Exploring the effects of workplace deviance on perpetrators’ own work outcomes: the role of benevolent leadership in regulating fear activation and implication

Abstract

The prevalence of workplace deviance has become an urgent issue for managers. Although increasing research has investigated the detriments of workplace deviance on other employees and organizations, limited research has studied the harm of workplace deviance on perpetrators themselves. This research drew from appraisal theories of emotion and sought to understand perpetrators' affective and behavioral consequences of engaging in deviance. Using a diary method, a survey consisting of 92 employees with 918 observations was conducted. The results reveal that employees' deviance is positively related to their feeling of fear and that fear overrides feelings of guilt, ultimately decreasing work engagement and organizational citizenship behavior (OCB). Importantly, perceived benevolent leadership weakens the effects of deviance on perpetrators themselves by relieving fear associated with past deviance and mitigating the negative influences of fear on OCB.

Introduction

The prevalence of workplace deviance – intentional acts that violate the legitimate interests of, or cause actual harm to, organizations or their stakeholders (Bennett & Robinson, 2003) – has become an increasingly urgent ethical issue for managers (Harvey, Martinko, & Borkowski, 2017). Ethics and Compliance Initiative (2018) revealed that ‘50% of employees reported observing misconduct in the workplace.’ And the employees' engagement in workplace deviance costs billions of dollars per year for organizations (Bennett, Marasi, & Locklear, 2019). Overall, extant empirical research and meta-analysis have revealed that workplace deviance is related to decreased business unit performance, productivity, and profit (Carpenter, Whitman, & Amrhein, 2021; Dunlop & Lee, 2004), as well as negative consequences for other employees in the organization, such as decreased job satisfaction, organizational commitment and well-being, and increased stress, turnover intentions, and abusive supervision (Bowling & Beehr, 2006).

Despite the various consequences of employees' deviant behaviors in the workplace, scholars have underestimated deviance's total cost. Indeed, past research has almost exclusively examined how an employee's deviance negatively affects their team or organizational performance. In other words, extant research has primarily employed a victim-centric perspective to understand how an employee's deviance affects his or her coworkers, teams, and organizations. The present research adopts a perpetrator-centric perspective to investigate how employees react to their deviant behaviors. Because deviance in the workplace is almost unavoidable (Jetten & Hornsey, 2014), it is imperative for practitioners and researchers to know more about the full range of consequences associated with workplace deviance.

Several scholars have taken a perpetrator-centric perspective of deviance; however, most discussions have been theoretical in nature and have linked deviance to perpetrators' own outcomes such as reduced job performance, health, and well-being (Spector & Fox, 2002). However, among the handful of empirical studies, some have surprisingly revealed several positive consequences of deviance for the perpetrator. For example, Ilies, Peng, Savani, and Dimotakis (2013) proposed an emotion-based model to demonstrate that perpetrators engaged in more organizational citizenship behaviors (OCBs) to compensate for prior deviant behaviors due to heightened feelings of guilt.

The current research drew from appraisal theories of emotion (Smith & Ellsworth, 1985; Smith & Lazarus, 1990) and sought to understand the perpetrators' affective and behavioral consequences of engaging in deviance. Specifically, this research extends the literature and suggests that perpetrators experience intense feelings of fear after participating in deviant behaviors – a negatively-valenced, high-arousal emotion arising from low certainty and control (Ellsworth & Scherer, 2003). Unlike the feeling of guilt, which is often associated with individual reparatory behaviors (Haidt, 2003), fear is generally related to behavioral withdrawal (Frijda, Kuipers, & Ter Schure, 1989). Specifically, drawing from and extending past research that has revealed the effects of deviance on perpetrators' feelings of guilt (Bonner, Greenbaum, & Quade, 2017; Ilies et al., 2013), this study theorizes that perpetrators would also feel fearful after engaging in deviant behaviors. As a result, they would devote less to their work (Clark & Loxton, 2012; Schaufeli, Salanova, González-Romá, & Bakker, 2002) and perform fewer OCBs. While guilt is a reasonable emotional response to deviance (Ilies et al., 2013), fear is a basic and universal human emotion associated with survival (Ekman, 1992). Therefore, its effect on the perpetrator will be much stronger than, and override, the effects of guilt.

Although the indirect relationships between deviance and perpetrators' work engagement and OCB via fear are expected to be negative in general, several boundary conditions of these relationships are likely. Notably, appraisal theories of emotion suggest that the social context is vital in shaping employees' emotional reactions and the behavioral consequences of their emotional expressions (Gross, 2002). In organizations, leadership creates a social context that can affect the strength of fearful feelings associated with one's prior deviance by shaping the perpetrator's cognitions of environmental certainty and controllability (Smith & Lazarus, 1993). Specifically, leader support can weaken the negative social influences of fear (Lebel, 2017). As such, this research aims to explore whether benevolent leadership, defined as ‘leader behaviors that demonstrate individualized, holistic concern for subordinates’ personal and family well-being’ (Pellegrini & Scandura, 2008), can attenuate the effect of deviance on fear.

Benevolent leadership emphasizes that leaders care about followers' own benefit in an individualized and holistic way (Farh & Cheng, 2000). Accordingly, benevolent leadership is expected to mitigate the influences of deviance on fear as benevolent leaders' high tolerance for subordinates' mistakes and wrong-doings can alleviate the uncertain and uncontrollable appraisals associated with deviance. This research also expects benevolent leadership to weaken the effects of fear on work engagement and OCB because benevolent leaders provide employees with individual care and support to manage their fear. Figure 1 illustrates our overall conceptual model.

