested that people vary in their expectations regarding their ability to change the environment (Zhao et al. 2013). Specifically, proactive individuals show initiative and persevere to make meaningful change, and they remain motivated by their internal desire to improve their skills, acquire knowledge, and realize their goal (Parker et al. 2006), whereas while non-proactive people have less intention to change things. An early study reported that workers with a strong proactive personality are more resilient in response to high job demands and report less job strain (Parker and Sprigg 1999; Hoyt and Murphy 2016).

Abusive supervision, peer harassment, and peer ostracism are all clear signs to the target employee that people in the team do not welcome him/her, suggesting a challenging environment into which to blend. In such an environment, individuals receive limited support from the supervisor and co-workers. Workers with a high proactive personality tend to engage in behaviors to (re)gain supports. For instance, they may be more determined in resolving peer harassment and peer ostracism because of attributes related to proactivity, including change orientation, self-starting, and confidence (Parker et al. 2010). They initiate more network building (Thompson 2005), a gesture that helps to “clear the air” during interpersonal constraints (Park and DeFrank 2018). Consistent with these arguments, some evidence have indicated that proactive workers show a less negative response to workplace mistreatment than non-proactive workers (Zhao et al. 2013). For instance, an interview transcript in Lee et al.’s (2013) study illustrated how proactive personality may buffer the negative influence of an abusive situation:

“I think my leader's harsh behaviors are effective in fostering my creativity to some extent… A certain degree of abusive supervision will give me a sense of crisis, and the negative mood from abusive supervision will make me uncomfortable. This discomfort frequently drives me to ask ‘What is the problem?’ or ‘How can I get rid of this mood?’ In that situation, I am better at finding new solutions to problems than in a comfortable and satisfying situation”.

An interview quote from Lee et al. (2013)

Reflecting upon the sample quote and research evidence, we argue that target employees’ proactive personality should reduce the damaging impact from the experience of peer mistreatments such that the level of creative performance and job performance remains good enough, if not better. In contrast, less proactive people may feel helpless about the hostile work environment and show less initiative in changing their relationship with the supervisor and co-workers (Parker et al. 2001). Their inactive attitude toward change strengthens co-workers’ impression of weakness and self-isolation from the group. For non-proactive employees, the situation of harassment and ostracism is unlikely to change. In the long term, less proactive workers suffer more psychological exhaustion, resulting in worse job performance and less creative performance. Therefore, we predict that the damaging effect of peer mistreatment is greater for non-proactive workers for proactive workers:

Hypothesis 6

The negative indirect effect of abusive supervision via peer harassment on the target employee’s performance will be weaker when the target employee’s proactive personality is higher rather than lower.

Hypothesis 7

the negative indirect effect of abusive supervision via peer ostracism on the target employee’s performance will be weaker when the target employee’s proactive personality is higher rather than lower.

Study 1

Methods

Sample and Procedure

In Study 1, we examined mainly the spillover effects of abusive supervision on the creative aspect of employee performance via third-party mistreatment. We collected data from MBA students in a business school in northeast China. We invited 95 MBA students who were leaders, as well as their 4–6 subordinates, to participate in a longitudinal survey study. Each participant signed a consent form. Then, we conducted data collection in three phases. In Time 1, these subordinates (total 456) completed a questionnaire to rate their leaders’ abusive supervision. We prepared an envelope with a stamp, with which they could mail their response directly back to us. This round, we received 428 subordinates’ replies. One month later, in Time 2, these 428 subordinates received a follow-up questionnaire to report their perceptions of harassment and ostracism from peers, as well as their proactive personality. We received 401 responses with the same mail-back-envelope approach. In Time 3, another month later, we sent the last questionnaire to the MBA student leaders to assess their subordinates’ creative performance. We used employees’ working IDs to match the supervisor–subordinate pairs. Hence, we had different data sources to prevent common method bias (Podsakoff et al. 2003). After deleting non-matched and incomplete questionnaires, we had paired data for 388 subordinates from 76 teams for further analyses. The average age of subordinates was 31.00 (s.d. = 7.10), their average tenure was 4.97 (s.d. = 5.05), and about 43% were males. For team leaders, the average age was 38.36 (s.d. = 7.02), their average tenure was 8.20 (s.d. = 5.16), and 70% were males.

