Political climate and workplace cyberbullying perpetration: a multilevel moderated mediation model

Abstract

Background

In the post-COVID-19 era, more work is being done online; more people are working remotely; and communication has shifted online. Increased concerns about workplace cyberbullying (WCB) have accompanied these changes in work, but the scholarly literature is limited in terms of understanding the characteristics of organizations and individual perpetrators that shape WCB behavior. Drawing upon social exchange theory and the literature on organizational politics, we proposed and tested a multilevel moderated mediation model to examine the cross-level direct and indirect relationships between political climate and WCB perpetration mediated through psychological contract violation, and we also investigated the moderating role of toxic online disinhibition in this process.

Methods

We collected multiphase and multilevel data from 416 white-collar employees nested within 30 organizations in the service sector in Islamabad, Pakistan. We used a multilevel structural equation modeling (MSEM) technique in Mplus to analyze the data.

Results

We found a direct and positive cross-level relationship between political climate and WCB perpetration. Multilevel mediation analysis revealed that psychological contract violation mediates the cross-level relationship between political climate and WCB perpetration. Moreover, multilevel moderated mediation analysis suggested that the conditional cross-level indirect effect of political climate on WCB perpetration via psychological contract violation was stronger and significant at higher levels of toxic online disinhibition, whereas it was weaker and non-significant at lower levels of toxic online disinhibition.

Conclusion

Studies exploring the situational antecedents of WCB perpetration are scarce, particularly at the organizational level. We proposed and tested a multilevel model of WCB perpetration indicating that political climate leads individuals to engage in WCB perpetration directly as well as indirectly through its impact on psychological contract violation. We outline a number of practical implications and suggest future research directions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-025-03239-1.

Introduction

In the post-COVID-19 era, more work is being done online than ever before; more people are working remotely; communication has shifted online, including through email and virtual meetings; and more organizations have adopted a hybrid workplace model [1]. Higher prevalence rates of workplace cyberbullying (WCB) have accompanied these changes, in part due to the increased use of information and communication technologies as a communication medium [2, 3]. The heavy reliance of organizations and their employees on information and communication technologies has led to the emergence of WCB, particularly after the COVID-19 outbreak [4, 5]. WCB is defined as “all negative acts stemming from working relationships and occurring through the use of ICTs that are either (a) carried out repeatedly and over a period of time or (b) conducted at least once but form an intrusion into one’s personal life, having the potential to expose private information to a wide online audience” [6, p. 29]. Consistent with previous research, we examine WCB perpetration, which comprises behaviors such as sending derogatory comments and gossiping or spreading rumors about someone at work while online, so as to better understand how to prevent this antisocial work behavior [7].

How organizations prevent and mitigate WCB is a significant business ethics problem [7]. WCB is considered an extension of traditional workplace bullying [8]; however, certain characteristics of WCB such as anonymity and privacy, boundarylessness, a possibly wider audience, concreteness, and permanence make it more hazardous compared to traditional workplace bullying [9]. The emerging issue of WCB has recently attracted increased attention from researchers and practitioners because WCB has severe and detrimental implications for individuals and organizations alike [10]. Researchers to date have predominantly focused on the prevalence of WCB [e.g., 11, 12], its degree of overlap with traditional workplace bullying [e.g., 8], its negative attitudinal, behavioral, and health outcomes for victims [e.g., 12, 13–15], and exploring individual and organizational risk factors that increase an individual’s vulnerability to WCB exposure [e.g., 16]. Moreover, there are only a handful of studies that have explored individual and organizational factors that foster WCB perpetration [e.g., 17, 18]. The major shortcoming in the WCB literature is that it has not fully considered the role of characteristics of the work environment, particularly at the organizational level, that might trigger employees to engage in WCB perpetration. There is also a lack of understanding as to why employees engage in WCB perpetration and the individual differences that might exacerbate the likelihood of doing so. This is a problem in terms of limited theoretical development and a lack of precision as to how characteristics of individuals and the work environment interact to shape WCB perpetration. It also limits our practical understanding as to how WCB can be prevented.

The overarching purpose of our study is accordingly to address these problems by examining political climate as a predictor of WCB perpetration, psychological contract violation as an underlying mechanism, and toxic online disinhibition as an individual difference moderator. The purpose of our study ties to a perennial discussion in and limitation of the organizational politics literature. Organizational politics, defined as “actions by individuals which are directed toward the goal of furthering their own self-interest without regard for the well-being of others or their organization” [19, p. 4]] is a pervasive phenomenon in the workplace [20] and an important business ethics concern in that organizational politics is associated with a wide variety of dysfunctional work-related outcomes, such as diminished job satisfaction and organizational commitment, job stress, counterproductive work behaviors, decreased job performance, and higher intentions to leave [e.g., 21, 22]. Organizational politics has been studied almost exclusively at the individual [23] and workgroup levels [24]. Researchers have theorized that organizational politics can be both an individual level perception as well as an organizational level reality [e.g., 23, 25, 26] and have called for more research on organizational politics as an organizational level phenomenon [26, 27]. Given the pervasive nature of organizational politics and the chances that its negative consequences will be witnessed or experienced by many organizational members, it seems reasonable to assume that employees may form a shared, collective political cognition or climate at the organizational level, known as the political climate [24, 28, 29]. We theorize that the experience of working in a highly political organization operationalized as a political climate may result in WCB perpetration.

To address the problems with the WCB and organizational politics literature described above, we theorize that political climate may encourage individuals to engage in WCB perpetration. We also propose that psychological contract violation (a negative emotional state) will mediate the cross-level association between political climate and WCB perpetration. Moreover, we suggest that toxic online disinhibition will act as a boundary condition of the relationship between psychological contract violation and WCB perpetration. More specifically, based on social exchange theory (SET; [30–32]), we propose a multilevel moderated mediation model [e.g., 33, 34], assuming a 2-1-1 mediation in which psychological contract violation at Level 1 mediates the relationship between political climate at Level 2 (organizational level predictor) and WCB perpetration at Level 1 (individual level outcome), with toxic online disinhibition as a second-stage moderator of the relationship between psychological contract violation and WCB perpetration at Level 1 (see Fig. 1).

