One of the primary concerns for organizational leaders is to provide workers with a productive and ethical work environment. Despite scholarly research and practitioners’ concerted efforts, organizations continue to struggle with ethical issues stemming from interpersonal mistreatment in organizations. Currently, workplace bullying (LaVan and Martin 2008) is a phenomenon generating a great deal of scholarly research, practical distress, and legal action. Workplace bullying occurs when individuals perceive that they have been the target of undesirable behaviors over a period of time that were difficult to defend against (Einarsen and Skogstad 1996). Bullying behaviors can occur in many forms, including teasing coworkers, gossiping about coworkers, excluding coworkers from activities, and reminding coworkers of past mistakes (Einarsen et al. 2003; Lutgen-Sandvik et al. 2007).

Bullying behaviors are particularly troublesome for organizations because they universally harm employees’ well-being and adversely impact job outcomes (Harvey et al. 2009; Nielsen and Einarsen 2012; Vega and Comer 2005). Meta-analytic evidence indicates that about 15 % of employees in the United States report being the targets of bullying (Nielsen et al. 2010), whereas less than 1 % of employees in the United States identify themselves as perpetrators of bullying (Namie 2014). The discrepancy between employees who reported being the victims of bullying (i.e., about 15 %) and those who report being culprits of bullying (i.e., less than 1 %) motivated us to learn more about employees’ individual differences and the features of the work environment that contribute to workplace bullying.

Despite the growing scholarly and practical interest in understanding and preventing workplace bullying, there are still several substantial gaps in our knowledge of this important phenomenon. Of particular importance, much of the extant research on workplace bullying has examined bullying from the victim’s perspective (Namie and Namie 2009), which has resulted in a lack of accumulation of knowledge regarding the individual characteristics and environmental factors that influence perpetrators of bullying (Sperry 2009). This is surprising because individual differences can fundamentally alter how employees perceive, process, and experience workplace interactions (George 1992). Thus, additional research is needed to explore how employees’ individual differences influence their tendencies to perceive and respond to interpersonal mistreatment in their work environments.

We fill the aforementioned gaps in knowledge by drawing from the tenets of social exchange and displaced aggression theories to explain why coworker bullying is driven by entitled employees’ perceptions of abusive supervision and a lack of felt accountability in their organizations. Specifically, we argue that employees who perceive they were the victims of interpersonal mistreatment (i.e., perceived abusive supervision) likely become the culprits of interpersonal mistreatment (i.e., coworker bullying). We focus on subordinates’ perceptions of social exchange relationships with their supervisors as the key mechanism that drives subordinates’ social exchange perceptions in the workplace because supervisors are key organizational representatives whose actions are reinforced by the organization and the people within it (Eisenberger et al. 2010). Thus, we contribute to recent research that examines the antecedents of workplace bullying by examining a novel multifoci social exchange perspective of bullying that incorporates subordinates’ individual differences (i.e., entitlement), perceptions of the work environment (i.e., felt accountability), and perceptions of supervisory treatment (i.e., perceptions of abusive supervision) into a hypothesized model that examines the antecedents of coworker bullying.

Also, we make an empirical contribution to abusive supervision research by examining coworker bullying as a behavioral outcome of perceptions of abusive supervision. Although numerous abusive supervision studies have examined various forms of interpersonal mistreatment (e.g., coworker aggression, coworker deviance; Martinko et al. 2013; Tepper 2007), prior abusive supervision studies have not directly examined the relationship between perceptions of abusive supervision and subordinates’ tendencies to bully their coworkers. Finally, we draw from the tenets of social exchange and displaced aggression theories in order to theoretically explain why employees’ individual differences and perceptions of the work environment interact to affect perceptions and perpetration of interpersonal mistreatment. Ultimately, we extend interpersonal mistreatment research by examining whether some employees are victims of perceived supervisory abuse and culprits of coworker bullying.

Theoretical Foundations

Social Exchange and Displaced Aggression Theories

Social exchange theory (Adams 1965; Blau 1964) explains that relationships develop between parties over time based on exchange relations with rules and norms that guide exchange processes (Cropanzano and Mitchell 2005). Exchange relations refer to informal expectations between parties that are founded upon trusting one another with regard to the level of benefits received (Parzefall and Salin 2010). Generally, employees respond to favorable work conditions and environments with favorable attitudes and behaviors (e.g., high levels of performance) and respond to unfavorable treatment with unfavorable adjustments in attitudes and behaviors (e.g., aggression; Robinson 2008). Researchers have used social exchange theory in prior abusive supervision research (e.g., Avey et al. 2015) to argue that a state of perceived imbalance is created when employees perceive adverse social exchange processes with their supervisors.

