It’s Not Personal: A Review and Theoretical Integration of Research on Vicarious Workplace Mistreatment

Third-Party Well-Being

Although predictions in this area are supplemented by other theoretical perspectives, occupational stress models are the most commonly invoked theories in the areas of well-being, attitudinal, and performance outcomes. Applying a stress perspective, vicarious mistreatment is conceptualized as a stressor that provokes psychological and physiological strain responses (Nesse, Bhatnagar, & Young, 2000; Sonnentag & Frese, 2003). Such responses begin to manifest almost immediately following stressors (e.g., stress hormones are released, physiological arousal increases) and when sustained over time, develop into more prolonged outcomes such as decreased mental and physical health, more negative job attitudes, and reduced performance. Though it is acknowledged that many models of occupational stress have been proposed, these theories are considered together throughout the review for two reasons. First, the reviewed studies draw on a multitude of stress models rather than relying on a single stress theory. Second, despite stress models adopting diverse conceptualizations of the process through which stress is perceived, stressors are assumed to evoke similar reactions regardless of the specific stressful stimuli (Selye, 1956).

Considering well-being outcomes more specifically, a common component of stress models is the notion that stressors engender mental and physical health decrements through the more proximal reactions that accompany exposure to stressors, such as excitation of the sympathetic nervous system, production of stress hormones, and mood disturbances (Sonnentag & Frese, 2003). Though many of the instantaneous reactions predicted by these models (e.g., increased cortisol, elevated blood pressure) have not been examined, ample empirical support has been found for the more distal outcomes of increased psychological strain (e.g., Cooper-Thomas et al., 2014; Dupré, Dawe, & Barling, 2014; Hoel, Faragher, & Cooper, 2004; Vartia, 2001), symptoms of anxiety and posttraumatic stress disorder (Fitzpatrick & Wilson, 1999; Hansen, Hogh, Persson, Karlson, Garde, & Ørbæk, 2006), increased emotional labor (Spencer & Rupp, 2009), lower life satisfaction (Glomb, Richman, Hulin, Drasgow, Schneider, & Fitzgerald, 1997; Richman-Hirsh & Glomb, 2002), increased burnout (Miner-Rubino & Cortina, 2007; Totterdell, Hershcovis, Niven, Reich, & Stride, 2012), reduced physical health (Dupré et al., 2014; Hoel et al., 2004; Lim, Cortina, & Magley, 2008), and lower health satisfaction (Glomb et al., 1997; Miner-Rubino & Cortina, 2004; Richman-Hirsh & Glomb, 2002). Further bolstering support for the conclusion that vicarious mistreatment at work is damaging for employee well-being, evidence suggests that the relationships between vicarious mistreatment and several indicators of health hold after accounting for other stressors, including nonwork stressors (Fitzpatrick & Wilson, 1999), general job stress (Lim et al., 2008; Miner-Rubino & Cortina, 2004), and personal mistreatment (Miner-Rubino & Cortina, 2007).

Importantly, recent longitudinal evidence suggests that the well-being consequences of vicarious mistreatment may not persist for long periods of time. Gullander et al. (2014) found no significant relationship between witnessed bullying at baseline and depression at follow-up 2 years later. Hansen, Hogh, Garde, and Persson (2014) similarly found that the relationship between witnessed bullying and sleep difficulties measured 2 years later was not significant after controlling for baseline sleep difficulties. Additionally, although Emdad, Alipour, Hagberg, and Jesen (2013) initially found that observed bullying predicted symptoms of depression 18 months later, Nielsen and Einarsen’s (2013) reanalysis of the data showed that the significant association between witnessed bullying at baseline and prospective depression disappeared when controlling for personal experiences of bullying. Therefore, although cross-sectional evidence supports the linkage between vicarious workplace mistreatment and third-party well-being outcomes, longitudinal evidence, though sparse, does not uphold these conclusions.