Fig. 1.

Conceptual model

Our study offers important contributions to theory, research, and practice. Our first contribution is to develop theoretical logic and rationale for predictors of WCB perpetration at the individual and organizational levels, as well as an individual difference moderator thereof. This is important in that we not only investigate what predicts WCB perpetration (i.e., political climate) but also unravel a potential mediating mechanism (i.e., psychological contract violation) and a boundary condition of this process (i.e., toxic online disinhibition). A related contribution is to build on previous scholarship [i.e., 26–28] to provide the conceptual foundation for political climate at the organizational level as a cross-level predictor of WCB perpetration. This can help spur future research on political climate, as well as organizational level predictors of WCB perpetration. Negative exchange patterns, such as those represented in our model, are studied less often in social exchange than positive patterns [35] and they are modeled almost exclusively at the same level of analysis. We contribute to the social exchange literature by examining an important negative exchange pattern between organizations and their employees. Our results offer important practical implications as they identify features of organizations (political climate) and individuals (toxic online disinhibition) that can be managed to reduce the likelihood of WCB perpetration.

Theoretical background and hypotheses

SET is one of the most influential conceptual paradigms for understanding workplace behavior [30]. SET is a theory of sequential transactions between two or more parties [32], wherein the nature of treatment received by one party is reciprocated to the other party [31, 36]. A basic tenet of this theory is that over time relationships evolve into trusting, loyal, and mutual commitments as long as the parties abide by certain rules of exchange [30, 31]. One such exchange rule is that of reciprocity [31], which is an implicit quid pro quo arrangement such that behavior from one party in an exchange relationship is met with an in-kind response from the other party [31, 35]. Scholars have posited that reciprocity represents positive and negative quid pro quo propensities [e.g., 37] such that favorable treatment is reciprocated with favorable treatment and unfair treatment is reciprocated with unfair treatment [e.g., 38].

Negative exchange patterns are studied less often than positive exchange patterns [35]. The negative hedonic model of social exchange [35] is useful for understanding negative exchange patterns. The model describes a series of exchanges between an actor and a target. This process begins with an initiating action wherein the actor commits harm to the target, which is followed by a reciprocating response from the target, i.e., the target reciprocally commits harm to the actor. A low-quality exchange relationship is the link between the initiating action from the actor and the negative reciprocal response from the target. This model maps well to our study. Negative workplace conditions, such as workplace incivility, have been theorized as initiating events that begin a negative social exchange [e.g., 39]. In the present study, we propose that political climate at the organizational level is an important initiating event that predicts WCB perpetration and that this relationship is mediated through psychological contract violation, which itself is an indicator of a negative social exchange [40].

Based on SET about a negative reciprocity norm [e.g., 31, 41] and individual differences that are predictive of this norm [38], we further propose that individual differences in toxic online disinhibition are an important boundary variable in our model. Negative reciprocity is a way to respond to unfair treatment and seek retribution [42]. Social exchange theorists have posited a negative reciprocity norm such that retribution can be viewed as an appropriate response to unfair treatment and that endorsement of this norm is contingent on individual differences [31, 38]. In support of this proposition, research has shown individual differences, such as in dominance and impulsivity, are distinct from the negative norm of reciprocity and that these individual differences predict a negative norm of reciprocity [38]. Since WCB perpetration occurs online, individual differences in toxic online disinhibition are especially relevant in terms of the extent to which as political climate leads to WCB perpetration via psychological contract violation.

Political climate and workplace cyberbullying perpetration

We theorize that the extent to which an organization is highly political, which we define and conceptualize later as political climate, will predict an individual’s reports of engaging in WCB perpetration. This is consistent with the basic tenet of SET that targets of a political climate, namely employees, will reciprocate this unfair treatment in kind [31]. This is also in line with the negative hedonic model of social exchange [35] in that a political climate is an initiating event that results in a negative reciprocating response from the target [39], i.e., in the form of WCB perpetration. It is important to note that we theorize it is political climate and not individual level perceptions of organizational politics that are relevant as an initiating event. In other words, the behavior is characteristic of the organization as an actor. Since employees are not necessarily able to reciprocate this unfair treatment directly at the organization itself, they may do so indirectly through WCB perpetration. As we argue below, WCB is a relatively low-risk way by which employees can reciprocate and receive a feeling of retribution for unfair treatment.

Consistent with our theorizing political climate as an initiating event resulting in WCB perpetration, early papers on organizational politics acknowledged that political behavior can occur at multiple levels, namely the individual, workgroup, and organizational [43], and conceptualized the construct from an organizational level perspective. For instance [44], viewed organizational politics as fitting into the basic conceptualizations of work-related stress. From their perspective, politics can be categorized as an organizational level stressor that leads to “individually-experienced strains” (p. 1207). When organizational politics are discussed in primary studies, the construct is often conceptualized at the organizational level. For instance, Djurdjevic et al. [45] explain that “highly political organizations are characterized by the actions of self-serving individuals who are competitive and inattentive to the needs of others, creating workplaces in which dominant coalitions, conflict, and uncertainty are commonplace” (p. 182). Findings in this literature are used to describe how firm-level decisions are made in political organizations. For instance, personnel decisions, such as those regarding pay and promotions, are based on favoritism rather than on merit [46]. The types of political behavior measured by leading perceptions of organizational politics scales, e.g., the extent to which favoritism not merit gets people ahead [47], are features of organizations and are studied at the organizational level across disciplines [e.g., 48, 49]. Research supports the hypothesis that individuals differentiate between organizational politics occurring at different levels [50].