Likewise, displaced aggression theory (i.e., the frustration–aggression hypothesis; Dollard et al. 1939; Marcus-Newhall et al. 2000) has been used to explain the cognitive and behavioral outcomes resulting from perceptions of supervisory mistreatment (e.g., Mitchell and Ambrose 2012). Displaced aggression theory explains that employees who engage in acts of aggression may not aggress against the source of their frustration (e.g., supervisors) because of social norms or fear of retaliation, but instead may aggress against convenient targets (e.g., coworkers) who are less able or likely to retaliate than supervisors. Thus, we draw from social exchange to argue that employees who perceive abusive supervision respond with unfavorable adjustments to their attitudes and behaviors, and draw from displaced aggression theory to argue that employees are likely to bully convenient targets (e.g., coworkers) who are less able to retaliate than supervisors. In the following section, we explain why entitlement indirectly affects employees’ perpetration of bullying behaviors through abusive supervision, conditional upon levels of felt accountability.

Entitlement and Perceptions of Abusive Supervision

Entitlement is the stable belief that an individual deserves and should receive more desirable treatment than others (O’Leary-Kelly et al. 2016), with little consideration given to actual deservingness (Campbell et al. 2004). A foundational premise within entitlement beliefs is the expectation of unequal reciprocity in social exchanges (Naumann et al. 2002). Employees’ levels of entitlement are positively associated with expecting special privileges, beliefs that they do not have to conform to social demands (Raskin and Terry 1988), and tendencies to experience unmet expectations and frustration (Snow et al. 2001).

Entitlement researchers argue that employees who possess a strong sense of entitlement have overly optimistic views of the world and themselves, avoid feedback that challenges their views (Snow et al. 2001), make undesirable interpersonal judgments (Levine 2005), experience personal dissatisfaction, exhibit overly self-centered behavior (Rosenthal and Pittinsky 2006), use destructive responses to criticism (Campbell et al. 2004), and experience interpersonal conflict with their supervisors (Harvey and Martinko 2009). Thus, entitlement is associated with imbalanced reciprocity perceptions that contribute to perceptions of mistreatment from supervisors (e.g., perceptions of abusive supervision) when social exchange expectations are unmet.

Abusive supervision is defined as “subordinates’ perceptions of the extent to which supervisors engage in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact” (Tepper 2000, p. 178). As such, subordinates’ perceptions of abusive supervision are subject to subordinates’ subjective perceptual biases (e.g., romance of leadership; Mathisen et al. 2011). Entitlement is positively associated with perceptions of abusive supervision due to entitlement priming subordinates to make unmet social exchange expectations salient (Harvey et al. 2014). For example, entitled employees might feel that they have been lied to or not given the credit/rewards they deserve by their supervisors when their supervisors do not praise them for simply fulfilling assigned duties adequately. Consistent with social exchange theory and prior abusive supervision research (e.g., Harvey et al. 2014), we argue that employees’ entitlement is positively associated with perceptions of abusive supervision. Therefore, we hypothesize that:

Hypothesis 1

Employees’ entitlement will be positively associated with perceptions of abusive supervision.

Felt Accountability

Behavioral responses associated with perceptions of abusive supervision likely are influenced by the degree to which subordinates feel accountable to their supervisors and organizations. Felt accountability is a subjective experience (Frink et al. 2008) defined as the degree to which employees feel they need to explain and/or justify their behavior to their supervisors (Hall et al. 2015). Accountability researchers have argued that there is no optimal degree of felt accountability (Ammeter et al. 2004), but that individual differences (e.g., entitlement; Frink et al. 2008) play a key role in how felt accountability affects workplace perceptions and behaviors.

Felt accountability can be detrimental to employees’ perceptions and behaviors when supervisors’ behaviors are inconsistent across time and contexts. Subordinates who know that they will be held accountable (i.e., experience higher felt accountability) assume and expect that their behavior will be closely scrutinized and questioned (Ferris et al. 1995). Alternatively, subordinates who are inconsistently held accountable for their behavior and outcomes (i.e., experience lower felt accountability) are unsure of how or when they will be asked by their supervisors for justification for their behaviors. The resulting ambiguity, feelings of insecurity, and low trust likely result in heightened critical assessments of supervisors (e.g., perceptions of abusive supervision).

We argue that subordinates’ entitlement and felt accountability interact to influence the extent to which subordinates perceive abusive supervision, primarily due to the restraining effect of felt accountability on social exchange expectations biased by entitlement. Specifically, higher levels of felt accountability may limit employees’ biased social exchange expectations by making salient the standards against which they are being judged. Alternatively, employees who perceive lower levels of felt accountability likely perceive supervisor attempts to hold them accountable as even less justified and more harmful in nature compared to subordinates who perceive higher levels of felt accountability due to the inconsistency in social exchange interactions inherent in lower levels of felt accountability. Thus, levels of felt accountability likely alter the magnitude of the relationship between subordinates’ entitlement and their perceptions of abusive supervision. Therefore, we hypothesize that:

Hypothesis 2

Employees’ felt accountability will moderate the relationship between subordinates’ entitlement and perceptions of abusive supervision such that the positive relationship between entitlement and perceptions of abusive supervision will be stronger for employees with lower levels of felt accountability than employees with higher levels of felt accountability.