Third-Party Attitudinal and Relational Outcomes

Vicarious mistreatment likewise appears to damage several attitudinal outcomes, and stress models are also employed in this domain. Stress models predict that stressors erode attitudinal outcomes because experiences of work stress become infused in appraisals of one’s job and organization (e.g., Dhanani, Beus, & Joseph, 2018). Predictions regarding vicarious mistreatment and attitudinal outcomes have also been supplemented by affective events theory (Weiss & Cropanzano, 1996), which posits that workplace events (e.g., vicarious mistreatment) affect job satisfaction and job performance through emotional reactions. Correspondingly, a number of studies support the empirical linkage between vicarious workplace mistreatment and the attitudinal outcomes of job satisfaction (e.g., Glomb et al., 1997; Lim et al., 2008; Richman-Hirsh & Glomb, 2002), organizational commitment (e.g., Cooper-Thomas et al., 2014; Dupré et al., 2014; Mathisen, Einarsen, & Mykletun, 2008), job involvement (Duffy, Ganster, Shaw, Johnson, & Pagon, 2006), and job frustration (Harris, Harvey, Harris, & Cast, 2013).

Moving beyond stress, the relational third-party response model was developed to account for the deleterious relational outcomes observed in third parties exposed to vicarious mistreatment. Building on the tenants of social exchange theory, Dunford, Jackson, Boss, Tay, and Boss (2015) argued that witnessing other-targeted mistreatment informs third-party perceptions of the organization’s value for justice, and third parties rely on that information to construct relational attitudes toward the organization. When employees perceive the organization to have little value for justice, employees formulate negative social exchange attitudes because of the organization’s failure to fulfill their basic needs (i.e., the need for belonging and affiliation; Tyler & Lind, 1992). This then manifests in more negative relational attitudes toward the organization and ultimately in more negative behaviors. Aligned with the predictions of the relational third-party response model, Dunford et al. demonstrated that witnessed injustice decreased the relational outcomes of organizational trust and organizational identification. Further support for this model comes from research linking vicarious workplace mistreatment to decrements in other indicators of social exchange quality, including trust in management (Duffy et al., 2006; Marrs, 1999) and perceived organizational support (Harris et al., 2013).

When considered together, the literature examining vicarious mistreatment and organizational outcomes has documented a number of deleterious effects on job attitudes and the relationship between the third party and the organization. It is important to note, however, that despite stress and emotions being invoked as the processes through which negative attitudinal outcomes are produced, neither were tested as mediators in any of the reviewed studies.

Third-Party Performance

The effects of vicarious mistreatment on performance outcomes are explained by the impairments in cognitive processing predicted by stress models. Consistent with conservation of resources theory (Hobföll, 1989), for example, stressors are resource consuming because they divert cognitive attention toward sensemaking, appraisal, and regulation processes that accompany exposure to stressors (Sonnentag & Frese, 2003). Two consequences of resource depletion are (a) that less capital is available to devote to resource-intensive tasks and (b) that individuals will seek to protect themselves against further resource loss (Hobföll, 1989). In line with the resource depletion and protection arguments outline above, vicarious mistreatment has been empirically associated with diminished task performance (Cooper-Thomas et al., 2014; Duffy et al., 2006; Woolum et al., 2017), creativity (Mathisen et al., 2008; Porath & Erez, 2009), goal progress (Woolum et al., 2017), and citizenship behaviors (e.g., Dunford et al., 2015), as well as positively linked to withdrawal behaviors and turnover intentions (e.g., Duffy et al., 2006; Dupré et al., 2014; Greenbaum, Mawritz, Mayer, & Priesemuth, 2013; Houshmand, O’Reilly, Robinson, & Wolff, 2012). Though cognitive resource loss has not been directly assessed as an explanatory mechanism underlying these relationships, the mediational pathway through emotional reactions proffered by affective events theory has been supported (Porath & Erez, 2009). The results reviewed above overwhelming support the theoretically expected relationships between vicarious mistreatment and third-party performance outcomes.