In their review of the organizational politics literature, Ferris et al. [51] concluded that “although multilevel studies are few, theories of individuals’ perceptions of organizational politics suggest that these phenomena are inherently multi-level in nature” (p. 214). Building on the work by Ferris et al. [51], Dipboye and Foster [26] argued that the literature on perceptions of organizational politics had not gone far enough by way of considering multilevel linkages with organizational politics at the organizational level. They build a case for studying and operationalizing organizational politics at the organizational level as a shared construct that emerges from lower levels of analysis and described “collective or aggregated perceptions of politics” as a feature of “… climate at the group and organizational levels…” (p. 265). Other scholars have posited that organizational political climate is a distinct or functionally specific type of climate [e.g., 25, 27]. Landells and Albrecht [28] built on these ideas to propose that perceptions of organizational politics can be sensibly measured at the organizational level and defined organizational political climate as “shared perceptions’ of practices, policies and procedures specific to organizational politics” (p. 358). In this regard, Drory [27] demonstrated that the majority of the variance in individual climate perceptions resided at the organizational level and concluded that organizational politics can be measured as “an organizational rather than a personal attribute” (p. 66).

When organizational members share such perceptions, a highly political climate emerges and organizational processes suffer as a result [52]. Political climate is a pervasively undesirable work context and a reality of organizational life [53] that captures people’s quest for self-interest [52]. At high levels of political climate, there may be shared perceptions that common practices in the workplace comprise the use of networks to acquire potentially vital information, individuals getting ahead based on who they know, individuals investing substantial time and effort trying to comprehend who can influence decisions, individuals striving to become members of powerful coalitions, and individuals exploiting policy loopholes to benefit themselves at the expense of others [28, 54]. Thus, a work environment is likely to be perceived as political wherein policies and practices are carried out unfairly and where the same rules do not apply equally to everyone [55]. Drory [27] who undertook a seminal venture into political climate research noted that employees who lack the power base and the means of influence needed to take advantage of the political game may perceive organizational politics as a source of frustration. Chang et al. [54] noted that a highly political climate places demands on employees to engage in political behavior to compete for scarce resources. Likewise, Ferris et al. [56] suggested that employees who lack power in a political work environment might choose to participate in such influencing or political behavior as bullying to regain some control over their environment.

Previous research suggests that employees who engage in bullying as an influence tactic, such as in response to a political work climate, may prefer WCB over traditional workplace bullying because the latter requires certain resources (e.g., physical and social power) and not all employees have access to and can use these resources in a highly political climate [7, 57]. Moreover, certain characteristics of digitally-mediated communication such as perceived anonymity, physical separation, intrusive nature, and viral reach may give power to employees to engage in cyberbullying perpetration who are otherwise powerless and would normally not engage in traditional workplace bullying [58]. Being anti-hierarchical, the Internet al.so enables more equal opportunities for self-expression and is viewed as an equal playground devoid of social signs indicating authority [59, 60]. To the best of our knowledge, no prior studies have investigated the cross-level association between political climate and WCB perpetration. Overall, we argue that shared organizational level perceptions of manipulative, self-serving, and non-sanctioned behaviors are an important contextual variable that may trigger employees to engage in WCB perpetration. Thus, we hypothesize the following:

H1

There is a positive cross-level relationship between political climate and WCB perpetration.

The mediating role of psychological contract violation

We theorize that the relationship between political climate and WCB perpetration is mediated through psychological contract violation. This is consistent with the negative hedonic model of social exchange [35] in that a negative social exchange is an intervening variable between a negative initiating action by the actor that results in a negative reciprocating response from the target. We theorize that the extent to which an organization is highly political– wherein policies and practices being carried out unfairly and where the same rules do not apply equally to everyone– the more individual employees will experience psychological contract violation. This is because employees have an expectation of safety in their employment relationship in that they will be protected from such unfair treatment. We argue that a political climate breaches this psychological contract resulting in feelings of psychological contract violation, which represents a negative social exchange [40].

Rousseau [61] defined psychological contract as the employees’ perceptions of the mutual obligations existing between themselves and the organization. An important attribute of the psychological contract is that it is inherently subjective, residing in the “eyes of the beholder” [62, p. 246]. Employees expect, as part of their psychological contract, to be protected from unfair treatment including high levels of organizational politics [63]. A psychological contract breach is the cognition that the organization has failed to fulfill one or more of its obligations [64]. We posit that a psychological contract breach is likely to occur in organizations in which their members agree that there are high levels of organizational politics. Psychological contract violation captures the emotional experience of disappointment, frustration, anger, and resentment that may arise from a breach [64–66]. In this study, we focus on psychological contract violation as an outcome of political climate as an initiating action because this response should be more predictive of our criterion of interest, WCB perpetration, compared to psychological contract breach itself. Research has demonstrated that psychological contract violation leads to a range of negative attitudinal and behavioral outcomes [e.g., 67–69].

We propose that organizational politics at the organizational level are positively associated with employees’ experience of psychological contract violation at the individual level. When an organization is highly political and when this is agreed upon by employees in an organization, it creates an impression that the organization is not capable of fulfilling its exchange obligations. Organizational agents are believed to be preoccupied with the quest for their self-interests and accumulating power, often without concern for how their behaviors impact other members [70]. Organizational politics also weaken performance-reward relationships [71], which signals that an organization is incapable or unwilling to fulfill its exchange obligations. In a highly political climate, employees may get the impression that they are not being treated fairly or their contributions are not being valued by the organization [72], which may lead to a greater likelihood of perceiving a violation of psychological contract [70].

We further propose that psychological contract violation mediates the cross-level relationship between political climate and WCB perpetration. This proposition is consistent not only with social exchange theory and the negative hedonic model [35] but also with the environmental responsiveness model [70] which specifies that psychological contract violation mediates the relationships between perceptions of organizational politics and employee outcomes. When a psychological contract is violated, employees are likely to become angry and frustrated with their employment relationship and may seek retribution [73, 74]. Thus, psychological contract violation may lead employees to engage in antisocial and aggressive behaviors, such as bullying and interpersonal deviance [75, 76]. Moreover, research suggests that anger and frustration associated with psychological contract violation could be displaced toward a more tangible and easy target (such as a coworker or a subordinate) rather than the organization itself, a phenomenon often termed “displaced aggression” [6, 77]. In addition to displaced aggression, previous research suggests that negative emotional reactions to unfair treatment can be expressed more directly and overtly in digitally-mediated communication as opposed to face-to-face [78]. This may be because digitally-mediated communication is likely to reduce negative social appraisal, fear of counter-retaliation, and social isolation [79]. Taken together, we propose that in a highly political climate employees are likely to perceive psychological contract violation that, in turn, may motivate employees to engage in WCB perpetration. Thus, we expect the following:

H2

Psychological contract violation mediates the cross-level relationship between political climate and WCB perpetration.