Workplace Bullying

Recently, workplace bullying research has shifted its focus from examining the outcomes of workplace bullying to examining its antecedents. For example, recent research has demonstrated that conflict in the workplace is an antecedent to workplace bullying (e.g., Baillien et al. 2016). If the conflict stems from their supervisors, subordinates may find it difficult to defend themselves from their supervisors’ actions, and are usually less likely to initiate adverse social exchange processes with supervisors than coworkers (Lord 1998). Rather, consistent with displaced aggression theory, employees are more likely to displace their adverse social exchanges on targets who find it difficult to defend themselves (e.g., coworkers) than retaliate against initial aggressors who have power over them (e.g., supervisors). Additionally, most employees will have more coworkers and a greater frequency of social exchanges with coworkers than supervisors, so they may choose to consistently target the same coworkers with mistreatment depending on the quantity and quality of the social exchange relationships they share with their coworkers (i.e., triggered displaced aggression; Miller et al. 2003), which likely is the case when employees choose targets for their bullying behaviors.

Ultimately, drawing from the tenets of social exchange and displaced aggression theories explains why employees who perceive abusive supervision may direct their responses from these adverse social exchanges with supervisors toward their coworkers. Specifically, coworkers are more convenient targets who employees interact with more frequently and who are less likely to retaliate or exercise power than supervisors. Based on prior research that has established relationships between perceptions of abusive supervision and dysfunctional subordinate behaviors (e.g., Nielsen and Einarsen 2012), we expect a positive association between perceptions of abusive supervision and subordinates’ bullying behaviors toward coworkers.

Hypothesis 3

Perceptions of abusive supervision will be positively associated with coworker bullying.

Prior research has established that individual differences influence social perception and behavior (e.g., Judge and Ilies 2002). Consistent with social exchange theory, employees’ entitlement and felt accountability may interactively combine to influence perceptions of social exchange imbalances between subordinates and their supervisors with regard to what each party contributes and receives in their relationships. Perceptions of imbalance may manifest as perceptions of abusive supervision if employees perceive that they are contributing more to the relationship with their supervisors than their supervisors provide in return. Thus, adverse social exchange relationships with supervisors may be associated with employees’ attempts to displace their aggression onto less powerful targets in their work environments (i.e., coworkers). In summary, we argue that subordinates’ entitlement indirectly affects tendencies to bully coworkers through perceptions of abusive supervision, and that felt accountability serves as a boundary condition that can exacerbate or mitigate this relationship.

Hypothesis 4

Employees’ entitlement will be positively and indirectly associated with tendencies to bullying coworkers through perceptions of abusive supervision and conditional upon levels of felt accountability, such that the conditional indirect effect of entitlement on coworker bullying will be stronger for employees with lower levels of felt accountability than employees with higher levels of felt accountability.

Plan of the Research

We used a two-sample constructive replication study design to examine the conditional indirect effect of employees’ entitlement on their tendencies to bully their coworkers through perceptions of abusive supervision, conditional upon levels of felt accountability. We used different sampling procedures across two samples in order to provide stronger evidence of the validity of the obtained results than possible through the use of a single sampling procedure (Wright and Sweeney 2016). Specifically, we collected cross-sectional data (i.e., single-source, self-report data collected at one point in time) for Sample 1 and data collected at two time periods for Sample 2.

Method

Sample and Procedure

Sample 1

During an academic semester, 396 working adults completed an online survey. Undergraduate business students at a university in the southeastern United States were given extra course credit to e-mail links to an online survey to adults they knew working at least part-time (i.e., 20 or more hours per week). Students were provided extra course credit for each e-mail they sent, up to a maximum of three emails. We used a student-recruited sample (Hochwarter 2014) in order to increase the generalizability of study findings with a heterogeneous sample of demographically diverse respondents (Demerouti and Rispens 2014). Student-recruited samples have been used to successfully obtain heterogeneous samples with generalizable results in prior abusive supervision and interpersonal mistreatment research (e.g., Wang et al. 2015).

All respondents provided informed consent prior to completing the survey. Respondents who indicated that they spoke English, resided in the United States, were at least 18 years old, worked for at least 20 h per week, responded from an independent IP address, and who exerted sufficient effort while completing surveys (i.e., spent less than 30 min to complete each survey and chose the correct response for instructed items) were included in the final sample. About 62 % of respondents provided complete data and met the inclusion criteria. Respondents’ average age was 44.15 years old (σ = 12.90), average organizational tenure was 3.5 years (σ = 1.4), and about 53 % of the sample was female.

Sample 2

We constructively replicated the sampling procedure used in the first sample by collecting data over two time periods via the student-recruited sampling technique described above. Specifically, in a data collection process completely independent of Sample 1, we invited respondents to complete two surveys approximately 3 weeks apart. This time frame has been successfully used in prior abusive supervision and interpersonal mistreatment research (e.g., Wang et al. 2015) to reduce the effects of common method bias (CMB) without excessive respondent attrition, which is a concern for respondents who report high levels of perceptions of abusive supervision (Tepper 2000). Initially, 963 respondents completed the first survey. Of the 963 respondents, 705 met the inclusion criteria described above. Then, 178 of the respondents who completed the first survey responded to the second survey. Respondents were matched based on their e-mail addresses, IP addresses, and codes they provided that included the initials for their first and last names. Of the 178 respondents at Time 2, 123 met the inclusion criteria used for this study. Thus, about 13 % of respondents provided complete data and met the inclusion criteria. All respondents provided informed consent prior to completing each survey.