Figure 1. Conceptual model.

The present research contributes to the literature on workplace deviance in several ways. First, this research moves beyond the victim-centric perspective of deviance and examines the deviance's consequences to the perpetrators themselves. In doing so, this research contributes to a greater theoretical understanding of workplace deviance, its full range of outcomes, and the emerging stream of perpetrator-centric studies on deviance (Bonner, Greenbaum, & Quade, 2017; Meier & Spector, 2013). Second, by identifying fear as a salient, and perhaps dominating, emotional reaction to deviance, the present work extends recent research by revealing that perpetrators do not always engage in compensatory behaviors to amend for deviant behaviors. When fear is considered, this research suggests that employees are likely to suffer from their deviance by disengaging from work and behaving less pro-socially. Finally, benevolent leadership is a type of positive leadership, and existing research has connected it directly to several positive outcomes at work for employees (Lin, Ma, Zhang, Li, & Jiang, 2018; Pellegrini & Scandura, 2008). This research advances the field's knowledge on how benevolent leadership works in the context of workplace deviance. Specifically, this research suggests that benevolent leadership has beneficial moderating effects on employees who have engaged in deviant behaviors by mitigating the detrimental effects of that deviance on the perpetrator's emotional and behavioral outcomes. These moderation findings provide practical implications for managers who must minimize the damaging consequences of deviance, which is usually unavoidable in the workplace.

Theory and hypotheses

Appraisal theories of emotion

Appraisal theorists of emotion (Arnold, 1960; Frijda, 1986) seek to understand the factors that determine people's emotional experience from a cognitive perspective. The central tenet of appraisal theories of emotion is that an event's appraisal, rather than the event itself, determines whether an emotion (and why a certain emotion) is felt in a specific situation. Appraisals detect and evaluate the significance of the environment for individual well-being, that is, whether individual interests, concerns, or goals are satisfied or obstructed (Frijda, 2007). According to Smith and Lazarus (1990), ‘each emotion expresses a person's appraisal of a person–environment relationship involving a particular kind of harm or benefit’ (p. 611).

There are two well-established dimensions of appraisal (Moors, Ellsworth, Scherer, & Frijda, 2013). The first is motivational congruence, which assesses the consistency between the events and a person's motives or goals (Lazarus & Smith, 1988). And the second is certainty, which assesses the extent to which the situation is unpredictable (Arnold, 1960). In the next section, this research suggests that fear will be activated after employees engage in deviance based on these appraisal processes.

Workplace deviance and perpetrators' feelings of fear

Workplace deviance is intentional behavior that seeks to harm the organization or its members (Mackey, McAllister, Ellen, & Carson, 2021). Prototypical deviance includes verbal aggression, theft, and various forms of interpersonal mistreatment (Bennett & Robinson, 2003). Because deviance induces actual harm to others or the organization, perpetrators of deviance will be, or will at least entail, a high risk of being retaliated against by others (Spector & Fox, 2005). Aquino, Tripp, and Bies (2001), for example, found that victims often cope with interpersonal deviance by retaliating against the perpetrators. According to appraisal theories of emotion, when an event is appraised as inconsistent with one's motives and goals and will likely result in uncertain outcomes, the event typically elicits the emotion of fear (Smith & Ellsworth, 1985; Smith & Lazarus, 1990). Accordingly, we suggest that perpetrators of deviance are likely to experience feelings of fear because their appraisals of prior deviance are associated with the cognitions of motivational incongruence and uncertainty.

First, deviance should arouse perpetrators' cognitions of motivational incongruence, which assesses the consistency between the events and a person's motives or goals (Lazarus & Smith, 1988). People typically strive to be moral persons or at least strive to be perceived as such (Aquino & Reed, 2002). Engaging in deviance jeopardizes such goals because deviance is often regarded as a form of unethical behavior (Moore, Detert, Treviño, Baker, & Mayer, 2012; Moore, Mayer, Chiang, Crossley, Karlesky, & Birtch, 2019), which leads to motivational incongruence. Besides, people fundamentally need to be accepted by social groups (Baumeister & Leary, 1995). In the work context, employees often desire to maintain good interpersonal relationships with others to maintain connectedness or climb the organizational ladder more effectively (Ferris, Lian, Brown, & Morrison, 2015). Once discovered, perpetrators may expect their deviant behaviors to lead others to dislike or ostracize them (Lian, Ferris, Morrison, & Brown, 2014). As such, this research suggests that engaging in deviance or appraising past behavior as deviant will lead to motivational incongruence.

Second, appraisals of past deviance induce high levels of uncertainty, which is the core appraisal dimension for the emotion of fear (Smith & Lazarus, 1990). Perpetrators' uncertainty comes from ruminating on whether, when, and by whom they will be caught or retaliated against at work (Aquino, Tripp, & Bies, 2001). Empirical findings demonstrate that those who observe workplace deviance are motivated to punish the perpetrator, and the perpetrators themselves will experience stress and constraints due to such uncertainties (Meier & Spector, 2013).