Measures

We adopted a 7-point Likert scale, with “strongly disagree” (1) to “strongly agree” (7) as scale anchors.

Abusive Supervision

We adopted Aryee et al.’s (2007) 10-item measurement of leader’s abusive supervision. An example item was “my supervisor makes negative comments about me to others.” The reliability for the scale was 0.91.

Peer Harassment

To measure harassment, we used O’Reilly et al.’s (2015) 10-item scale. An example item was “Team members often spread gossip and rumors about me.” The reliability for the scale was 0.88.

Peer Ostracism

We measured peer ostracism with the 10-item Workplace Ostracism Scale (Ferris et al. 2008). An example item was “Team members treat me as nonexistent.” The reliability for the scale was 0.93.

Proactive Personality

To measure proactive personality, we adopted Seibert et al.’s (1999) 6-item scale. An example item was “If I see something I don’t like, I fix it.” The reliability for the scale was 0.89.

Employee creativity

We used Farmer et al.’s (2003) 4-item scale to measure employee creativity. An example item was “The employee always takes new ideas first.” The reliability for the scale was 0.92.

Control Variables

We included subordinates’ age, gender, and tenure as well as leaders’ age, gender, and tenure, and team size as control variables.

Analytic Strategy

As the subordinates’ data were nested in teams, we should consider the non-independence of the data. In addition, peer harassment and peer ostracism had large ratios of team-level variance (the ICC(1) values of both peer harassment and peer ostracism were 18.8%), indicating the necessity to partition their individual- and team-level variances in the analyses. Therefore, we applied the multilevel technique (hierarchical linear modeling, HLM) for hypotheses testing (Raudenbush 2004). HLM can control for non-independence in data by dividing the entire variance into within-group and between-group sections. Furthermore, to test the significance of conditional indirect effects, we adopted the Monte Carlo bootstrapping method with 20,000 repetitions for the parameter estimations (Preacher and Selig 2012).

Results and Discussion

Multilevel Confirmatory Factor Analyses

We adopted multilevel confirmatory factor analyses with Mplus to check the quality of our measures. The item parceling method was used for the multiple-item constructs to reduce the parameters to sample size ratio, which might cause a non-convergence problem in CFA analyses (Chen et al. 2018; Marsh et al. 2009; Muthén and Asparouhov 2012). The baseline model in Table 1 yielded a good model fit, and all the factor loadings were above 0.74, indicating good convergent validity. We also created several rival models that combined different factors. However, none of these rival models had a better model fit than the baseline model. Thus, our five measures had good discriminant validity as well. The summary in Table 2 shows the means, standard deviations, and correlations among the variables. Abusive supervision has a negative association with employee creative performance and positive associations with peer harassment and ostracism; creative performance has negative associations with peer harassment and peer ostracism.

Table 1 Results of Multilevel Confirmatory Factor Analyses in Study 1
Full size table
Table 2 Means, standard deviations, and correlations of the variables in study 1
Full size table

Testing the Hypotheses

Table 3 shows the HLM results of the hypotheses testing. Hypothesis 1 predicted a negative relationship between abusive supervision and target employee performance. The results of M2 in Table 3 reveal that leader abusive supervision had significantly negative impact on employee creative performance, lending support to H1. Hypotheses 2 and 3 proposed that abusive supervision had positive relationships with peer harassment and peer ostracism, respectively. The coefficients of abusive supervision in the regressions of peer harassment (M4) and peer ostracism (M6) are both significant and positive. Thus, H2 and H3 are also supported.

Table 3 Results of HLM Analyses in Study 1
Full size table

Hypotheses 4 and 5 predicted the mediating roles of peer harassment and peer ostracism between abusive supervision and employee performance. In M2 of Table 3, we found a significant relationship between abusive supervision and employee creative performance. In M7, we introduced both abusive supervision and the mediators (i.e., peer harassment and peer ostracism) in the regression. The results indicate that peer harassment and peer ostracism had significant and negative relationships with creative performance, and the effect of abusive supervision on creative performance became insignificant in M7. We used the Monte Carlo bootstrapping method (Bootstrap = 20,000) to test the mediation effects. The indirect effects through peer harassment and peer ostracism were − 0.07 (95% CI [− 0.12, − 0.02], not including zero) and − 0.05 (95% CI [− 0.09, − 0.01], not including zero), respectively, providing support for mediation Hypotheses 4 and 5 when the outcome was employee creative performance (Preacher and Selig 2012).