The moderating role of toxic online disinhibition

We posit that individual differences in toxic online disinhibition are an important boundary condition in our research model. That is, toxic online disinhibition strengthens the positive relationship between psychological contract violation and WCB perpetration at the individual level, as well as the cross-level indirect effect of political climate on WCB perpetration. Negative reciprocity is a way to respond to unfair treatment and seek retribution [42]. Negative reciprocity– seeking retribution for unfair treatment– may be normative [31] but this is conditional on the characteristics of individuals. We propose that the extent to which WCB is perpetrated as a retribution for unfair treatment in response to experiencing a political climate, and the psychological contract violation in which this results, is contingent on individual differences in toxic online disinhibition. Toxic online disinhibition is particularly relevant to the context of this study as compared to, for instance, dominance and impulsivity [38] as it should be more closely tied to WCB [60].

Online disinhibition is a phenomenon where people in cyberspace do or say things that they would not ordinarily say or do in the face-to-face world as they feel less retrained and able to express themselves more directly [59]. This disinhibition can work in two seemingly opposite directions. The first is benign disinhibition which is when people are encouraged to share hidden emotions, fears, and wishes. The second is toxic disinhibition which is composed of more hostile behaviors such as passing rude remarks, harassment, trolling, and even hate speech [59, 80, 81]. Udris [82] in his study demonstrated that benign disinhibition can be empirically separated from toxic disinhibition and that the latter is more strongly related to WCB perpetration. Since we theorize that political climate will predict WCB perpetration, in part through psychological contract violation, toxic disinhibition is most relevant to our research model.

Previous research has demonstrated that online disinhibition is a risk factor for antisocial online behaviors, such as cyberbullying, cyberhate, and cyberharassment [e.g., 83–85]. The most commonly argued aspects of online disinhibition related to cyberbullying are anonymity, invisibility, asynchronicity, and the absence of rules or authority [e.g., 13, 85–87]. The anonymity inherent in online communication, in particular, has been associated with disinhibited online behavior [88]. Research suggests that individuals with high levels of online disinhibition express themselves more freely in cyberspace, do not sense the need to take responsibility for their actions, are low in self-control, and show a lack of empathy and an inability to recognize social cues [85, 89], which may increase the risk of WCB perpetration when such individuals feel that their psychological contract has been violated as a result of being exposed to a political climate. It is reasonable to expect that toxic online disinhibition as an impellor would exacerbate the psychological effects of exposure to a political climate and experiencing psychological contract violation as they relate to WCB perpetration [85, 90]. Denson et al. [90] noted that the most powerful aggressive urges arise when both instigation and impellance are strong. Our proposition is consistent with research that has found that toxic online disinhibition moderates the relationship between psychological distress and cyberbullying perpetration [60]. We accordingly predict that toxic online disinhibition will strengthen the relationship between psychological contract violation and WCB perpetration. We further predict that toxic online disinhibition will strengthen the indirect effect between political climate and WCB perpetration through psychological contract violation. This leads us to the following hypotheses:

H3

Toxic online disinhibition moderates the relationship between psychological contract violation and WCB perpetration, such that the relationship is stronger when toxic online disinhibition is high rather than low.

H4

Toxic online disinhibition moderates the cross-level indirect relationship between political climate and WCB perpetration via psychological contract violation, such that the indirect relationship is stronger when toxic online disinhibition is high rather than low.

Method

Participants and procedure

To test our hypotheses, we collected multiphase and multilevel data from 416 white-collar employees in 30 service sector organizations, including higher education, banking, and telecommunications in Islamabad, Pakistan. We collected our data from March 2023 to May 2023. We conducted a three-phase data collection to reduce potential common method variance. Using our personal and professional contacts with organizations’ leaders, we contacted the HR executives of 50 organizations and asked each organization to randomly select 20 employees as participants in our study. Of the organizations initially contacted, 20 denied their collaboration. Following prior multilevel research [e.g., 91], we utilized a multistage sampling design in which organizations as the primary sampling unit were first pre-selected in Stage 1, followed by recruitment of employees as the secondary sampling unit from each organization in Stage 2. The paper-and-pencil survey instrument was personally administered to the selected employees from each organization during working hours along with a cover letter stating the purpose of the study. Anonymity and confidentiality were ensured and informed written consent was obtained from each participant before their participation. At Time 1, we asked employees about their demographics, perceptions of organizational politics, and toxic online disinhibition. We received 511 responses from employees in 30 organizations, with an overall response rate of 51.1% and 60% at the individual and organizational levels, respectively. One month later, at Time 2, we asked these 511 employees to self-report their perceptions of psychological contract violation. We received 448 responses, representing a response rate of 87.6%. After another month, at Time 3, we asked these 448 employees to self-report their WCB perpetration behavior in the past 6 months and the time they spent on digital media on a typical weekday. We received 428 responses, representing a response rate of 95.5%. After discarding incomplete questionnaires, our final sample included 416 employees from 30 organizations. We linked surveys across all three points using the codes that were generated by the participants. The average number of employees in each organization was 13.87. Generally, it is suggested to have a large number of organizations (at least 30) and a small number of individuals (5 or even fewer) within each organization to conduct multilevel analysis [92, 93]. In terms of demographics, 51.2% of the participants were men, their average age was 35.44 years (SD = 6.935), and their average organizational tenure was 4.74 years (SD = 3.523). All the participants were well-educated with at least an undergraduate degree. In terms of ownership, 63.5% of the organizations were from the public sector.

Measures

All constructs in our study were operationalized using validated and widely used measures with established psychometric properties.