Third-Party Interpersonal Responses

The next outcome domain considers the interpersonal responses that third parties may utilize following exposure to vicarious workplace mistreatment. Within this section, we review the literature as it corresponds to two sets of responses: intervention decisions and retributive and restorative behaviors. Intervention decisions are defined as third-party behaviors that attempt to change the mistreatment event, its outcomes, or the future occurrence of mistreatment (Bowes-Sperry & Powell, 1999; Ryan & Wessel, 2012). Intervention strategies can include interrupting the event while it is occurring, reporting mistreatment to the organization, and advising the victim to avoid the perpetrator, among others (Bowes-Sperry & O’Leary-Kelly, 2005). Retributive and restorative behaviors refer to the ways that third parties may seek to restore justice, either through retaliation against the perpetrator or through compensatory behaviors directed toward the victim. Though these domains both represent interpersonally directed behaviors, they are considered separately because each of these behavioral responses is explained by unique theoretical perspectives. In the following subsections, we detail the theoretical explanations and corresponding empirical evidence for both sets of interpersonal responses, beginning first with intervention decisions.

Personal and Vicarious Mistreatment

One interesting line of work is the examination of the effects of being exposed to both personal and vicarious mistreatment on employee outcomes, and expectations and empirical findings in this area have been mixed. Some propose that vicarious mistreatment can amplify the harm of being personally targeted by mistreatment because both personal and vicarious mistreatment represent stressors that elicit simultaneous strain responses (Walsh & Hitlan, 2007). In support of amplification, employees who were targeted by racial harassment and also knew about their coworkers’ racial harassment had poorer attitudinal and health outcomes than employees who had been exposed only to personal or vicarious racial harassment (Low et al., 2007). In a study of the interaction between personal abusive supervision and vicarious abusive supervision, Harris et al. (2013) also found the strongest effects on enacted abuse and perceived organizational support when employees had high levels of personal abusive supervision and high levels of vicarious abusive supervision. In further support, Hitlan, Schneider, and Walsh (2006) assessed the impact of both personal sexual harassment and bystander sexual harassment on ratings of how upset women were by their personal experiences of sexual harassment. Although a multiplicative effect was not supported, women found their personal experiences of sexual harassment to be more upsetting when they also observed the sexual harassment of others.

The opposing perspective has proposed that witnessing negative treatment of others may buffer against the harm of personal negative treatment because it provides employees with normative information that allows them to contextualize their mistreatment. That is, being personally targeted in a context where few others are being mistreated may cause employees to feel singled out and increase internal attributions for the mistreatment, whereas being targeted in an environment with high vicarious mistreatment facilitates external attributions (Farh & Chen, 2014; Schilpzand, Leavitt, & Lim, 2016). In support of a buffering effect, personal abusive supervision had a stronger effect on employee outcomes through organization-based self-esteem when team-level abusive supervision was low rather than high (Farh & Chen, 2014). Furthermore, Duffy et al. (2006) indicated that personal undermining was more damaging for attitudinal, behavioral, and well-being outcomes when there were lower levels of undermining in the workgroup. Extending these findings, Schilpzand, Leavitt, and Lim found that high team levels of vicarious incivility reduced employee self-blame for personal experiences of incivility, which in turn mitigated the impact of experienced incivility on rumination, stress, and withdrawal behaviors.

There are two distinctions between the studies that have found exacerbating and dampening effects that might explain this discrepancy. First, two of the three studies that found that personal and vicarious mistreatment are more detrimental in combination focused on forms of mistreatment that are based on social identities (i.e., discrimination and sexual harassment), whereas none of the studies supporting the buffering effect examined this type of mistreatment. Vicarious discrimination and sexual harassment may magnify the effects of personal experiences because both devalue the employee’s identity from which a sense of self-worth is derived (Tajfel & Turner, 1986). Second, studies that indicate that vicarious mistreatment reduces the harm of personal mistreatment have operationalized vicarious mistreatment as a team-level construct. In a team context, vicarious and personal mistreatment are enacted by a limited number of perpetrators: the members of the team. Experiencing mistreatment from a perpetrator that has been observed to normatively engage in positive interpersonal behaviors signals that the relationship between the perpetrator and the victim is more dysfunctional than the relationship between the perpetrator and others on the team. Outside of a team context, employees may be more likely to witness and experience mistreatment from a different perpetrator or set of perpetrators and may not have social information to make the same normative comparisons. A critical recommended direction for future research is the clarification of the effects of exposure to both personal and vicarious mistreatment and an analysis of additional variables upon which the effect may be contingent, such as those suggested above.