Respondents’ average age was 48.15 years old (σ = 10.26), average organizational tenure was 10.75 years (σ = 9.49), average working week was 46.78 h (σ = 9.69), and about 59 percent of the sample was female. The sample respondents reported working in a variety of job functions (e.g., 23.6 % management, 14.6 % sales), organizational levels (e.g., 27.6 % middle management, 20.3 % staff/associate level), and industries (e.g., 30.9 % educational, health, and social services, 10.6 % finance, insurance, real estate, rental, and leasing). Also, the respondents reported varying degrees of highest levels of education obtained (e.g., 35.8 % Bachelor’s degree, 27.6 % Master’s degree). Thus, the student-recruited sampling technique resulted in a heterogeneous sample of demographically diverse respondents.

The results of independent samples t tests (Schwab 1999) demonstrated that respondents included in the final sample (n = 123) did not significantly differ from respondents who provided full information for both surveys but did not meet the inclusion criteria for the study (n = 55) with regard to mean reported level of any variable. Thus, non-respondent bias likely did not pose a threat to the validity of the obtained results.

We incorporated numerous procedural remedies recommended by prior research (Podsakoff et al. 2003, 2012) into the study design for Sample 2 in order to strengthen the study design and confidence in the findings, as well as to limit concerns stemming from CMB before, during, and after the data collection process (Aguinis and Edwards 2014; Aguinis and Vandenberg 2014). Specifically, we procedurally addressed CMB concerns by altering the number of scale points and response formats between constructs, protecting respondents’ anonymity, limiting respondent fatigue by designing the surveys to be brief, concealing the true purpose of the study by including short measures of constructs not examined in this study, including instructed items to ensure that study participants exerted an appropriate amount of cognitive effort while filling out the surveys, and counterbalancing question order in order to avoid priming effects (Desimone et al. 2015; Johnson et al. 2011; Meade and Craig 2012; Podsakoff et al. 2003, 2012).

Measures

All measures were created such that higher scores reflected greater levels of each construct than lower scores. For Sample 2, entitlement, felt accountability, and perceptions of abusive supervision were collected at Time 1, whereas coworker bullying was collected at Time 2.

Entitlement

Entitlement was measured using Campbell et al.’s (2004) nine-item measure (α Sample 1 = .90; α Sample 2 = .81). Sample items included “I deserve more things in my life” and “Things should go my way.” One item was originally negatively worded, so we reverse-scored that item prior to aggregating items. Responses were recorded on a seven-point agreement scale (1 = “strongly disagree,” 7 = “strongly agree”).

Felt Accountability

Felt accountability was measured using Hall et al.’s (2003) eight-item measure (α Sample 1 = .78; α Sample 2 = .75). Sample items included “I often have to explain why I do certain things at work” and “I am held accountable for my actions at work.” Responses were recorded on a seven-point agreement scale (1 = “strongly disagree,” 7 = “strongly agree”).

Perceptions of Abusive Supervision

Perceptions of abusive supervision were measured using Tepper’s (2000) 15-item measure (α Sample 1 = .95; α Sample 2 = .94). Sample items included “Ridicules me” and “Is rude to me.” Responses to the statement “My boss…” were recorded on a 5-point frequency scale (1 = “I cannot remember him/her using this behavior with me”, 5 = “He/she uses this behavior very often with me”).

Coworker Bullying

Coworker bullying was measured by adapting ten items from the person-related bullying dimension of the Negative Acts Questionnaire-Revised (NAQ-R; Einarsen et al. 2009) in order to measure bullying from the perpetrator’s perspective (α Sample 1 = .87; α Sample 2 = .80). We chose the person-related dimension of the NAQ-R because we explicitly examine interpersonal mistreatment in the hypothesized model. Ultimately, the results obtained from the coworker bullying measure we used likely are nearly identical to the results we would have obtained using the full NAQ-R measure because the person- and work-related bullying behaviors’ dimensions are so highly correlated (e.g., r = .96, Einarsen et al. 2009). Example items were “I sometimes make fun of others about their work” and “I enjoy hearing gossip and sharing rumors about others.” A full list of the coworker bullying items is shown in the “Appendix” section. Responses were recorded on a seven-point agreement scale (1 = “strongly disagree,” 7 = “strongly agree”).