Political climate

Employees’ perceptions of organizational politics were measured using 6 items developed by Kacmar and Ferris [47] that captured general political behavior. A sample item is “Favoritism not merit gets people ahead in this organization.” Scale anchors ranged from 1 (strongly disagree) to 7 (strongly agree). Composite reliability (CR) was 0.965. The perceptions of organizational politics scale (POPS) has been used extensively in previous research [e.g., 94, 95].

We created the political climate variable using the direct consensus composition approach [96, 97], as it is the most familiar and popular form of composition among researchers. Consensus models are appropriate when the existence of the higher-level construct (e.g., political climate) is said to be conditioned on the lower-level units (e.g., individuals) demonstrating sufficient agreement in their scores [98]. This is consistent with previous research wherein political climate was operationalized and measured at the organizational level [e.g., 23, 27]. With respect to aggregation, there was statistical evidence that perceptions of organizational politics varied significantly across organizations [F(29,386) = 4.475, p = 0.000]. The values of intraclass correlation coefficients ICC (1), ICC (2), and the average r_WG(J) using a rectangular (uniform) null distribution were 0.20, 0.78, and 0.87, respectively. The ICC (1), ICC (2), and r_WG(J) values were greater than the suggested cut-off values of 0.12, 0.70, and 0.70, respectively [99–101]. These results provided evidence in support of aggregating political climate at the organizational level of analysis.

Psychological contract violation

Psychological contract violation was measured using a 4-item scale developed by Robinson and Morrison [102]. A sample item is “I feel a great deal of anger toward my organization.” Scale anchors ranged from 1 (strongly disagree) to 7 (strongly agree). CR was 0.957. The scale has been used extensively in previous research [e.g., 75].

Workplace cyberbullying perpetration

Consistent with previous research [7], WCB perpetration was assessed using a 3-item scale developed by Ybarra et al. [103]. A sample item is “In the last 6 months, how many times did you send rude or nasty comments to someone at work while online.” Scale anchors ranged from 1 (never) to 7 (almost every day). CR was 0.943.

Toxic online disinhibition

Toxic online disinhibition was assessed using a 4-item scale developed by Udris [82]. A sample item is “I don’t mind writing insulting things about others online, because it’s anonymous.” Scale anchors ranged from 1 (strongly disagree) to 7 (strongly agree). CR was 0.963. The scale has been used extensively in previous research [e.g., 60].

Control variables

At the individual level, we controlled for respondents’ gender, age, and time spent on digital media as they are related to WCB perpetration [17]. Gender was coded ‘0’ for females and ‘1’ for males. Age was measured as a continuous variable in years. Time spent on digital media during a typical weekday was assessed using a 7-point Likert scale ranging from 1 (none) to 7 (6 h or more). At the organizational level, we controlled for ownership of the current employer organization coded ‘0’ for private and ‘1’ for public [104].

Analytical strategy

We applied multilevel structural equation modeling (MSEM) in Mplus 7.0 to accommodate the multilevel nature of our study and the need to model top-down relationships. MSEM decomposes the variance of Level 1 variables into within-level and between-level components [105]. Relationships between these variance components can be modeled simultaneously and independently at each level through the specification of measurement and structural models [106]. In the present study, we adopted the multilevel moderated mediation model [e.g., 33], assuming a 2-1-1 mediation (X_j-M_ij-Y_ij) combined with a Level 1 moderator (Z_ij). In a 2-1-1 design, the indirect effect occurs only at the between-level because the predictor is a Level 2 variable [107]. To test our research model, we utilized the procedures outlined by Preacher et al. [108] for multilevel mediation models combined with the procedures of Bauer et al. [109] for moderated mediation. MSEM avoids potential problems of conflated within- and between-level relationships and can estimate indirect relationships more precisely than the traditional multilevel approach [108]. MSEM also allows for the inclusion of traditional latent variables to account for measurement error [110].

We used a Monte Carlo simulation with 5000 iterations to determine the 95% confidence interval for the indirect effect. The Monte Carlo method is considered the most appropriate approach for assessing mediation in the multilevel context [111]. If the Monte Carlo confidence interval does not contain the value of zero, the indirect effect is considered to be significant [112].

Given the cross-level nature of our research model, we computed the intraclass correlation coefficients [i.e., ICC (1) and ICC (2)] for our Level 1 variables to justify the use of multilevel modeling. ICC (1) provides an indication of whether there is a organizational level effect on the variable of interest, whereas ICC (2) provides an estimate of the reliability of the organizational level mean [98]. We found that the ICC (1) values for psychological contract violation, WCB perpetration, and toxic online disinhibition were 0.23, 0.37, and 0.40, respectively, indicating that an adequate amount of variances in these Level 1 variables resided at the organizational level. The ICC (2) values for psychological contract violation, WCB perpetration, and toxic online disinhibition were 0.81, 0.89, and 0.90, respectively, indicating high reliability for the organizational level means [113]. Overall, these results justified the use of multilevel analysis.

Results

Descriptive statistics

Table 1 presents the means, standard deviations, Cronbach’s alpha coefficients, and correlations among the main variables and controls. It is noteworthy that correlations do not take into account the non-independent nature of the data; as a result, relationships and significance tests should be interpreted with caution until estimated with multilevel analysis [114]. As shown in Table 1, none of the control variables were significantly correlated with the main variables. Thus, to increase the statistical power of our tests and reduce the probability of a Type II error, we excluded these impotent control variables from further analysis [115].

Table 1.

Descriptive statistics, Cronbach’s alpha, and correlations

Construct	1	2	3	4	5	6	7	8
1. Gender	-
2. Age	-0.030	-
3. Time spent on digital media	-0.043	0.052	-
4. Organizational ownership	0.023	-0.036	-0.125*	-
5. Political climate	0.005	-0.045	0.015	0.078	0.965
6. Psychological contract violation	-0.003	-0.014	0.075	0.072	0.384**	0.957
7. WCB perpetration	0.007	0.044	0.092	-0.017	0.353**	0.593**	0.943
8. Toxic online disinhibition	-0.002	-0.067	0.076	0.069	-0.053	-0.172**	-0.084	0.963
Mean	0.52	35.44	4.07	0.63	4.563	4.511	4.448	3.234
SD	0.500	6.935	1.747	0.482	1.331	1.335	1.313	1.223

Note. N = 416, ^**p < 0.01, ^*p < 0.05, Cronbach’s alpha along the diagonal in boldface

Common method bias

Constructs in our model were operationalized such that targets in social exchanges with their employer organizations self-reported their perceptions of organizational politics, psychological contract violation, toxic online disinhibition, and WCB perpetration. Individual targets were the relevant source of data in that they were embedded in the exchange, experienced political climate, and were most privy to their own experience of psychological contract violation, toxic online disinhibition, and the extent of their WCB behavior. Although our data source, individual employees, is the most relevant to our theorizing and is consistent with previous research [e.g., 7], we followed the recommendations of Podsakoff et al. [116] and used the unmeasured latent method construct (ULMC) technique to rule out the possibility of common method bias (CMB).