Third-Party Moderators

Responses to vicarious workplace mistreatment may also be a product of dispositional traits of the bystander. In addition to the third-party characteristics already discussed (i.e., just-world beliefs and moral identity), evidence from a recent study on witnessed rudeness also supports the moderating role of third-party core self-evaluations. Individuals’ core self-evaluations reflect their levels of neuroticism, self-efficacy, self-esteem, and locus of control—characteristics that are all associated with resilience to stressors. Employees with high core self-evaluations were thus expected and found to experience less harm when exposed to vicarious mistreatment (Woolum et al., 2017). More specifically, core self-evaluations weakened the negative effect of witnessing rudeness on third-party task performance, goal progress, interaction avoidance, and withdrawal. Moreover, third-party prosocial orientation (Ghumman et al., 2016) and political liberalism (Espinoza & Cunningham, 2010) have also been linked to increased intentions to intervene following vicarious mistreatment.

The moderating role of third-party demographics has also been examined, though results yield an unclear conclusion. Predictions in this area focus on forms of mistreatment that are based on demographic group membership. Women and minorities are expected to have stronger reactions to sexual harassment and discrimination because it devalues a salient social identity and increases perceptions that the third party might also be targeted by future mistreatment (Miner-Rubino & Cortina, 2004). However, despite the conceptual rationale being limited to discrimination and sexual harassment, research has examined the moderating role of third-party demographics across a variety of forms of mistreatment, including those that are not motivated by group membership.

Across all studies, the sex of the third party has been found to both moderate (e.g., Bradley-Geist, Rivera, & Geringer, 2015) and not moderate (Lim et al., 2008; Miner-Rubino & Cortina, 2007) the relationships between observed mistreatment and outcomes, and the nature of the moderation has also varied across studies. Some studies concluded that the harmful effects precipitated by vicarious mistreatment were stronger for women than men (Bradley-Geist et al., 2015; Emdad et al., 2013), whereas one study evidenced a stronger effect for men than women (Miner & Cortina, 2016). Interestingly, the discrepancies do not appear to be explained by differences in the types of mistreatment examined. That is, differentiating between sex-based and non-sex-based mistreatment does not reconcile the inconsistent findings. Sex did (witnessed sexism; Bradley-Geist et al., 2015) and did not (witnessed incivility toward women; Miner-Rubino & Cortina, 2007) moderate the relationships between witnessed sex-based mistreatment and outcomes, and sex did (witnessed bullying; Emdad et al., 2013) and did not (witnessed incivility; Lim et al., 2008) moderate the relationships between non-sex-based mistreatment and outcomes. Finally, only a single study examined witness race as a moderator and found that race did not affect the magnitude of the relationships between ambient racial harassment and attitudinal and well-being outcomes (Chrobot-Mason et al., 2013).

When taken together, the results do not consistently support the notion that women and minorities are more affected by vicarious mistreatment than men and majority group employees and suggest that the moderating effect of third-party demographics may be dependent upon other variables. For example, the outcomes of vicarious mistreatment may be exacerbated only for minority employees working in noninclusive climates or for minority employees who occupy positions of low power as both may increase perceived risk of future personal mistreatment. Subsequent research should test these and other possible explanations for the contradictory findings.

Victim Moderators

Automatic Versus Deliberative Processing Predicts Different Third-Party Reactions

We believe each of the theories of vicarious workplace mistreatment can be described by the extent to which they align with either System I (automatic and heuristic processing) or System II (deliberative and analytical processing). To illustrate, we discuss these linkages and the unique mediational mechanisms and outcome domains associated with each system below. As shown in Figure 1, the vicarious mistreatment process begins with a mistreatment event that can then actuate System I or System II. Beginning with System I, we propose that the stress processes and morally motivated responses associated with vicarious mistreatment align with the automatic and emotional nature of System I processing. In support of this notion, stressor models maintain that immediate strain reactions, such as the release of stress hormones and other biological stress processes, are automatically activated responses that enable rapid and evolutionarily beneficial interpretations of and reactions to threatening stimuli (Nesse et al., 2000). Given that these responses are unconscious and largely fall outside of the third party’s control, the emotional and physical health symptoms elicited through these automatic stress responses thus align with the underlying processing mechanisms involved in System I.