Taken together, engaging in or appraising prior deviance leads to motivational incongruence and high uncertainty levels, which are the core appraisal contents for the negative emotion of fear (Smith & Ellsworth, 1985). Hence, the following hypothesis is proposed:

Hypothesis 1: Workplace deviance is positively related to the perpetrator's negative emotion of fear.

Implications for the perpetrator's work engagement and OCB

Once the appraisal dimensions of deviance trigger feelings of fear, we suggest that such feelings would lead to behavioral withdrawal (Fox & Spector, 1999). The experience of fear entails high coping potential levels; that is, individuals are motivated to expend effort to deal with unpleasant emotions (Smith & Ellsworth, 1985). The coping strategy of fear is generally associated with ‘flight,’ withdrawing, or avoidant behavioral tendencies as an attempt to seek protection (Frijda, Kuipers, & Ter Schure, 1989). More specifically, fear is a typical form of avoidance emotion that activates avoidance motivation (Lerner & Keltner, 2001). Driven by avoidance motivation, individuals experiencing fear usually withdraw from their current activities and seek to preserve their efforts (Öhman, 2008). In the organizational context, an employee suffering from fear is likely to disengage from their work (Clark & Loxton, 2012). Organizational behavior research has suggested that employee work engagement can vary significantly across workdays, and emotions play a key role in affecting daily work engagement (Bledow, Schmitt, Frese, & Kühnel, 2011; Park, Kim, Jung, Yun, & Hai, 2021). For example, using a diary study involving 1,203 day-level data from 117 workers, Miralles, Navarro, and Unger (2015) empirically demonstrated that general negative emotions negatively influenced state work engagement. Hence, the following hypothesis is proposed:

Hypothesis 2a: Fear is negatively related to the perpetrator's work engagement.

Social psychology research also suggests that, in addition to withdrawing from current activities, the avoidance motivation activated by fear will motivate individuals to avoid social interactions with others (Plant & Devine, 2008). In the organizational context, recent research has revealed that avoidance emotions (i.e., anxiety) are associated with employees' avoidance-oriented interpersonal behaviors that ‘include any action that seeks to remove an individual from, or otherwise minimize, interaction with situations or individuals’ (Ferris, Yan, Lim, Chen, & Fatimah, 2016). Drawing on these arguments and empirical findings, this research postulates that experiencing fear will also decrease employee OCB.

Broadly defined, OCB is a form of discretionary behavior that goes beyond formal job requirements but promotes organizational effectiveness (Podsakoff, MacKenzie, & Organ, 2006). Engaging in OCBs is an effortful social interaction process (Fehr, Yam, He, Chiang, & Wei, 2017); it requires strong psychological resources (e.g., self-control; for a review, see Lian, Yam, Ferris, and Brown, 2017) to enact altruistic behaviors. Thus, employees are less likely to do OCB when they experience fear, the action tendencies of which are avoidance-oriented (Öhman, 2008). Indirect support for this argument comes from Milliken, Morrison, and Hewlin (2003), who revealed that employees experiencing fear were more likely to remain silent at work, which is negatively related to OCBs such as suggesting new ideas to improve organizational effectiveness. Following these theoretical arguments and findings, the hypothesis is as follows:

Hypothesis 2b: Fear is negatively related to the perpetrator's OCB.

Integrating Hypothesis 1 with Hypotheses 2a and 2b, this research suggests that deviance will have indirect and negative relationships with work engagement and OCB via heightened feelings of fear. At first glance, our hypotheses seem to directly contrast past research, which has revealed that deviance has an indirect, positive relationship with OCB via the negative moral emotion of guilt (Ilies et al., 2013). Although individuals can experience both guilt and fear due to prior deviance, we argue that the detrimental effects of fear on perpetrator OCB and work engagement will override the positive effects of guilt for several reasons.

First, as aforementioned, fear is widely recognized as a basic human emotion (Izard, 1992). Specifically, Frijda (1986) stated that the significant difference between basic emotions and nonbasic emotions was action readiness. He regarded basic emotions as behavior-oriented. Also, he considered that basic emotions are not composed of different other emotions. Fear is considered a basic emotion because the corresponding action readiness (to avoid a threat) cannot be reduced to any other action tendency. Guilt, however, is not basic because it is defined by its object (i.e., the negative evaluation of a specific behavior), and its correlated action tendencies (e.g., reparative behaviors) are composed of sorrow and disgust, which belong to other basic emotions (Ekman, Friesen, & Ellsworth, 1982). Therefore, fear tends to have more direct and stronger influences on individual actions than guilt because basic emotions are biologically primitive; that is, they are essential in evolution and adaptation (Izard, 1992). Indeed, Öhman and Mineka (2001) demonstrated that fear activation and fear responses are automatic and impenetrable to cognitive control. In other words, after fearful feelings are activated, the resulting responses and actions are biologically natural, fast, and strong; indeed, they are likely stronger than the responses and actions are driven by nonbasic emotions such as guilt.

Second, whereas one's avoidance motivation generally drives the negative effect of fear on OCB, the positive effect of guilt on OCB is generally driven by one's approach motivation (Elliot & Covington, 2001). The classic research on social psychology has suggested that when both motivational tendencies are activated, the self is more likely to be dominated by avoidance motivation than approach motivation (Tversky & Kahneman, 1991). In directly comparing the appraisal differences of various emotions, Smith and Ellsworth (1985) revealed that the avoidance tendencies associated with fear were significantly stronger than the approach tendencies associated with guilt and shame. The hypothesis is as follows:

Hypothesis 3: Controlling for guilt, workplace deviance has indirect and negative relationships with the perpetrator's (a) work engagement and (b) OCB via the negative emotion of fear.