Hypotheses 6 and 7 predicted that proactive personality would weaken the indirect negative effects of abusive supervision on employee performance. We first checked the simple moderation effect in Model 9 of Table 3. For H6, the results show that the interactive effect of target employee’s proactive personality and peer harassment on employee creative performance was significant but in a different direction (negative, rather than positive), indicating that, when target employees have a higher level of proactive personality, peer harassment leads to much lower creative performance. Following Aiken and West (1991), we depicted this moderating effect in Fig. 2, with 1-SD above and below the mean values of peer harassment and proactive personality. The simple slope analyses showed that, when proactive personality was high, the effect of peer harassment was significantly and negatively related to employee creative performance (simple slope = − 0.35, p < 0.05), whereas, when proactive personality was low, the effect became non-significant (simple slope = − 0.04, p > 0.10). For the conditional indirect effects through peer harassment, we applied the Monte Carlo bootstrapping method (repetition = 20,000) to test the indirect effects at different levels of proactive personality. The results in Table 4 show that, when proactive personality was high, the indirect effect of abusive supervision on target employees’ creative performance via peer harassment was significant; the indirect effect became non-significant when proactive personality was low; the difference between the two conditions was significant. Thus, our findings do not support H6, but provide significant evidence in an opposite way. For H7, as we did not find a significant interaction between proactive personality and peer ostracism in M9, H7 receives no support.

Fig. 2
figure 2

The interactive effect on employee creative performance

Full size image
Table 4 Results of conditional indirect effects in study 1
Full size table

The result of Study 1 supported the mediation hypothesis that abusive supervision exerted a negative and indirect effect on the victim employee’s creative performance through third-party employees’ mistreatment behaviors (both harassment and ostracism). However, the result opposed the hypothesized direction of the moderated mediation of H6. It showed that victim employee’s proactive personality accentuated, rather than buffered, the negative impact of abusive supervision through peer harassment on the victim employee’s creative performance. Proactive employees, who are supposed to actively and determinedly seek solutions until meaningful change occurs, are not able to keep positive in an abusive environment (Bin et al. 2020). A possible explanation is that peer harassment may excessively exceed the individual’s expectations regarding his/her ability to change the environment (Guo et al. 2020). The feeling of inability to make a change amplifies the negative emotions and perceptions of self-determination, thus stifling creative performance (S. Lee et al. 2013; Harvey et al. 2006). On this occasion, one can expect an alternative H6, that is, the negative effect of abusive supervision on the victim’s performance becomes stronger when the victim has a high proactive personality. To avoid making a conclusion from statistical errors, we are open to the possibility of an opposite direction of H6. Therefore, we conduct a second study to retest our model.

Study 2

Methods

Sample and Procedure

We collected a second sample to replicate the findings of Study 1 with a different form of performance, i.e., employees’ objective job performance. We drew the sample from service staff in one big Chinese airline company. We contacted the HR director and randomly selected 265 service employees from the names provided. Each employee received a reward of 100 RMB to encourage participation. We obtained their consent form before the survey and assured them that their input would be confidential. As mentioned, one difference was that we used objective job performance in Study 2, rather than the subjective creative performance used in Study 1. The change allowed us to replicate the results with both objective and subjective performance measures, as well as two types of employee performance, thereby greatly enhancing the robustness of our findings. We conducted data collection with online surveys in three phases. In Time 1, we distributed an online questionnaire to these service employees to rate their leaders’ abusive supervision. This round, we received 247 replies. One month later, in Time 2, we sent a second online questionnaire to these 247 employees to ascertain their perceptions of harassment and ostracism from peers, as well as their own proactive personality. We received 234 responses this round. In Time 3, another month later, we asked the HR department to provide the objective job performance evaluations of these 234 employees. The company evaluated its employees’ objective job performance on a scale of “A” to “D,” i.e., “A” as excellent, “B” as good, “C” as normal, “D” as below normal. After deleting non-matched data, we finally obtained 228 datasets for further analyses. The average age of the 228 service employees was 28.54 (s.d. = 6.29), their average tenure was 5.96 (s.d. = 3.61), and about 71% were females.