Furthermore, the central tenants of the deontic and morally motivated response models (Folger et al., 2005; O’Reilly & Aquino, 2011) also comport with System I because of the intuitive, heuristic, and emotional nature of these theories. These models argue that vicarious mistreatment exposure activates unconscious, heuristic moral judgments that occur almost immediately after the mistreatment event. The proximal outcome of these automatic and intuitive moral judgments is the elicitation of moral emotional reactions (i.e., moral anger and empathy), which in turn prompts third-party retributive and restorative interpersonal responses. Thus, there is a clear linkage between the underlying theoretical specifications of the deontic and morally motivated response models and System I processing. The proposition that such behavioral responses align with System I has also been corroborated by empirical evidence showing that third parties were more likely to engage in perpetrator punishment when primed to use emotional and unconscious rather than cognitive and deliberative processing (Skarlicki & Rupp, 2010).

Shifting to System II, we propose that the deliberative processing associated with third-party relational outcomes and intervention decisions corresponds with System II processing. First, the relational third-party response model states that third parties use the treatment of others to formulate social exchange attitudes toward the organization. Such evaluations require drawing inferences about the organization’s values, the likelihood the third party will be treated fairly in the future, and the risk of affiliating with the organization (Dunford et al., 2015). Each of these processes require purposeful and active deliberation regarding the third party’s relationship with the organization and are not simply evoked through unconscious emotional reactions to the mistreatment event. This proposition is also consistent with more general theoretical explanations of how justice cognitions are formulated, which predominantly conceptualize justice perceptions as the outcome of a series of cognitive appraisals, social comparisons, and/or counterfactuals (Colquitt, Greenberg, & Zapata-Phelan, 2005). We therefore believe the mediating process of relational justice cognitions as well as the relational outcomes of organizational trust, identification, and other indicators of social exchange quality are best modeled as outcomes of System II.

Furthermore, models of bystander intervention are likewise a product of complex cognitive deliberations that require slower, more analytical appraisals of a number of variables that may change the relative costs and benefits associated with intervening. These deliberations include appraising one’s own personal responsibility to act, drawing inferences about other bystanders’ appraisals of the situation, and appraising whether inaction will enable future instances of mistreatment, among others. Bowes-Sperry and O’Leary-Kelly propose that bystanders appraise each of these factors in a series of decisionmaking stages and describe this sequence as “a rational, controlled decision-making process” (2005: 292). That is not to say that intervention decisions or parts of the intervention decisionmaking process are never decided in a more automatic way. Rather, this argues that bystander intervention decisions are most often a result of a series of deliberative and controlled processes and therefore are most likely a product of System II cognitions.

Finally, we contend that some third-party consequences may be generated through both automatic, emotional processes and deliberative, cognitive processes. This is particularly likely in the case of third-party attitudinal and performance outcomes. Prior theorizing suggests that performance decrements and negative job attitudes could result from immediate emotional responses that divert resources away from task performance and change evaluations of one’s job (i.e., a stressor theory or affective events theory approach to vicarious mistreatment; Sonnentag & Frese, 2003; Weiss & Cropanzano, 1996) or a more rational and deliberative decision to withhold positive inputs from the organization (i.e., a social exchange theory approach to vicarious mistreatment; Dunford et al., 2015). Thus, exposure to vicarious mistreatment may engender immediate attitudinal and performance decrements through emotional and stress processes associated with System I but may show more cumulative damage in the attitudinal and performance domains through more conscious and purposeful cognitive and behavioral responses that align with System II processing.