Analysis

We used PASW/SPSS 23.0 to examine means, standard deviations, and intercorrelations for study variables, as well as to run Model 8 of Hayes’ (2013) PROCESS procedure for estimating mediated moderation effects (http://afhayes.com/spss-sas-and-mplus-macros-and-code.html). Hayes’ procedure examines the conditional indirect effects used to explain how and when an effect exists. Hayes’ approach uses bootstrapping (n = 5000) to test results in a single step, which is a stronger approach than Baron and Kenny’s (1986) piecemeal approach for testing mediation (Edwards and Lambert 2007; Preacher et al. 2007). We report unstandardized effect sizes and standardized effect sizes, which were calculated by multiplying unstandardized effect sizes by the quotient of the standard deviations for the predictor (X) variable and the outcome (Y) variable (i.e., B × [σ X /σ Y ]). Independent variables (i.e., entitlement and felt accountability) were mean-centered when computing interaction terms (Aiken and West 1991), but, to aid interpretation, were not mean-centered when creating the interaction graphs. We used hierarchical moderated multiple regression (Cohen et al. 2003) to confirm the obtained results.

We hypothesized that employees’ entitlement affects employees’ bullying behaviors toward coworkers through perceptions of abusive supervision, conditional upon employees’ levels of felt accountability. Thus, we estimated the following equations:

$${\text{AS }} = a_{0} + a_{ 1} E \, + a_{ 2} {\text{FA }} + a_{ 3} E \, \times {\text{ FA }} + r,$$
(1)
$${\text{CB }} = b_{0} + b_{ 1} {\text{AS}} + E \, + {\text{FA }} + E \, \times {\text{ FA }} + r.$$
(2)

The first equation enables us to estimate various paths of interest, including a 1 [i.e., the strength of employees’ entitlement (E) on perceptions of abusive supervision (AS)], a 2 [i.e., the strength of felt accountability (FA) on perceptions of abusive supervision (AS)], a 3 (i.e., the strength of the entitlement × felt accountability interaction on perceptions of abusive supervision), and r (i.e., the error term). The second equation estimates the strength of the path from perceptions of abusive supervision (AS) to coworker bullying (CB) while controlling for the effect of entitlement, felt accountability, and the entitlement × felt accountability interaction. First, \(c^{\prime}_{1}\) represents the direct effect of entitlement on coworker bullying. Next, \(c^{\prime}_{2}\) refers to the strength of the felt accountability-coworker bullying path. Then, \(c^{\prime}_{3}\) reflects the strength of the interaction of entitlement and felt accountability on coworker bullying. Finally, r is included as the error term. By estimating these equations, we were able to assess whether the effect of entitlement (E) on coworker bullying (CB) through perceptions of abusive supervision (i.e., AS; the mediator) was conditional on felt accountability (i.e., FA; the moderator).

Results

Measurement Model Results

Confirmatory Factor Analyses

Confirmatory factor analyses (i.e., CFAs) were conducted using the maximum likelihood estimation method in AMOS 23.0 (Arbuckle 2005) to examine the distinctiveness of the study variables. The measurement model consisted of the four primary variables of interest: entitlement, felt accountability, perceptions of abusive supervision, and coworker bullying. Due to the number of items measured for some of the constructs (e.g., 15 items were used to measure perceptions of abusive supervision), we used partial disaggregation (i.e., parceling) techniques when conducting the CFAs and exploratory factor analyses (EFAs) described below. We used the factorial algorithm technique (i.e., item-to-construct balance; Little et al. 2002; Rogers and Schmitt 2004) to create parcels for the four variables of interest. This technique balances the best and worst items across parcels in order to equally balance parcels with regard to their difficulty and discrimination (Williams et al. 2009). Parceling techniques are appropriate for unidimensional constructs with non-normally distributed item-level data (Bandalos 2002), such as perceptions of abusive supervision and coworker bullying.

Typically, root-mean-square error of approximation (RMSEA) and standardized root-mean-square residual (SRMR) scores below .10 (Hoyle and Panter 1995), as well as comparative fit index (CFI) and Tucker–Lewis index (TLI) scores above .90 (Bentler and Bonett 1980; Bollen 1989) indicate acceptable fit (Hu and Bentler 1999). The results indicated acceptable model fit to the data (Sample 1: χ 2 (48) = 73.99, p < .01, CFI = .99, TLI = .99, RMSEA = .04, and SRMR = .03; Sample 2: χ 2 (48) = 92.42, p < .01, CFI = .95, TLI = .93, RMSEA = .09, SRMR = .08).

Further, the CFAs indicated that acceptable levels of convergent validity (Fornell and Larcker 1981) were present because all parcels loaded significantly on the intended constructs and demonstrated standardized item loadings greater than .5. Also, there were acceptable levels of discriminant validity (Bagozzi et al. 1991) present because all correlations were significantly different from 1 (i.e., the highest correlation between non-demographic variables was r = .37 between perceptions of abusive supervision and coworker bullying in Sample 1). Finally, there were acceptable levels of internal consistency because all estimates of Cronbach’s alpha (α) were above Nunnally’s (1978) standard of .70.