Following the ULMC technique, the hypothesized measurement model (Model 1) was compared to the measurement model with an unmeasured latent method factor (Model 2). The two models were compared using the chi-square difference test since Model 1 was a subset of Model 2 [117]. Model 2 was identical to Model 1, but the former also included freely estimated paths from the method construct to all manifest indicators of the latent variables. According to Richardson et al. [118], a significant chi-square difference between Model 1 and Model 2 may indicate that a CMB issue exists. The result of the chi-square difference test showed that the difference in χ² between Model 1 [χ² (113) = 139.774] and Model 2 [χ² (96) = 112.364] was non-significant [Δχ² (17) = 27.410, p > 0.05], indicating that Model 1 was not significantly worse than Model 2 and might be retained. Thus, CMB was not a potential validity threat to this study.

Confirmatory factor analysis

Following the suggestions of [119], we first performed a conventional confirmatory factor analysis (CFA) on the total sample covariance matrix in Mplus 7.0 using the maximum likelihood estimation method, ignoring the multilevel nested data structure. Results demonstrated that the proposed four-factor measurement model provided a good fit to the data [χ² (113) = 139.774, p = 0.045; RMSEA = 0.024; CFI = 0.997; TLI = 0.996; SRMR = 0.020] compared to alternative measurement models (see Table 2). Moreover, the standardized factor loadings of all items were greater than the cut-off value of 0.708, ranging from 0.898 to 0.947. Next, we performed a multilevel confirmatory factor analysis (MCFA) by decomposing the total sample covariance matrix into both within- and between-level variance. All individual level items were loaded on their respective constructs (i.e., psychological contract violation, WCB perpetration, and toxic online disinhibition) at both the within- and between-levels, whereas organizational level items were loaded on the organizational level variable (i.e., political climate). Results indicated that the proposed four-factor multilevel measurement model provided a good fit to the data [χ² (154) = 278.583, p = 0.000; RMSEA = 0.044; CFI = 0.981; TLI = 0.976; SRMR_within = 0.020; SRMR_between = 0.043].

Table 2.

Comparison of alternative measurement models

Model	χ2 (df)	p	RMSEA	CFI	TLI	SRMR
Four-factor model: Hypothesized model	139.774 (113)	0.045	0.024	0.997	0.996	0.021
Three-factor model a: Combined PC and PCV items	1917.202 (116)	0.000	0.193	0.780	0.742	0.178
Three-factor model b: Combined PC and WCBP items	1314.523 (116)	0.000	0.158	0.854	0.829	0.149
Three-factor model c: Combined PCV and WCBP items	965.838 (116)	0.000	0.133	0.896	0.878	0.0742
Two-factor model: Combined PC, PCV, and WCBP items	2945.684 (118)	0.000	0.240	0.655	0.602	0.199
One-factor model: All factors combined	4984.564 (230)	0.000	0.314	0.406	0.322	0.263

Note. N = 416, PC = political climate, PCV = psychological contract violation, WCBP = workplace cyberbullying perpetration

Further analysis based on the results of our MCFA revealed good reliability and validity of the measurement model. As depicted in Table 3, the standardized factor loadings of all items were greater than the cut-off value of 0.708 at both within- and between-levels, ranging from 0.863 to 1.000. The CR estimates for all constructs at both levels ranged from 0.915 to 0.999, which were significantly higher than the cut-off value of 0.70. Likewise, all the average variance extracted (AVE) values were greater than 0.5 at both levels, ranging from 0.781 to 0.996, lending support to the convergent validity of the scales [120–122]. Finally, as shown in Table 3, the square root of each construct’s AVE value was higher than its correlation coefficients with other constructs at both levels, lending support to discriminant validity [123].

Table 3.

Reliability, convergent, and discriminant validity

Construct	Item	FL	CR	AVE	Discriminant Validity
Individual level ( N  = 416)					PCV w	WCBP w	TOD w
PCVw	4	0.886–0.922	0.949	0.823	0.907
WCBPw	3	0.863–0.907	0.915	0.781	0.572**	0.884
TODw	4	0.875–0.919	0.937	0.789	-0.080	-0.059	0.888
Organizational level ( N  = 30)					PC b	PCV b	WCBP b	TOD b
PCb	6	0.895–0.944	0.971	0.846	0.920
PCVb	4	0.967–0.995	0.990	0.962	0.678**	0.981
WCBPb	3	0.993–0.998	0.997	0.992	0.726**	0.771**	0.996
TODb	4	0.991-1.000	0.999	0.996	-0.108*	-0.427**	-0.172*	0.998

Note. ^**p < 0.01, ^*p < 0.05, Square root of the average variance extracted value along the diagonal in boldface with the correlation coefficients (off-diagonal)

PCV_w = within-level (i.e., individual level) psychological contract violation, WCBP_w = within-level workplace cyberbullying perpetration, TOD_w = within-level toxic online disinhibition, PC_b = between-level (i.e., organizational level) political climate, PCV_b = between-level psychological contract violation, WCBP_b = between-level workplace cyberbullying perpetration, TOD_b = between-level toxic online disinhibition, FL = factor loading, CR = composite reliability, AVE = average variance extracted

Hypothesis testing

H1 predicted that there is a positive cross-level relationship between political climate and WCB perpetration. As shown in Fig. 2, political climate had a direct and positive cross-level relationship with WCB perpetration (β = 0.385; t = 2.995; p = 0.003), lending support to H1. Political climate also had a positive cross-level relationship with psychological contract violation (β = 0.686; t = 4.566; p = 0.000). At the individual level, psychological contract violation was positively associated with WCB perpetration (β = 0.575; t = 8.599; p = 0.000). Likewise, psychological contract violation was positively associated with WCB perpetration (β = 0.502; t = 4.117; p = 0.000) at the organizational level.