In addition to the differential outcomes associated with each of the processing modes, it is also possible that the previously summarized moderators align more closely with System I or System II processing. Findings surrounding similarity between the victim and the third party illustrate this potentiality and may explain the unexpected finding that demographic similarity does not appear to affect bystander intervention. Critically, whereas similarity or dissimilarity between the victim and the third party are linked to associative processing (which is automatic in nature and comports with System I), decisions to intervene are more deliberative and are proposed to operate through System II. Thus, the lack of support for demographic similarity may be a product of a misalignment of processing systems. Other boundary conditions may likewise activate more deliberative processing, such as the competitive or cooperative relationship between the victim and the third party. That is, being in competition with the victim may trigger cognitive appraisals (e.g., considering the ways mistreatment may benefit the third party) that supersede moral intuitions of or emotional reactions to the event. Given the nascence of the vicarious mistreatment literature and the lack of prior emphasis on the dual system perspective, more research is needed to understand whether the currently identified moderators (a) affect one or both of the processing routes or (b) influence which processing path is initiated.

Fast Versus Slow Processing Predicts When Third-Party Reactions Develop

Another critical processing characteristic that separates System I from System II is that System I processing is expected to happen quickly, whereas in comparison, System II processing happens more slowly. Because of this feature, the dual processing perspective can also inform the time frame in which each outcome domain is expected to develop. Outcomes associated with System I (i.e., emotional, well-being, and retaliatory responses) are proposed to emerge rapidly through the nearly instantaneous automatic processes that underlie this system. However, because System I is congruent with emotional responses and emotions are theorized to be intense but short lived (Barsade & Gibson, 2007), System I outcomes may dissipate relatively quickly. Findings that vicarious mistreatment is associated with concurrent but not prospective well-being outcomes seem to be consistent with this notion (e.g., Nielsen & Einarsen, 2013). In contrast, because System II requires an active contemplation of the mistreatment event and surrounding conditions, processes associated with this may be characterized by delayed rather than immediate action (Epstein & Pacini, 1999), and associated responses (e.g., appraisals of one’s social exchange relationship with the organization, the decision to intervene) may take longer to manifest in third parties. This is important for guiding future research on vicarious mistreatment because study designs should be consistent with the outcome and processing system researchers are interested in understanding. For example, though emotional reactions may be most appropriately measured minutes or even seconds after exposure to vicarious mistreatment, measuring organizational trust in the same time frame may fail to capture decrements that are slower to develop.

As illustrated by the sections above, a dual process model of vicarious mistreatment reconciles existing theories through the inclusion of both System I and System II processing and can perhaps clarify the processing strategies (i.e., heuristic vs. deliberative processing) and processing speed (i.e., fast vs. slow processing) associated with each of the consequences of vicarious mistreatment. We next consider the likelihood that third parties will utilize System I or System II processing and potential ways to change the processing pathway adopted.

Activating Default Versus Nondefault Processing

System I and System II have thus far been discussed as independent processing routes, but System I and System II are interrelated in multiple ways, as indicated in Figure 1. The interrelationship can derive from System I and System II processing becoming activated in parallel, through System I responses informing subsequent System II processing, or by System II processing intervening to change automatic System I responses to vicarious mistreatment (Evans, 2008), the latter of which is the focus of the current section.

System I is described as the “default” mode of processing because it is the system most frequently engaged and is thought to exert control over most of our behavior (Evans, 2008). System II, however, which is a deliberative form of processing, can interrupt, alter, or overcome the default processing system. We contend that accounting for default versus nondefault processing in vicarious mistreatment is important because most processing of mistreatment events may simply occur within the default processing mode. However, such automatic processing may have unintended and detrimental consequences for organizations, third parties, and victims. These preconscious, heuristic reactions to vicarious mistreatment may include behaviors that are unacceptable in the workplace (e.g., aggression and violence), are counterproductive to performance (e.g., withholding information to punish a perpetrator), or actually punish victims (e.g., negative treatment of victims of mistreatment), all with little to no deliberation or analysis (i.e., without using System II). Thus, we argue an important theoretical and practical consideration is when and how System II may intervene in the processing of vicarious mistreatment and the potential divergent consequences of default versus nondefault processing.