Measures

We adopted the same scales as in Study 1 for abusive supervision, peer harassment, peer ostracism, and proactive personality. The reliabilities for these scales were 0.95, 0.94, 0.95, and 0.81, respectively. For the objective performance evaluation, we coded “A” to “D” as 4 to 1. We included age, gender, and tenure as control variables.

Analytic Strategy

As the service employees were from different teams, the data were independent in nature. Therefore, we applied multiple regression with SPSS 22.0 for the data analyses. To test the significance of conditional indirect effects, we adopted Hayes’ PROCESS MACRO tool (Version 3.3) to conduct bootstrap estimations of the parameters (bootstrap = 20,000) (Hayes 2017; Hayes and Rockwood 2020).

Results and Discussion

Confirmatory Factor Analyses

Table 5 illustrates the results of the confirmatory factor analyses. The baseline model (including abusive supervision, peer harassment, peer ostracism, and proactive personality) had a good model fit, and all the factor loadings were above 0.57. As shown in Table 5, none of the rival models yielded a better model fit than the baseline model. Thus, our measures in Study 2 had good convergent and discriminant validities. The summary in Table 6 shows the means, standard deviations, and correlations among the variables. We found positive relationships between peer harassment, peer ostracism, and abusive supervision, and negative relationships between abusive supervision, peer harassment, peer ostracism, and employee job performance.

Table 5 Results of confirmatory factor analyses in study 2
Full size table
Table 6 Means, standard deviations, and correlations of the variables in study 2
Full size table

Testing the Hypotheses

Table 7 provides the results of the multiple regression analyses. Similar to the findings in Study 1, we found significant and negative relationships between abusive supervision and objective job performance (in M2), and significant and positive relationships between abusive supervision and peer harassment (in M4) and peer ostracism (in M6), H1, H2, and H3 are again supported.

Table 7 Results of multiple regression analyses in study 2
Full size table

Hypotheses 4 and 5 predicted the mediating roles of peer harassment and peer ostracism between abusive supervision and employee performance. In M7, we introduced both abusive supervision and the mediators (i.e., peer harassment and peer ostracism) as predictors of job performance, and we found that only peer harassment had a significant relationship with job performance, and there was not a significant relationship between peer ostracism and job performance. Again, the significant effect of abusive supervision on job performance in M2 became insignificant in M7. The Monte Carlo bootstrapping method (Bootstrap = 20,000) used for the mediation test indicated that the indirect effect through peer harassment was -0.06 (95% CI [− 0.08, − 0.01], not including zero) (Preacher and Selig 2012). Thus, the results again supported H4 when the outcome was job performance.

For H6 and H7, the results in M9 of Table 7 indicate that proactive personality negatively moderated the relationship between peer harassment and employee job performance, but not the relationship between peer ostracism and job performance. This is consistent with the finding in Study 1 that, when target employees have a high proactive personality, peer harassment leads to much lower employee performance, confirming the accentuating role of proactive personality in moderating the mediation path. We depicted this moderating effect in Fig. 3, and the simple slope analyses showed that, when proactive personality was high, the effect of peer harassment on job performance was significant (simple slope = − 0.23, p < 0.05); when proactive personality was low, the effect became non-significant (simple slope = − 0.03, p > 0.10). For the conditional indirect effect of peer harassment, we applied Hayes’ PROCESS MACRO tool (Version 3.3) for the moderated mediation test (Bootstrap = 20,000). The results in Table 8 show that, when proactive personality was high, the indirect effect of abusive supervision on employee job performance via peer harassment was significant, whereas when proactive personality was low, the indirect effect was not significant. Overall, the above results generally replicate the findings in Study 1, in particular, it revalidates the opposite direction of H6 in Study 1.