The moderating role of benevolent leadership

This research highlights the role of fear in the link between workplace deviance and work engagement and OCB. Besides, it is important to understand that this mediated effect is unlikely to be consistent across all employees. To identify factors most likely to moderate the mediated effect, this research draws from Gross (1998) theory of emotion regulation. Specifically, benevolent leadership, referring to leaders demonstrating individualized, holistic concern for subordinates' personal welfare within or beyond the work domain (Farh & Cheng, 2000; Li, Rubenstein, Lin, Wang, & Chen, 2018; Lin et al., 2018), will shape the influences of deviance on the perpetrator's fearful feelings and work outcomes by affecting fear's effects on employees' behaviors.

Our theorizing is consistent with Gross' theory of emotion regulation (Gross & Barrett, 2011), which explains to what intensity an emotion is felt and whether the felt emotion would lead to behavioral outcomes. Briefly stated, Gross (1998) suggested that emotion regulation is a two-stage process. The first stage focuses on interpreting the emotion experienced, whereas the second stage focuses on behavioral responses, if any, to the emotion experienced. Next, this research explicates the role benevolent leadership plays at both stages (i.e., fear activation and fear implication).

Antecedent-focused emotion regulation

In the antecedent-focused stage of emotion regulation, employees could respond to their deviant behaviors differently based on their specific leader. Although the experience of fear is common across all employees who committed deviance, the intensity of fear differs. Specifically, Gross and Barrett (2011) suggested that individuals can alter their interpretations of an emotion-eliciting cue based on social contexts. For example, an employee who perceives strong organizational sanctions against unethical behavior would likely experience more fear after committing deviance than an employee who perceives weak organizational sanctions against unethical behavior.

The positive relationship between a perpetrator's deviance and his/her own emotion of fear will be weakened by high benevolent leadership, which reduces the perpetrator's uncertainty appraisals after deviance. Specifically, benevolent leaders have a high tolerance for subordinates' mistakes and wrongdoings, are likely to forgive employees' deviant behaviors, and allow opportunities to correct mistakes (Farh & Cheng, 2000; Lin et al., 2018; Wang & Cheng, 2010). On the one hand, benevolent leaders will neither adopt nor allow others to adopt a tit-for-tat policy to retaliate against employees who engaged in deviance. On the other hand, those employees would be certain that retaliation against their deviant behaviors is unlikely, and thus feelings of fear are likely less intense. Chen, Eberly, Chiang, Farh, and Cheng (2014), for example, discovered that benevolent leadership was beneficial in building a forgiving climate in an organization, thus reducing the likelihood of perpetrators receiving retaliation (Fehr & Gelfand, 2012). Taken together, the hypothesis is as follows:

Hypothesis 4: Benevolent leadership moderates the intra-individual level relationship between workplace deviance and the perpetrator's negative emotion of fear such that the relationship is less positive when benevolent leadership is high rather than it is low.

Response-focused emotion regulation

After experiencing an emotion, in our case fear, individuals turn to the response-focused stage of emotion regulation (Gross & Barrett, 2011). In this stage, emotion-driven behavioral consequences can be amplified or suppressed based on social contexts. For example, whereas experiencing gratitude might lead to increased prosocial behavior (Spence, Brown, Keeping, & Lian, 2014), a cut-throat organizational culture might suppress emotion-driven prosocial behavior. Although experiencing fear might lead to reduced job engagement and OCB in a similar vein, a supportive social context might suppress such avoidance tendencies (Lebel, 2017).

In the context of individual deviance related to fear and the two work outcomes this research theorized, we suggest that benevolent leaders can mitigate the negative behavioral effects of fear by providing employees with individualized care (Li et al., 2018; Lin et al., 2018; Pellegrini & Scandura, 2008). Typical benevolent leadership behaviors include treating employees as family members and showing holistic concern about employees' emotional and attitudinal states (Wang & Cheng, 2010). By providing tangible and intangible resources and social support, benevolent leaders give employees the confidence to deal with negative events or situations and manage their fearful emotions (Lebel, 2017). In short, we expect that the relationships between fear and work engagement and OCB will be less negative when individuals perceive high rather than low benevolent leadership. Thus, the hypothesis is as follows:

Hypothesis 5: Benevolent leadership moderates the intra-individual level relationships between the perpetrator's negative emotion of fear and (a) work engagement, and (b) OCB, such that the relationships are less negative when benevolent leadership is high rather than it is low.

Integrating Hypothesis 4 with Hypotheses 5a and 5b, the following hypothesis is proposed:

Hypothesis 6: Benevolent leadership moderates the indirect relationships between workplace deviance and the perpetrator's (a) work engagement and (b) OCB via the negative emotion of fear, such that the indirect relationships will be less negative when benevolent leadership is high rather than low.