One area that highlights the potential pitfalls of default processing is the degree of similarity or dissimilarity between victims and third parties. As we reviewed, increased similarity between the victim and the third party strengthens affective reactions to vicarious mistreatment (Blader et al., 2013). We believe this bias to react to mistreatment on the basis of the degree of similarity reflects System I processing, which is heuristic and associative in nature. Furthermore, default processing also privileges increased recognition of mistreatment of similar others (i.e., shared group membership is antecedent to knowledge of vicarious mistreatment; Chrobot-Mason et al., 2013) and may therefore result in bystanders failing to acknowledge mistreatment or the need for intervention when victims are dissimilar. In the absence of an intervening deliberative process, it may be the case that biases in recognizing and reacting to vicarious mistreatment are ubiquitous and terminate potentially positive bystander behaviors. For example, even sports fan group membership (a relatively superficial group division) has been shown to suppress helping behaviors for outgroups (Levine, Prosser, Evans, & Reicher, 2005), and these biases have been linked at a neurobiological level to decreased activation of brain areas associated with empathic concern (Hein, Silani, Preuschoff, Batson, & Singer, 2010)—an effect we associate with System I.

Further illustrating the disadvantages of defaulting to System I processing, automatic rather than deliberative processing is associated with responding to vicarious mistreatment with retaliatory behaviors (Skarlicki & Rupp, 2010). Retaliating by withholding resources from or assigning negative work tasks to the perpetrator, for example, is not a productive response to vicarious mistreatment and can exacerbate negative organizational consequences. Triggering deliberative processing that actively considers the harmful effects of retaliation can potentially override these reflexive behavioral tendencies.

Given that System II may offer some advantages over default processing, it is important to understand how to facilitate the activation of System II to mitigate the potential irrational, harmful, or biased outcomes of System I. Bystander intervention training may be one path toward encouraging more conscious and active interpretations of vicarious mistreatment for two reasons. First, bystander intervention training can work to overcome personal schemas of the behaviors that constitute mistreatment and facilitate the recognition of a greater number of inappropriate behaviors (or increase recognition of mistreatment targeting a greater number of victims, including those who are dissimilar to the third party). Second, bystander intervention training can help to define the third party’s role in the vicarious mistreatment process (Holland et al., 2016). That is, bystander intervention training can formally assign third parties a responsibility to intervene in mistreatment. Tasking bystanders with mistreatment prevention can then serve as a rational prime (see Skarlicki & Rupp, 2010, for an example of rational vs. experiential priming), which triggers more deliberative and analytical processing.

In addition to factors that may facilitate deliberative processing, it is also important to consider factors that may inhibit System II. For example, working environments with high demands and high stress may preempt deliberative processing because employees have too few resources available to devote to analytical appraisals. This would suggest that characteristics of the working environment might be both antecedent to the occurrence of vicarious mistreatment and change the processing system activated by vicarious mistreatment. Finally, we note that System II is not without its own cognitive biases (Evans, 2008). However, System II processing may still be preferable in organizational settings (compared to System I) because it leads third parties to attend to more information and perhaps more carefully consider the consequences of their reactions. The above arguments remain speculative, and future research is encouraged to test these and other intervention methods that may redirect third-party processing toward System II.