Fig. 3
figure 3

The interactive effect on employee job performance

Full size image
Table 8 Results of Conditional Indirect Effects in Study 2a
Full size table

General Discussion

Researchers have previously reported the negative impact of abusive supervision on the victim employee’s (Harris et al. 2007) and third-party employees’ creative performance and job performance (Jiang et al. 2019). What is less clear is whether third-party employees become evil transporters for the abusive supervisor to hurt the victim. To address this research gap, the present study examined how abusive supervision was related to the victim’s creative performance and job performance through third-party employees’ mistreatment behaviors and when proactive personality moderated this relationship. Below, we discuss the findings and the theoretical and managerial implications.

Theoretical Implications

The current research makes contributions to the abusive supervision and employee performance literature in four ways. The primary contribution is that our research investigates the correlation between abusive supervision and the target employee’s performance from a third-party perspective. Past research regarding the correlation between abusive supervision and employee performance focused on either the victim (W. Liu et al. 2016) or the third-party employees (e.g., Jiang et al. 2019; Porath and Erez 2009). Research about whether and how third-party employees who witness abusive supervision affect the victim’s creative performance and job performance is missing. The current research addresses this gap by incorporating the third-party employees’ mistreatment behavior into the already established negative relationship between abusive supervision and victim employee’s performance. To generalize the model, this study incorporates both subjective creative performance and objective job performance in the investigation. Our results support the contention that third-party employees’ vicious behaviors, i.e., both harassment and ostracism, act as channels for the link between abusive supervision and employees’ creative performance. Investigating these potential channels advances the understanding of the dark side of leadership and how this darkness contaminates the work environment among employees.

Second, explaining third-party employees’ workplace abusive behaviors, this study extends the research on third-party employees by proposing a social learning mechanism (Bandura and Walters 1977). The very few studies that apply social learning theory have found that abusive supervision passes through organizational hierarchies in such a way that team leaders view their managers as social role models to learn from, thereby abusing their team members (Liu et al. 2012; Tu et al. 2018). However, it should be noted that, as co-workers do not hold a position of authority, they cannot abuse the target with supervisory power as the leaders do. Rather, as supported by our study, the outcomes of third-party employees’ social learning from abusive supervisors can result in two distinct forms of mistreatment: peer harassment and peer ostracism. In this regard, we show that social learning theory makes a notable contribution to the literature on abusive supervision and peer mistreatment.

Third, we examine the influence of two types of peer mistreatment, i.e., peer ostracism and peer harassment, on employees’ creative performance and job performance. Several studies have found that, compared to harassment, ostracism is more strongly related to negative outcomes, including health problems, lower affective commitment and job satisfaction, higher psychological withdrawal, and a tendency to quit (Ferris et al. 2017). Therefore, researchers have called for studies to distinguish workplace ostracism from other mistreatment behaviors (Balliet and Ferris 2013; Hitlan et al. 2006). Our research addresses this call. However, unlike O’Reilly et al.’s (2015) study, which claimed that ostracism has a much stronger effect on employees’ work-related attitudes, our Study 1 found that harassment had a comparatively stronger impact on employee creative performance. Study 2 confirmed this result by showing the strong impact of harassment on job performance. The results of Study 2 failed to find a significant influence of ostracism on job performance. One possible explanation for this may relate to the underlying psychological mechanisms of harassment and ostracism (Ferris et al. 2017). Ostracism creates a context of social isolation in such a way that peers subtly deplete the victim’s sense of belonging (Ferris et al. 2017). We agree that a sense of belonging is critical for one’s psychological health (Rubin et al. 2019), negatively affecting the attitudinal outcomes such as job satisfaction, a tendency to quit, or psychological withdrawal. However, we speculate that the need to belong may be less crucial for actual performance (e.g., creative performance and job performance). This is because individuals are obliged to fulfill their job requirements even though they do not feel that they belong to the group. Moreover, creativity is about thinking outside of the box and challenging the environment rather than looking to belong. Thus, ostracism may be less directly harmful in affecting behavioral performance, whereas harassment, when it appears in the workplace, displays an obvious intention to harm the target, causing the target to have negative emotions and feelings of unfairness (Andersson and Pearson 1999; J. Lee and Jensen 2014), furthermore distracting the target from fulfilling the job requirement normally or from coming up with new ideas. Thus, in a context of abusive supervision, peer harassment has a worse impact on employee performance than peer ostracism does. Our study contributes to the debates regarding these two distinct forms of workplace mistreatment.