Methods

Participants and procedures

This research collected data from employees working at a Chinese firm located in Central China. This firm is a large corporation with diversified business areas such as manufacturing, construction, and real estate. The chief executive officer (CEO) of this firm is an EMBA alumnus who graduated from the business school with which the first author is affiliated. The CEO acted as the ‘gatekeeper’ in the data collection process. With his permission and support, the research team asked the HR manager of this firm to post an advertisement to all full-time employees. In this advertisement, the research team provided information about our university affiliation and invited employees to participate in a work attitudes and behavior-related survey. Also, the research team stated the survey's time requirement (10 consecutive working days' participation) and indicated each participant would receive 50 Chinese Yuan (about $7.70) for participating.

A total of 102 employees from different departments and work roles voluntarily participated in this diary survey. One week before starting the diary survey, the research team gathered participants in a large meeting room and repeatedly emphasized that the daily questionnaires had to be filled out after work. Besides, the research team emphasized voluntary participation and full confidentiality again. Then, the research team asked the participants to complete an initial survey in which they reported their demographic information and rated their leaders' benevolent leadership. A week after the initial survey, participants started the diary survey assessing their daily performance of deviance, negative emotions of fear and guilt, and work engagement and OCB.

Ten employees did not return any of their daily surveys; therefore, the final sample consisted of 92 employees with 918 observations, indicating the response rate was 90.20% in this study. Of the 92 participants, the average age was 30.17 (SD = 7.11), and 46.55% were male. Most of them were well-educated (81.52% had a college education or higher). In terms of their occupation, 11.96% of participants were from production departments, 51.09% were from administrative departments, 19.57% were from research and development departments, and 17.39% were from marketing departments.

Measures

The scales this research used in this study were originally in English. Hence, to guarantee our questionnaires' accuracy, which were finally in Chinese, this study followed Brislin (1980) translation and back-translation procedure. Except for perceived benevolent leadership, all of the variables were measured using 7-point Likert scales (1 = strongly disagree, 7 = strongly agree).

Deviance

This study measured employees' daily deviance using a 5-item short scale of workplace deviance adapted from Bennett and Robinson (2000). The original version of Bennett and Robinson (2000) scale has 19 items and makes it difficult to use in a diary survey. This study selected five items, complying with the following standard (1) the behaviors described in items could vary daily, and (2) the behaviors described in items composed of not only individual-targeted deviance but also organization-targeted deviance. These five items are ‘Worked on a personal matter instead of work for my employer,’ ‘Spent too much time fantasizing or daydreaming instead of working,’ ‘Made fun of someone at work,’ ‘Took an additional or a longer break than is acceptable at my workplace,’ and ‘Lost my temper while at work.’ This 5-item short scale was validated in a prior diary study conducted in the Chinese context (Fehr et al., 2017). Cronbach's α was .80 in this study.

Fear

This study measured employees' daily negative emotion of fear using four items, including ‘afraid,’ ‘scared,’ ‘nervous,’ and ‘jittery.’ This scale was adapted from the PANAS-X scale (Watson & Clark, 1994). Compared with the original 6-item scale of fear, the item ‘frightened’ was not used because its Chinese translation is virtually identical to the Chinese translation of ‘scared.’ Besides, the item ‘shaky’ was not used because its Chinese translation has vague meanings. Cronbach's α was .87 in this study.

Work engagement

This study measured employees' daily work engagement using Schaufeli, Bakker, and Salanova (2006) 9-item scale. Sample items included, ‘I felt happy when I was doing this task intensely,’ and ‘I was enthusiastic about my task.’ Cronbach's α of these nine items was .95.

Organizational citizenship behavior

Consistent with Fehr et al. (2017), this study measured employees' daily OCB using a 5-item short scale adapted from Lee and Allen (2002) 16-item scale. Like our logic in shortening the deviance scale, these five items were selected because they were most likely to vary each day and the behaviors described in items composed of individual-targeted OCB and organization-targeted OCB. Specifically, these five items are ‘Expressed loyalty toward the organization,’ ‘Took action to protect the organization from potential problems,’ ‘Demonstrated concern about the image of the organization,’ ‘Willingly gave my time to help others who had work-related problems,’ and ‘Showed genuine concern and courtesy toward my coworkers.’ In this study, Cronbach's α was .87.

Perceived benevolent leadership

In the entry survey, participants evaluated their supervisors' benevolent behaviors using an 11-item subscale of the Paternalistic Leadership scale (Chen et al., 2014). Sample items included, ‘My supervisor tries to understand the cause when I do not perform well,’ and ‘My supervisor will help me when I am in an emergency’ (1 = strongly disagree, 6 = strongly agree). Cronbach's α was .94.

Control variables

This study controlled for the possible influences of employees' negative emotion of guilt on their state work engagement and OCB; indeed, Ilies et al. (2013) have demonstrated the mediating role of guilt in linking individual deviance with increased OCB. Guilt was measured using the first two items adapted from the PANAS-X scale (Watson & Clark, 1994). Specifically, the research team asked the participants to evaluate their feeling of ‘guilty’ and ‘ashamed’ at work. Cronbach's α was .80.

Analytic approach

Following Bolger, Davis, and Rafaeli (2003) recommendations for how to analyze diary data, and in line with prior diary studies (Wang, Liu, Liao, Gong, Kammeyer-Mueller, & Shi, 2013), this study controlled for previous days' measures of fear (i.e., the mediator), and state work engagement and OCB (i.e., the dependent variables) at the within-individual level in all our analyses.