Supplementary Factor Analyses

Next, we examined the results of hierarchically nested covariance structure analyses (Cote and Buckley 1987) in order to examine the degree to which CMB was present in the data for the four substantive variables in the model. The results indicated that more of the variance was explained by the trait factor (i.e., Sample 1: 73.49 %; Sample 2: 62.35 %) than the method factor (Sample 1: 3.05 %; Sample 2: 12.91 %) or random error (Sample 1: 23.46 %; Sample 2 = 24.74 %). We concluded that CMB likely did not present a threat to the interpretability of the results or the validity of the inferences drawn from the study.

Prior research (e.g., Martinko et al. 2014) recommends conducting alternative CFA models and EFAs in order to provide evidence of construct independence. Thus, we conducted alternative CFAs that loaded all the parcels used for the CFAs onto one latent factor. The results indicated that the model fit decreased dramatically (Sample 1: χ 2 (54) = 448.51, p < .01, CFI = .47, TLI = .36, RMSEA = .25, SRMR = .21; Sample 2: χ 2 (54) = 425.67, p < .01, CFI = .56, TLI = .46, RMSEA = .24, SRMR = .20), which provided evidence of construct independence. Then, we ran EFAs using principal axis factoring and promax rotation with Kaiser normalization to examine the parcels used for the CFAs. We found four independent factors with Eigenvalues above 1.00 (Field 2005) that each accounted for the parcels associated with one of the substantive variables. Thus, the results of alternative CFA models and EFAs provided additional evidence of construct independence.

Descriptive Statistics and Correlations

Means, standard deviations, and zero-order bivariate correlations are presented in Table 1. The zero-order bivariate correlations were in the expected directions and of the approximately expected magnitudes. Of particular interest, the means for perceptions of abusive supervision (i.e., \(\bar{x}_{{{\text{Sample}}\;1}}\) = 1.48; \(\bar{x}_{{{\text{Sample}}\;2}}\) = 1.39) were low, which is consistent with meta-analytic evidence that abusive supervision is a low base-rate phenomenon (Mackey et al. 2015). Variance inflation factor (VIF) scores for predictive variables were all below 1.16 in both samples, which is considerably below the acceptable standard of 10 or less needed to demonstrate that multicollinearity likely did not substantively affect the study’s findings (Montgomery et al. 2001; Ryan 1997).

Table 1 Descriptive statistics and zero-order bivariate correlations for Samples 1 and 2
Full size table

Hypothesized Model

We argued that employees’ entitlement had an indirect effect on coworker bullying through perceptions of abusive supervision, and that this relationship was conditional upon employees’ levels of felt accountability. The results of the mediated moderation analysis are presented in Table 2, as well as visually depicted in Fig. 1. Entitlement significantly predicted perceptions of abusive supervision in Sample 1, but not Sample 2 (Sample 1: B = .11, β = .18, p < .01; Sample 2: B = .06, β = .11, ns). Thus, Hypothesis 1 was partially supported. The entitlement × felt accountability interaction term significantly predicted perceptions of abusive supervision in both samples (Sample 1: B = −.09, β = −.15, p < .01; Sample 2: B = −.14, β = −.19, p < .05). Thus, Hypothesis 2 was supported.

Table 2 Regression results for Samples 1 and 2
Full size table
Fig. 1
figure 1

Hypothesized model and results of model estimation. N Sample 1 = 396, N Sample 2 = 123. All estimates reflect standardized effect sizes. The effect sizes for control variables were omitted for clarity. *p < .05; **p < .01

Full size image

We used the two-way plotter with all options available on Jeremy Dawson’s interaction effects website (i.e., http://www.jeremydawson.co.uk/slopes.htm) to plot the interaction effect at lower (i.e., one standard deviation below the mean) and higher (i.e., one standard deviation above the mean) levels of felt accountability across the range of scores for perceptions of abusive supervision (Stone-Romero and Liakhovitski 2002). The interaction plots are shown in Figs. 2 and 3. The plots depict that increases in entitlement are associated with increases in perceptions of abusive supervision, but that this effect is stronger for employees who reported lower levels of felt accountability than employees who reported higher levels of felt accountability. The results of simple slopes tests provided by the interaction plotter demonstrated that the lower felt accountability slope was significant in Sample 1 (Sample 1: b = .19, t = 2.59, p < .05), but not Sample 2 (Sample 2: b = .18, t = 1.35, ns). Additionally, simple slopes tests demonstrated that the higher felt accountability slopes were not significant in either sample (Sample 1: b = .03, t = .31, ns; Sample 2: b = −.06, t = −.35, ns). Overall, the interaction effects (see Figs. 2, 3) were significant and in the hypothesized direction.