Fig. 2.

Results of the structural model. Organizational level variables are based on n = 30, and individual level variables are based on n = 416. ^***p < 0.001, ^**p < 0.01

H2 predicted that psychological contract violation mediates the cross-level relationship between political climate and WCB perpetration. As mentioned earlier, in a 2-1-1 design, the mediating effect occurs only at the between-level because the predictor variable resides at Level 2 [33, 107]. Results of 2-1-1 multilevel mediation analysis revealed that the indirect effect of political climate on WCB perpetration through psychological contract violation at the organizational level was positive and significant (unstandardized indirect effect = 0.427, p = 0.029; 95% CI [0.043, 0.810]) because the confidence interval did not contain zero, lending support to H2.

H3 proposed that toxic online disinhibition positively moderates the relationship between psychological contract violation and WCB perpetration at the individual level. The interaction between psychological contract violation and toxic online disinhibition was added to the individual level model. The results of multilevel moderation analysis indicated that the interaction effect of psychological contract violation and toxic online disinhibition on WCB perpetration was positive and significant at the individual level (β = 0.103; t = 2.224; p = 0.026) as well as at the organizational level (β = 0.438; t = 2.930; p = 0.003). We plotted the relationships between psychological contract violation and WCB perpetration at high and low levels of toxic online disinhibition (1 SD above and below the mean). In Fig. 3, the results of simple slopes analysis showed that the effect of psychological contract violation on WCB perpetration was more positive at higher levels of toxic online disinhibition (β = 0.710; t = 8.517; p = 0.000 at + 1 SD) than at lower levels of toxic online disinhibition (β = 0.505; t = 4.972; p = 0.000 at − 1 SD). Thus, H3 was supported at both the individual and organizational levels.

Fig. 3.

The moderating effect of toxic online disinhibition (TOD) at the individual level

Finally, H4 proposed that toxic online disinhibition moderates the indirect effect of political climate on WCB perpetration via psychological contract violation at the organizational level. Results of multilevel moderated mediation analysis showed that the index of moderated mediation was positive and statistically significant (β = 0.325, p = 0.000; 95% CI [0.158, 0.492]). Moreover, as expected, the conditional indirect effect of political climate on WCB perpetration via psychological contract violation was stronger and significant at higher levels (+ 1 SD) of toxic online disinhibition (β = 0.840, p = 0.000; 95% CI [0.361, 1.319]), whereas the conditional indirect effect was weaker and non-significant at lower levels (–1 SD) of toxic online disinhibition (β = 0.191, p = 0.164; 95% CI [-0.077, 0.459]) at the organizational level. Thus, H4 was also supported. Overall, the proposed multilevel moderated mediation model was tenable.

Discussion

Work and business have shifted more online in the post-pandemic era than ever before. More communication is done virtually using information and communication technologies as well. WCB has become a serious concern, in part as a result of these shifts [3]. A problem with this literature is the lack of understanding of the predictors of WCB perpetration, particularly in terms of how characteristics of organizations and individuals interact to shape this antisocial online behavior. This is a key issue for the study of business ethics, as information such as that provided in our study is necessary to prevent and mitigate WCB perpetration. The purpose of our study was to address this problem and, in so doing, we demonstrated that political climate is an organizational level predictor of WCB perpetration linked through psychological contract violation and bounded by toxic online disinhibition. The results of our study have important implications for theory, research, and practice.

Theoretical implications

Research has shown that WCB is related to a host of negative attitudinal, behavioral, and health outcomes for victims [e.g., 13–15]. Most of the research on WCB has focused on individual and organizational level risk factors for victimization and attitudinal and behavioral outcomes of WCB [16, 124, 125]. There has been a lack of theoretical development when it comes to the predictors of WCB perpetration, especially when it comes to the characteristics of organizations. Based on SET and the theoretical development of the political climate construct, our study demonstrates that political climate is predictive of WCB perpetration. This is an important insight in the WCB literature in that it demonstrates how differentiations between organizations in terms of work environment characteristics are a useful way to predict WCB perpetration. Our results suggest that future research should focus on predictors at multiple levels of theory and analysis.

In addition to a lack of clarity as to what characteristics of organizations predict WCB perpetration, there is also a lack of clarity as to why WCB perpetration occurs. In other words, and as applied to the current study, there is a lack of understanding as to why political climate might predict WCB perpetration. Based on SET and the literature on psychological contracts, we found that psychological contract violation mediates the cross-level relationship between political climate and WCB perpetration. This provides an important explanation as to why incidents of WCB perpetration occur among highly political organizations. Specifically, negative emotions experienced in connection to psychological contract violation, such as frustration and anger, may lead employees to engage in aggressive behaviors such as bullying. Moreover, consistent with previous research, our results suggest that negative emotional reactions to unfair treatment, such as exposure to a political climate, can be expressed more directly and overtly in digitally-mediated communication as opposed to face-to-face [78].

Negative exchange patterns are studied much less often than positive ones in the social exchange literature [35] and are typically limited to exchanges between parties at the same level of analysis (i.e., at the individual level). We offer evidence that there is a negative exchange pattern between organizations and employees. This pattern begins with the experience of political climate as an initiating act, which is responded by a negative reciprocating behavior in the form of WCB perpetration. In addition to explicating this negative cross-level exchange pattern, something which itself is novel to the literature and promotes studying negative cross-level social exchanges, we also explain based on SET when this will be exacerbated. Based on literature suggesting that negative reciprocity is contingent on individual differences [38], we found that toxic online disinhibition moderates (exacerbates) the relationship between psychological contract violation and WCB perpetration. We also found that the cross-level indirect relationship between political climate and WCB perpetration via psychological contract violation is contingent on toxic online disinhibition. This is important in that it provides a foundation for studying individual differences that moderate relationships between characteristics of organizations and WCB behavior.