Finally, regarding the moderating effect, both Study 1 and Study 2 evidenced that proactive personality worsens the negative mediation path of abusive supervision on the victim’s performance via peer harassment. The finding enriches our understanding of the “double-edged” nature of proactive personality, especially in a constrained environment (Altura et al. 2020). In addition, the accentuated negative impact of peer harassment on the victim’s creative performance and job performance depends on a proactive personality, whereas the power of peer ostracism on the victim’s creative performance does not depend on a proactive personality. Proactive employees are those who actively and determinedly seek challenges until meaningful change occurs in their goals (Crant 2000). It is possible that employees with a proactive personality are so enthusiastic about making changes that they blind themselves from sensing subtle or even invisible social signals, such as ostracism. As a result, the moderating effect of proactive personality between ostracism and creative performance is insignificant. Unlike ostracism, harassment is direct and explicit. For employees with a high proactive personality, confronting blunt harassment puts a damper on their enthusiasm, and this in turn exaggerates the negative link between peer harassment and creative performance. Thus, our results suggest that examining the victim’s personality as a boundary condition may generate valuable insights into the research on abusive supervision processes.

Practical Implications

This study provides some suggestions for practitioners. As employee performance is crucial to the success of firms in a competitive environment (Amabile et al. 2002), it is important to have a comprehensive understanding of the factors, especially negative factors, that may motivate or obstruct employee performance. Our study uncovers an intermediate route whereby third-party employees channel the negative link between abusive supervision and employee performance. We show that abusive supervision not only directly diminishes employees’ creative and job performance, but also is perceived by third-party employees as a behavioral model. Therefore, team leaders should pay attention to their leadership style, because negative role modeling can cultivate a vicious team. We would go so far as to suggest that very strict strategies banning abusive supervision should be adopted in an organization (Sutton 2007). That is, leaders should be monitored to ensure a suitable management style toward employees, and the organization should offer leadership training to leaders to improve their leadership approach.

Second, our results confirm that both peer harassment and peer ostracism have strong and negative impacts on the victim’s creative performance. To create friendly working conditions for employees to engage in creative behavior, organizations need to cultivate a culture of fairness and sincerity. On the one hand, organizations should develop and enforce strict policies against workplace harassment and ostracism. On the other hand, organizations should apply soft management techniques, such as organizing internal activities to help improve interpersonal relations among employees, providing training for employees to better manage conflicts with their co-workers, or establishing an office where employees can consult about workplace conflicts. All these measures may be helpful in breaking the cycle of learning unethical behaviors. For the victim abused by his/her leader and peers, the office could also offer psychological counseling help.

Third, our research reveals that, if the victim has a proactive personality, this accentuates the negative impact of peer harassment on creativity. Thus, when harassment incidents are reported, organizations may need to pay extra attention and offer help to employees with a proactive personality, because they may have a strong negative reaction to harassment incidents.

Limitations and Directions for Future Research

Although the current research offers notable contributions to the literature on abusive supervision and workplace mistreatment, it is not without limitations. First, although the multi-source data followed a time-lagged design (Podsakoff et al. 2003), we could not definitively rule out the possibility of common method bias, nor can we be sure about the direction of causality. For example, peer employees’ harassment or ostracism may also result from the target’s persistence in challenging the status quo (i.e., proposing new ways of working) and seeking to differentiate him/herself from the group. Such behavior may annoy third-party employees, resulting in their continuously ignoring the target. Future studies should use a longitudinal design that measures the same variables in each time period to establish causality. Second, in this study, supervisors rated employees’ creative performance, but multiple factors could lead to a biased rating, such as liking. Although the second study generally replicated the model by using an objective variable (i.e., job performance), we suggest that future research should combine subjective ratings and objective data, such as numbers of patents to measure employee creative performance (Ng and Feldman 2012), because creative performance is nowadays becoming crucial for organizational development.

The third concern is that the operationalization of the peer harassment and peer ostracism variables was sourced from target employees rather than from third-party employees. Therefore, we cannot guarantee that the target employees had no perception bias against their peers. Future studies would benefit from examining and replicating our models using experimental designs or the diary method. Employees could record their workplace behavior, such as whether they are abused by others, treat others badly, or witness abusive supervision. Researchers could cross-check the diaries and observe the frequencies of workplace mistreatment to have a more objective scoring of harassment and ostracism. By doing so, researchers may ascertain whether an individual has gradually become more hostile to others—a learning process—in the context of abusive supervision.