Fig. 2
figure 2

Felt accountability moderating the relationship between entitlement and perceptions of abusive supervision in Sample 1

Full size image
Fig. 3
figure 3

Felt accountability moderating the relationship between entitlement and perceptions of abusive supervision in Sample 2

Full size image

Next, perceptions of abusive supervision (Sample 1: B = .59, β = .35, p < .01; Sample 2: B = .38, β = .26, p < .01) were positively associated with coworker bullying. Thus, Hypothesis 3 was supported. Finally, the indirect effect of entitlement on coworker bullying through perceptions of abusive supervision, conditional upon felt accountability, was significant in Sample 1 (B = .06, β = .10, p < .05), but not Sample 2 (B = .02, β = .03, ns) for the mean level of felt accountability. However, the predicted pattern of results was obtained because the conditional indirect effect at lower levels of felt accountability was significant (Sample 1: B = .11, β = .19, p < .05; Sample 2: B = .07, β = .12, p < .05), whereas the conditional indirect effect at higher levels of felt accountability was not significant (Sample 1: B = .02, β = .03, ns; Sample 2: B = −.02, β = −.03, ns). Thus, Hypothesis 4 was partially supported.

Discussion

Our findings provide support for the hypothesized mediated moderation model, which demonstrate that not only is there an indirect relationship between entitlement and coworker bullying, but that this relationship occurs through perceptions of abusive supervision and is stronger for employees who report lower levels of felt accountability than employees who report higher levels of felt accountability. The findings provide empirical support for conceptual research (e.g., Samnani and Singh 2015) that argues that there is a complex relationship between employees’ individual differences (i.e., entitlement), perceptions of their environment (i.e., felt accountability), and perceptions of supervisory treatment (i.e., perceptions of abusive supervision) in predicting their bullying behaviors directed toward coworkers.

This study contributes to abusive supervision and bullying research in at least two ways. First, little research has empirically examined the role of employees’ felt accountability in interpersonal mistreatment research, or how felt accountability may serve as a boundary condition in the relationship between individual differences (e.g., entitlement) and interpersonal mistreatment perceptions and behaviors. Thus, this study makes a novel contribution that reconciles inconsistencies in prior research regarding which employees are most likely to be victims of perceived abusive supervision and culprits of coworker bullying. Second, this study adds value to the bullying literature by examining the characteristics that are associated with bullies, rather than solely or primarily focusing on perceptions of being bullied. This study draws attention to the lack of research examining the characteristics of individuals who bully others and corroborates prior research that has found positive associations between employees’ perceptions of being mistreated and tendencies to mistreat others via displaced aggression. Finally, we applied a novel extension of social exchange and displaced aggression theories to develop the research model examined in this study.

Strengths and Limitations

There are several strengths of the present study. First, the large sample size for Sample 1 (n = 396) provided ample power to detect relationships in the hypothesized model. Further, the time-separated data examined in Sample 2 helped alleviate some concerns of CMB. Next, both samples of data examined in this study potentially allow for broad generalizations of study results because of the sampling technique, which resulted in a demographically diverse, heterogeneous sample of employee respondents (Demerouti and Rispens 2014).

This study also has a few limitations that warrant discussion in order to appropriately contextualize the inferences drawn from our study (Brutus et al. 2013). First, this study relied on the use of self-reported data, which may increase the likelihood of CMB (Johnson et al. 2011). We limited the effects of CMB by following procedures recommended for survey design (e.g., protecting the anonymity of respondents, counterbalancing question order in order to avoid priming effects, varying the number and type of response scale points, including instructed items; Podsakoff et al. 2003, 2012). The use of self-reports was necessary for our study because employees themselves are the most qualified individuals to evaluate and report their perceptions of workplace environments (i.e., felt accountability) and social exchange relationships in the workplace (i.e., perceptions of abusive supervision and coworker bullying behaviors), especially because bullying behaviors can be covert and not readily or consistently witnessed by others. Thus, it was appropriate to use self-report data to pursue the objectives of this study (Conway and Lance 2010). Regardless, our measurement of perceptions of abusive supervision from subordinates’ perspectives rather than objective assessments of abusive supervision preclude us from drawing inferences regarding how abusive the supervisors in our study actually were.

Importantly, Siemsen et al. (2010, pp. 469–470) stated “researchers should not be criticized for CMV [common method variance] if the main purpose of their study is to establish interaction effects” because “CMV cannot create an artificial interaction effect. CMV can only deflate existing interactions.” Indeed, CMV, which stems from CMB, is not likely to artificially create interaction effects, and in some cases can attenuate actual interaction effects (Evans 1985; Siemsen et al. 2010). Thus, it is not likely that CMV had a strong biasing effect on the results obtained from this study because examination of the interaction between entitlement and felt accountability was a crucial part of the hypothesized model we tested. Finally, the non-experimental data we examined precluded conclusive claims of causality. Thus, we relied on prior research and theory to develop the hypothesized model, but encourage future research to use longitudinal experimental data that can bolster claims of causality.

Directions for Future Research

There are numerous immediate and incremental research opportunities to meaningfully extend our study’s findings in order to advance theory and research (Brutus et al. 2013). First, future research would benefit from examining data from varying sources and across multiple time periods. For example, combining self-reports with some form of objective data (e.g., other reports of supervisory and coworker bullying, number of formal complaints or reprimands, behaviors observed on surveillance cameras) may provide valuable insight to researchers. Objective measures may not capture the full range of employees’ bullying behaviors, but they can provide useful information, enable measurement triangulation, and provide additional evidence of the validity of self-report measures. Also, future research would benefit from examination of other-reported data when assessing employees’ levels of entitlement because entitlement is considered a stable individual difference. Future research could also examine whether employees’ levels of entitlement change over time, and if so, to what extent their social exchange perceptions change accordingly. It is possible that employees who are new to an organization have different social exchange expectations from employees who have established social exchange relationships within their organizations. We encourage future research to explore this possibility.