Practical implications

Our results have important applied implications. WCB is becoming an increasingly relevant issue in the post-pandemic world in which we live [5]. WCB is related to a number of negative outcomes for individuals, such as job strain and negative job attitudes [126] as well as for organizations, such as reputational and potential financial costs [13]. The intent of our paper, from a practical perspective, was to understand how to prevent WCB by understanding perpetration behavior. Our results identify features of organizations (political climate) and individuals (toxic online disinhibition) that can be addressed to reduce the likelihood of WCB perpetration. They suggest that workplace politics, political climate, and perceptions of psychological contract violation are potential targets for intervention and that it is important to consider individual differences in WCB perpetration.

Political climate is shaped by how policies and practices are implemented, as well as the extent to which perceptions of such implementation are shared. Our results suggest that organizations can reduce WCB perpetration to the extent that they ensure that policies and practices are implemented fairly, that they do not involve favoritism of any kind, and that rewards for good performance are based on merit and not on unequal treatment. Disrupting the extent to which negative perceptions of organizational policies and practices are shared might reduce WCB perpetration. This can be done, for instance, through messaging from top executives that organizational politics will not be tolerated, by the design and implementation of human resource management programs, and through training front-line managers to communicate how their decisions are based on merit and not favoritism.

To the extent that workplace politics are mitigated and perceptions of political climate are more favorable, our results suggest that perceptions of psychological contract violation will be lower. Our results also suggest that lower levels of psychological contract violation are related to lower levels of WCB perpetration. There are other ways, in addition to organizational politics and political climate, to reduce perceptions of psychological contract violation. Main among these, as related to our paper, is trust-building. To the extent that organizations develop perceptions of trust among employees, psychological contract violation should be lower [127] as should unethical behavior that results from psychological contract violation [128]. Research also suggests that resource depletion and organizational identification are important to consider as related to behavioral outcomes of psychological contract violation [129]. This suggests that organizations should consider the types of support that their employees need, especially when it comes to adapting to change in a post-pandemic world [130], to prevent resource depletion and enhance organizational identification.

Finally, our results suggest that toxic online disinhibition is a boundary condition of the relationship between psychological contract violation and WCB perpetration. In other words, WCB perpetration is more likely among individuals higher in toxic online disinhibition. Online disinhibition is a phenomenon wherein people say or do things in cyberspace that they would normally not otherwise do in face-to-face interactions [59] and this is especially likely among individuals high in toxic online disinhibition. To reduce WCB perpetration, our results suggest that online disinhibition could be addressed as could individual differences in toxic online disinhibition. Time spent online is a key predictor of online disinhibition [131], and thus it is important to consider how much work in an organization is done online. Organizations also need to take steps to help employees feel that they are protected online [132]. Such steps could include training around norms and expectations for online behavior, as well as consequences for toxic online disinhibition. Moreover, organizations should develop workshops aimed at educating employees about netiquette [133], including teaching employees the proper way of communicating online and increasing their digital literacy [17].

Limitations and future research directions

A strength of our study is that it captured data from multiple employees across multiple organizations so as to be able to measure organizational level political climate. We measured our outcome variable separately from the independent, mediator, and moderating variables. Having said that, a limitation of our study is that it is not truly longitudinal in nature. It would be useful for future research to study the processes in our model across time. In this connection, it would be interesting to study indicators of relationship formation as outcomes of the processes captured in our model at a later time point. For instance, it would be worthwhile to understand the extent to which employees engaged in WCB leave the organization– voluntarily or not– and to compare them to employees who stay. Among those who stay, it would be interesting to study how WCB affects others in the organization, such as coworkers or team members.

We measured WCB perpetration from the perspective of targets in a social exchange. We did so because they are most privy to their own WCB behavior. It could be useful for future research to consider studying WCB from a network perspective. In other words, multiple individuals could report on their experience of WCB with other actors (e.g., coworkers) with whom they are embedded (e.g., in workgroups). This was obviously outside of the scope of our study and would require an additional level of theory and analysis. But, it would help to flesh out a fully multilevel– and networked– perspective to WCB perpetration. Even if a network perspective were not used, it would be an ideal case to study organizational politics as they relate to WCB perpetration from the individual, workgroup, and organization perspective in a single study. This would be extremely difficult in terms of data sourcing, collection, and design, but it would be a big next step in the study of WCB. Notably, data in this study were obtained from white-collar employees of service sector organizations in Pakistan; therefore, the results may lack generalizability. Thus, we encourage future researchers to collect data from multiple samples, industries, and cultures to cross-validate our findings.

Conclusion

WCB is increasingly common and a relevant issue in the post-pandemic world in which we live, where a significant amount of activity occurs online during work hours. It is therefore important to understand the characteristics of organizations and individual perpetrators that shape WCB behavior. A problem here, as it pertains to scholarship, is the lack of consideration about the role of the work environment, especially at the organizational level, that can trigger employees to engage in WCB perpetration. Therefore, the purpose of our study was to address this fundamental problem and thereby offer practical recommendations for organizations to prevent and mitigate WCB perpetration. Our results suggest that when non-sanctioned political behavior is encouraged in the workplace, psychological contract violation is a common occurrence, and when individuals have higher levels of toxic online disinhibition, incidents of WCB perpetration are more likely to occur. We hope that our research will inspire more multilevel studies in the WCB literature.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

OFM, SP, and AS were involved in the conception and design of the study. The literature review was conducted by AS and NJ. Data were collected and analyzed by OFM and AS. OFM prepared the original manuscript draft that was reviewed and edited by SP and NJ. All authors have read and approved the final manuscript.

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

The datasets generated during and/or analyzed during the current study have been submitted to the journal’s reveiew system as supplementary materials.

Declarations

Ethical approval and consent to participate

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the Ethical Review Committee of the Department of Management Sciences, COMSATS University Islamabad, though no specific approval number was issued. Written informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.