This study tested our research model using multi-level path-analysis because our data were nested. The Mplus 7.0 software (Muthén & Muthén, 2010) was used in the analyses of this study. To test the main effects and mediation relationships (i.e., Hypotheses 1–3), this study initially estimated a multilevel mediation model that specified the Level 1 random slope effect of deviance on fear, the Level 1 random slope effects of fear on state work engagement, and OCB.

This study also calculated the direct influence of deviance on the outcome variables. Moreover, the cross-level effects of benevolent leadership on the mediator of fear and the two dependent variables were estimated as controls. Besides, as noted, guilt was added as a control variable, mediating the relationships of deviance with state work engagement and OCB. All previous days' measures of fear, state work engagement, and OCB were included in this model as control variables. Before conducting these analyses, this study first group-mean centered Level 1 predicting and control variables (Hofmann, Griffin, & Gavin, 2000).

Based on this multilevel mediation model, this study tested our multilevel moderated mediation hypotheses by additionally estimating the cross-level moderating effects of benevolent leadership on the random slope for deviance predicting fear and on the random slopes for fear predicting state work engagement and OCB. This multilevel moderated mediation model was used to demonstrate the moderation and moderated mediation relationships (i.e., Hypotheses 4–6). To alleviate potential problems relating to multicollinearity, the moderator in Level 2 (i.e., benevolent leadership) was grand-mean centered before conducting these analyses (Hofmann, Griffin, & Gavin, 2000).

In further testing the indirect effect of deviance on state work engagement and OCB through fear (i.e., Hypotheses 3a and 3b), Monte Carlo simulations were conducted with 20,000 replications to compute 95% confidence intervals (Preacher & Selig, 2012). Besides, to examine the moderated mediation effect further (i.e., Hypotheses 6a and 6b), this study tested the indirect effects of deviance on state work engagement and OCB through fear under two conditions (high vs. low level of benevolent leadership) using a method Bauer, Preacher, and Gil (2006) introduced.

Results

Preliminary analysis

Means, standard deviations, reliabilities, and correlations are presented in Table 1. This study conducted a multilevel confirmatory factor analysis (MCFA) to confirm the hypothesized 6-factor structure of employee perceived benevolent leadership, daily deviance, fear, state work engagement, OCB, and guilt, while also accounting for the nested structure of the data. Before conducting MCFA, state work engagement was packed into three indices representing the three distinct sub-dimensions: vigor, dedication, and absorption.

Table 1. Descriptive statistics and correlations among variables

Note: Correlations above the diagonal represent between-individual correlations (N = 92).

Correlations below the diagonal represent within-individual correlations (N = 918).

SD_w and SD_b are standard deviations computed within and between individuals separately.

Coefficient α estimates of reliability are in parentheses on the diagonal. (The same below).

The hypothesized 6-factor model demonstrated a good fit to the data: χ² = 563.28, df = 186, p < .001, CFI = .90, RMSEA = .05; SRMR (within) = .05, SRMR (between) = .10. This 6-factor model fit the data better than a 5-factor model grouping the two dependent variables of state work engagement and OCB: χ² = 1,085.84, df = 190, p < .001, CFI = .76, RMSEA = .07; SRMR (within) = .07, SRMR (between) = .10; Δχ² = 680.72, Δdf = 4, p < .001. Another 5-factor model grouped the independent variable (deviance) and the mediator (fear): χ² = 1,197.86, df = 190, p < .001, CFI = .74, RMSEA = .08; SRMR (within) = .11, SRMR (between) = .10; Δχ² = 189.13, Δdf = 4, p < .001. Overall, the results of the MCFA support the discriminant validity among our daily focal constructs.

Hypotheses testing

Hypothesis 1 proposed the direct effect of daily deviance on fear, and Hypotheses 2a and 2b proposed the direct effects of fear on state work engagement and OCB, respectively. This study tested these three hypotheses simultaneously in the same multilevel mediation path-analytical model. Here, fear was included as the mediator, and state work engagement and OCB were included as the dependent variables while controlling for guilt, previous day's fear, work engagement, and OCB. Results as shown in Table 2 revealed that daily deviance was negatively related to fear (β = .12, SE = .05, p < .01), supporting Hypothesis 1. Consistent with prior research, this study also found a direct relationship between deviance and guilt (β = .12, SE = .05, p < .05). In addition, when the effects of guilt on state work engagement (β = –.07, SE = .05, n.s.) and OCB (β = –.00, SE = .05, n.s.) were controlled, fear was still negatively related to state work engagement (β = –.15, SE = .06, p < .01) and OCB (β = –.12, SE = .05, p < .01). Hence, the results supported Hypotheses 2a and 2b.

Table 2. Multilevel path analysis results for testing Hypotheses 1, 2a, and 2b

Note: ‘t’ reflect the evaluation ‘today’; ‘t−1’ reflects the evaluation on the previous day compared to ‘today.’

The above estimates represent unstandardized path coefficients.

All hypothesized effects were calculated in the same path-analytical model and specified as random slopes.