Future research could continue to explore the association between subordinates’ individual differences (e.g., entitlement), perceptions of the work environment (e.g., felt accountability), adverse social exchange processes with supervisors (e.g., perceptions of abusive supervision), and adverse social exchange processes between coworkers when examining why bullying occurs in the workplace and how cultures of bullying are created and maintained (Samnani 2013). It is possible that abusive supervision creates a culture of bullying that enables and/or encourages subordinates to engage in coworker bullying. We encourage future research to explore this possibility. We tested the hypothesized model with two samples drawn from the United States. We encourage researchers to replicate and extend our findings in samples drawn from different cultures because prior research has demonstrated that different contexts may alter how workplace bullying occurs (e.g., Giorgi et al. 2015).

Finally, we encourage additional research that examines the antecedents of perceptions of abusive supervision and coworker bullying. Zhang and Bednall (2015) meta-analytically demonstrated that supervisor-related antecedents, organization-related antecedents, subordinate-related antecedents, supervisors’ demographic factors, and subordinates’ demographic factors all contribute to subordinates’ perceptions of abusive supervision. A similar investigation of bullying would be useful. Also, abusive supervision and bullying research would both benefit from a nuanced examination of boundary conditions that can reconcile inconsistencies in prior research that explores the antecedents of perceptions of abusive supervision and coworker bullying. For example, how do strong versus weak ethical organizational climates affect perceptions of abusive supervision and perpetration of coworker bullying? Are ethical organizational leaders able to create strong ethical cultures that deter interpersonal mistreatment? We encourage future research that explores these questions and the general role of ethics perceptions on interpersonal mistreatment perceptions and behaviors.

Implications for Practice

This study has several important implications that can inform practice in order to meaningfully bridge the gap between researchers and practitioners. First, organizations that hire entitled employees may have to combat issues associated with perceptions of abusive supervision and coworker bullying if clear accountability mechanisms are not in place. Further, many millennials have recently entered the workplace, and many more millennials will enter the workplace in the near future. Sometimes termed “generation me” or the “entitlement generation,” millennials present a considerable challenge for organizational leaders as they work to understand the impact of millennials’ attitudes and behaviors on the workplace environment. For example, millennials’ perceptions of interpersonal mistreatment and ethical codes of conduct that affect perpetration of interpersonal mistreatment may differ from employees who belong to different generations. Thus, organizations may want to assess prospective employees’ levels of entitlement. Also, organizational representatives who detect coworker bullying or other interpersonal mistreatment behaviors may be able to prevent the formation of cycles of unethical behaviors (e.g., perceptions of abusive supervision and coworker bullying) by implementing programs designed to monitor interpersonal mistreatment if they wish to control systematic instances of adverse social exchange processes between subordinates and supervisors.

Another important finding in this study was that felt accountability has an effect on how employees perceive their social exchange relationships with supervisors. Thus, organizations likely could benefit from establishing explicit accountability mechanisms for employees and ensuring that they clearly communicate the mechanisms in place to employees. Managerial efforts to provide detailed expectations and clearly describe accountability mechanisms may help employees establish consistent and realistic expectations for ethical behavioral conduct. For example, organizations can develop and enforce strict policies against interpersonal mistreatment in the workplace, train employees to engage in constructive conflict management techniques with their supervisors, and establish human resource hotlines that can be used to anonymously report mistreatment in the workplace (Sutton 2007). All of these measures likely would help prevent cycles of unethical behavior from becoming reinforced by adverse social exchange processes and acts of displaced aggression over time. Ultimately, such precautionary measures could help support well-functioning organizations with strong ethical leadership that promotes proper workplace environments characterized by minimal bullying behaviors (e.g., Stouten et al. 2010).

Conclusion

Results across two samples of data demonstrate that the conditional indirect effect of entitlement on coworker bullying occurs through perceptions of abusive supervision, and that this effect is stronger for employees who report lower levels of felt accountability than employees who report higher levels of felt accountability. Importantly, this study helps clarify the individual differences (i.e., entitlement), perceptions of the workplace (i.e., felt accountability), and perceptions of supervisory treatment (i.e., perceptions of abusive supervision) associated with the perpetration of coworker bullying. Also, the novel extension of social exchange and displaced aggression theories we presented in this paper provides a broader view of employees’ perceptions of being a victim of perceived abusive supervision and culprit of coworker bullying than offered by prior research. We hope this study stimulates additional research into the role employees’ individual differences and perceptions of their workplace have on employees’ tendencies to perceive and perpetrate interpersonal mistreatment.