We relied on the social learning mechanism to explain the plausible process of third-party employees’ harassment and ostracism via an abusive supervisor, but alternative explanations for our hypotheses may also exist. One example is the social exchange theory (Blau 2017), the core of which is the principle of reciprocity. That theory suggests that employees conduct either good or bad behaviors to reciprocate those who engage in good or bad behaviors (Eissa et al. 2019). Thus, we may also assume that co-workers engage in peer harassment or peer ostracism of the target employee as a gesture to reciprocate their relationship with the supervisor, because the supervisor probably encourages peer mistreatment. Researchers could further explore this possibility.

Moreover, we encourage future research to examine other plausible moderating variables that may strengthen or weaken the abusive supervision–peer mistreatment relationship and the peer mistreatment–victim’s performance relationship. For example, for the abusive supervision–peer mistreatment relationship, third-party employees may differ in their learning level of harassment and ostracism, because not everyone will have the same evil thoughts toward the victim. Mitchell et al. (2012) showed an emotional mechanism whereby angered third parties might attempt to fight against the abusive supervisor and help the victim, whereas contented third parties may agree with the supervisor and engage in more hostile behaviors against the victim (Qiao et al. 2019). Thus, third-party employees’ emotional reactions may be a factor in deciding their actions toward the victim employee. Besides emotions, third-party employees’ personality or moral identity may also moderate the learning outcome. For example, psychopathy, a personality disorder, promotes antisocial behavior and peer mistreatment (Khan et al. 2019; Hurst et al. 2019). Witnessing abusive supervision and having psychopathic characteristics may have a joint effect in third parties’ learning process. Questions include: What contextual factors may reinforce the learning mechanism between abusive supervision and third-party employees’ vicious behavior? Do third-party employees differ in their choice of whether to harass or to ostracize the target? Because third party employees’ choice of harassment and ostracism behavior also represents their directions in treating the victim—that of moving toward or moving away from abusing the victim employee, respective—their directions of motion indicate approach and avoidance tendencies (Ferris et al. 2016). Accordingly, the framework of approach–avoidance framework will be useful to understand the comprehensive learning process of different peer mistreatment constructs. Furthermore, what are the underlying mechanism through which peer mistreatment inhibits the target employee’s performance? Because a negative environment impacts largely on one’s psychological status (Wu et al. 2012), the target employee’s self-worth (Jian et al. 2012) and organization-based self-esteem (Farh and Chen 2014) are both possible mediating factors in the peer mistreatment–victim’s performance relationship. We encourage researchers to explore these questions and the boundary conditions related to peer mistreatment with empirical and/or vignette studies.

Last but not least, we tested the hypothesized model in China. This raises the concern that the findings may not be generalizable to Western countries, because interpersonal relations in China are greatly influenced by traditional cultures. Supervisor–employee relationships are likely to generate hierarchies of superiority and inferiority (Cheng 1995; Liu et al. 2016) that cultivate a hotbed of abusive supervision practices (Cheng 1995; Liu et al. 2016). Holding traditional values, most Chinese tend to be submissive to authority figures, and this could potentially strengthen individuals’ learning tendency toward authority figures. Therefore, we encourage researchers to test the generalizability of our findings by conducting cross-cultural studies.

Conclusion

Building on social learning theory, this study proposes a spillover pathway via third-party employees to the relationship between abusive supervision and the victim employee’s creative performance and job performance. We found support for the contention that third-party employees emulate abusive supervisors’ behaviors and engage in harassment and ostracism toward victims. These behaviors subsequently harm the target employee’s creative performance and job performance. Also, we showed that an individual difference (i.e., proactive personality) has a worsening impact on the relationship between peer mistreatment and creative performance. We call for future studies to continue exploring the effect of abusive supervision on peer mistreatment behavior from a social learning perspective, taking individual differences into account.

Note: We thank the anonymous reviewer very much for insightful suggestions of collecting new data to revalidate the result of H6 and adding a short discussion to highlight the opposite finding.