**p < .01; ***p < .001. (The same below).

This research proposed in Hypotheses 3a and 3b that fear mediated the effects of daily deviance on perpetrators' state work engagement and OCB after controlling for the influences of guilt feelings. Following Bauer, Preacher, and Gil (2006), this study computed the indirect effects as the product of path a (daily deviance to fear) and path b (fear to state work engagement and OCB), plus the covariance between them (i.e., indirect effect = path a × path b + cov [a, b]). Bootstrapping results revealed that after controlling for the mediating effects of guilt, the indirect relationship between daily deviance and state work engagement via fear was significant (indirect effect = –.02, 95% CI = [–.004, –.001]). Besides, the indirect relationship between daily deviance and OCB via fear was also significant (indirect effect = –.01, 95% CI = [–.035, –.001]). Thus, Hypotheses 3a and 3b are supported.

Hypothesis 4 stated that employees' perceptions of benevolent leadership would moderate the relationship between daily deviance and fear. This relationship would be stronger for employees whose direct leaders are less (vs. more) benevolent. This study tested this cross-level moderation effect in a model that included the Level 2 moderator (i.e., perceived benevolent leadership) as a predictor of the random slope of the effect of daily deviance on fear. As the data in Table 3 and Figure 2 show, the relationship between daily deviance and fear was moderated by employee's perceived benevolent leadership (β = –.12, SE = .06, p < .05). Also, the moderating effect was plotted in Figure 3 (Cohen, Cohen, West, & Aiken, 2013). Simple slope tests indicated that daily deviance was more positively related to fear when employee perceived low benevolent leadership (simple slope = .64, p < .01) compared to high benevolent leadership (simple slope = .49, p < .01). Thus, Hypothesis 4 is supported.

Figure 2. Unstandardized estimates of the path coefficients (H4–H6).

Figure 3. The moderating effect of benevolent leadership on the relationship between deviance and fear.

Table 3. Results of the cross-level moderating effects of perceived benevolent leadership

Hypotheses 5a and 5b proposed that the effects of fear on state work engagement and OCB were moderated by benevolent leadership. This study tested these two hypotheses following the same procedures we used to test Hypothesis 4. Results revealed that employee perceived benevolent leadership moderated the relationship between fear and OCB (β = .13, SE = .05, p < .01), but could not moderate the relationship between fear and state work engagement (β = .10, SE = .08, p > .05). The interactive effects of fear with perceived benevolent leadership on OCB are plotted in Figure 4. Simple slope tests showed the association between fear and OCB was more negative when employees perceived low benevolent leadership (simple slope = –.67, p < .001), compared to high benevolent leadership (simple slope = –.50, p < .001). In sum, Hypothesis 5a is not supported, and Hypothesis 5b is supported.

Figure 4. The moderating effect of benevolent leadership on the relationship between fear and OCB.

Finally, Hypotheses 6a and 6b proposed moderated mediation effects that the indirect effects of daily deviance on state work engagement and OCB via fear would be stronger when employees perceived lower (vs. higher) levels of benevolent leadership. Following Bauer, Preacher, and Gil (2006), this study calculated the indirect influences of daily deviance on the two outcomes through fear at low perceived benevolent leadership (one SD below the mean) versus high (one SD above the mean) levels of perceived benevolent leadership. Specifically, the indirect effect was calculated as the product of path a (daily deviance to daily fear) and path b (fear to state work engagement/OCB) plus the covariance between them (i.e., indirect effect = path a × path b + cov [a, b]). The effect sizes of path a and path b will change at high versus low levels of the moderator (perceived benevolent leadership).

Bootstrapping results revealed that for state work engagement, the indirect effect was –.35 (95% CI = [–.82, .11]) when perceived benevolent leadership was low, and the indirect effect was –.21 (95% CI = [–.47, .05]) when perceived benevolent leadership was high. These results suggested that benevolent leadership does not moderate the mediated relationship between deviance and engagement via fear. Hence, Hypothesis 6a is not supported. For OCB, bootstrapping results showed that the indirect effect was –.43 (95% CI = [–.81, –.05]) when perceived benevolent leadership was low, and this indirect effect was –.25 (95% CI = [–.45, –.04]) when perceived benevolent leadership was high. The difference between the two indirect effects was significant (difference = .18; 95% CI = [.01, .36]). These results indicated that the indirect effects of daily deviance on OCB through fear were stronger when perceived benevolent leadership was low than high, thereby supporting Hypothesis 6b.

Discussion

Through a diary study, this research demonstrated the conceptual model regarding the intra-individual relationships among employee deviance, the negative emotion of fear, and the individual's work engagement and OCB, as contingent on benevolent leadership at the between-individual level. Specifically, this research found that employee deviance decreases work engagement and OCB through the increased negative emotion of fear. Moreover, benevolent leadership weakens these negative effects of deviance by mitigating both the first-stage relationship between deviance and fear (i.e., fear activation) and the second-stage relationships between fear and work outcomes (i.e., fear implication).

Theoretical implications

Several theoretical contributions and implications of the current research are worth highlighting. First, compared with empirical investigations focused on how employees' deviance affects other organizations' members (Eissa, Lester, & Gupta, 2020; Foulk, Woolum, & Erez, 2016; Moon, Morais, de Moura, & Uskul, 2021) and organizational effectiveness (Carpenter, Whitman, & Amrhein, 2021; Jetten & Hornsey, 2014), this research contributes to the workplace deviance literature by theorizing and validating a perpetrator-centric perspective that reveals the effects of deviance on the perpetrator's own emotional, psychological, and behavioral outcomes in the workplace. Specifically, this study demonstrated that workplace deviance also triggers the perpetrator's negative emotion of fear, with negative downstream behavioral consequences.

Second, this research extends recent literature on deviance's consequences by revealing that the perpetrator conducting deviance might engage in withdrawal behaviors because of their negative emotion of fear. Extant research showed that perpetrators may always conduct compensatory behaviors to amend deviant behaviors (e.g., Ilies et al., 2013). Specifically, Ilies et al. (2013) found guilt motivates the perpetrator of deviance to engage in more reparatory behavior such as OCB. Our research showed that when we controlled for the mediating effect of guilt, an employee's prior deviance negatively affected his or her psychological state (decreased work engagement) and pro-social behavior (decreased OCB) via the negative emotion of fear. These findings suggest that although prior deviance can lead to the perpetrator's guilt and fear simultaneously, fear might be a more salient and perhaps dominating emotional reaction to deviance and further demotivate the perpetrator to engage in OCB.

The third contribution is to the benevolent leadership literature by identifying that benevolent leadership works in the context of workplace deviance. Specifically, prior benevolent leadership research has focused on its main, beneficial effects on employee outcomes, such as task performance and creativity (Chen et al., 2014; Lin et al., 2018; Wang & Cheng, 2010). This research extends these findings by showing the indirect, beneficial effects of benevolent leadership on employee outcomes. Our results show that benevolent leaders can reduce deviance-driven fear and prevent employees from decreasing pro-social behaviors (OCB) after they experience fear. These findings support Toegel, Kilduff, and Anand (2013) theoretical argument that leadership plays a vital role in alleviating employees' negative emotions.

Moreover, these findings suggest another beneficial effect of benevolent leadership: it prevents employees from generating unpleasant feelings (i.e., fear) concerning their prior deviant behaviors at work. The mitigating effects of benevolent leadership on both the fear activation stage and the fear implication stage may be attributed to the complex content of the individualized care that benevolent leaders express. Whereas general support and providing resources can mitigate the detrimental influences of fearful feelings on one's future behaviors, the mistake-forgiving and face-saving aspect of benevolent leadership may also mitigate the intra-person effect of prior deviance on an individual's experience of fear.

Practical implications

Clearly, employee deviance is unavoidable in the workplace, and the occurrence of deviance is usually beyond managers' control given the complex motives within employees who engage in deviance. Thus, the results suggest more managerial attention should be paid to the affective reactions of individuals who execute deviance in the workplace. Specifically, the findings revealed that perpetrators' negative emotion of fear is an inevitable affective outcome of prior deviance, with significant downstream implications for their psychological and behavioral outcomes. The findings highlight the beneficial role of leadership, particularly benevolent leadership, in reducing employees' fearful emotions associated with prior deviance. Moreover, benevolent leadership can also help perpetrators regulate their fearful emotions by preventing them from decreasing future OCB after they experienced fear.

According to Farh, Liang, Chou, and Cheng (2008), specific behaviors managers can display to show benevolence include forgiving employees' mistakes, avoiding the public embarrassment of employees, providing coaching and mentoring, and showing concern for employees' career development. In addition to individualized care in the work context, managers can also show personal care to employees beyond the work domain, such as providing financial or social support to employees during personal crises and maintaining high-quality relationships with employees by treating them as family members (Farh, Tsui, Xin, & Cheng, 1998). In sum, although employee deviance is almost unavoidable in the workplace, leaders can display benevolent leadership to deter the corrosive effects of deviance on perpetrators' work outcomes.

Limitations and future research directions

The present research has several limitations. One limitation is the sole focus of fear as the underlying mechanism linking deviance with the perpetrator's work outcomes. Although this study controlled for guilt as a parallel mediator in all regression analyses, we did not consider other emotions and did not control their influences. For example, past research has suggested that people might feel momentarily excited after engaging in unethical behavior (Ruedy, Moore, Gino, & Schweitzer, 2013). Thus, it is possible that workplace deviance leads to employees' positive emotions as well, which might motivate other outcomes beyond work engagement and OCB. Future research needs to take a more nuanced theoretical perspective to examine the influences of different discrete emotions in the relationships between deviance and the perpetrator's work outcomes.

Another limitation of the present research is that the data were collected in a specific Chinese firm, thus limiting our work's generalizability in two ways. On the one side, for example, China is still a transition economy, and moral regulation and penalty systems are not well established in Chinese firms. In other words, it is usually the leaders rather than formal policies that decide how deviant employees are treated. This condition strengthens the uncertainty perceptions associated with employees' past deviant behaviors. For this reason, the finding that fear overrides guilt in linking prior deviance with future work outcomes should be replicated in other cultural contexts. On the other side, although no evidence has been found in workplace deviance literature that collecting data from a certain firm might contaminate the results, we still encourage future research to retest our theory among more generalized sample participants.

Conclusion

Prior research has revealed the detrimental effects of deviance on both organizational functioning and its employees. More recent work has started to examine the surprisingly positive effects of deviance on the perpetrators themselves. The present research extends this literature by exploring the relationship between deviance and the perpetrator's work engagement and OCB via the mediating role of fear, as contingent on the contextual influence of benevolent leadership. Controlling for guilt, this research demonstrates that deviance is generally associated with reduced work engagement and OCB for the perpetrators because the emotion of fear overrides feelings of guilt. This research emphasizes the roles of emotion and leadership in understanding the effects of deviance on perpetrators' work outcomes.