Characteristics of Vicarious Mistreatment and the Dual Process Model

Prior to concluding, it is important to note that the current review adopted a broad definition of workplace mistreatment that includes manifestations of negative interpersonal behaviors that vary in ambiguity, severity, source, and motivation. Though these are considered together throughout this paper, it is possible that the proposed dual process model differs as a function of the specific form of mistreatment. Differentiating between group-based (i.e., mistreatment based on demographic group membership) and non-group-based (i.e., mistreatment not motivated by group membership) forms of vicarious mistreatment may be of particular importance because group-based mistreatment is proposed to be associated with unique automatic and deliberative processes more so than other forms of mistreatment. For example, research demonstrates that implicit biases (which are activated automatically and unconsciously) can alter associative processing and have been shown to result in poorer outcomes for minority groups (Green et al., 2007; Ziegert & Hanges, 2005). We contend that implicit bias is one mechanism that is unique to group-based mistreatment and can reduce recognition of and affective responses to mistreatment as well as increase victim blaming when negatively stereotyped groups are the targets of mistreatment. Group-based mistreatment also has unique considerations that may affect System II processing, though the result is expected to be the same as the inhibitory processes described for System I. Primarily, third-party deliberative appraisal processes may be influenced by explicit negative beliefs about women and minorities or any number of legitimizing cognitions that are used to support the pervasive social-structural inequities between minority and majority groups. Third parties may sanction sex discrimination in promotion decisions, for example, through the explicit belief that women do not possess the agentic traits stereotypically associated with leadership. This can allow third parties to frame the mistreatment as a product of merit rather than a product of discrimination, a key legitimizing myth pinpointed in social dominance theory (Pratto, Sidanius, Stallworth, & Malle, 1994). Third parties may therefore be less affected by mistreatment justified by such ideologies and may even be more likely to hold the victim rather than the perpetrator accountable.

Group membership is not the only characteristic of vicarious mistreatment that may affect third-party responses. Consistent with work in the area of bystander intervention (e.g., Ryan & Wessel, 2012), we assert that severity and ambiguity are two other features of vicarious mistreatment that may affect the salience of mistreatment for third parties. Witnessing severe forms of mistreatment may trigger stronger third-party emotional responses because (1) victims may have stronger emotional reactions that become diffused to third parties through empathy or emotional contagion processes and (2) third parties may intuit a stronger moral violation in response to severe mistreatment. System II cognitions that can justify vicarious mistreatment or inhibit third-party responses (e.g., being in competition with the victim) may also be less likely to overcome System I processing in the case of more severe behaviors. Ambiguous mistreatment may operate in the opposing way; it may be more difficult to conclude whether ambiguous mistreatment violates moral principles, and witnesses may be more likely to justify ambiguous mistreatment. As an example, knowing that the victim has a relationship with the perpetrator—a variable thought to decrease bystander intervention—may avert emotional and behavioral responses to witnessed rudeness (e.g., a form of mistreatment low in severity and high in ambiguity) because the third party may dismiss the behavior as unintentional. However, third-party responses to physical violence (e.g., a form of mistreatment high in severity and low in ambiguity) may not be curbed when bystanders have the same knowledge of the victim-perpetrator relationship because the behavior is severe enough that an informal relationship cannot justify it. These proposed differences, though not exhaustive, illustrate the importance of empirically examining the extent to which the proposed model of vicarious mistreatment is stable across different forms of mistreatment.

In closing, our proposed dual process framework advances that bystanders can interpret vicarious mistreatment using automatic versus deliberative, fast versus slow, and default versus nondefault processing and that each processing system is associated with different outcomes and biases. We further maintain that the specific manifestation of vicarious mistreatment may act on the dual systems in distinctive ways. The review next concludes with recommendations for ways to test and build upon the model described above to advance our understanding of vicarious workplace mistreatment.

Extending Potential Antecedents and Consequences

Extant work has devoted comparatively little attention to the antecedents of vicarious mistreatment and, as such, future research would benefit from examining the variables that precede vicarious mistreatment. In particular, subsequent theoretical and empirical efforts would be advantaged by explicating the variables that affect whether third parties recognize an interaction as mistreatment. Drawing from an important distinction in the justice literature (Cropanzano, Byrne, Bobocel, & Rupp, 2001), we propose that future work should differentiate between and simultaneously consider event appraisals (e.g., perceptions that a recent organizational decision was fair) and entity appraisals (e.g., perceptions that a supervisor is a fair or unfair person). Despite extant work focusing on perceptions of the mistreatment event itself, third parties may also use information about the actors or entities involved to determine whether mistreatment has occurred, and entity appraisals may even supersede event appraisals.

Applying the dual process framework, entity perceptions can set expectations about the types of behaviors a person typically engages in and, through heuristic processing, may enable third-party judgments about the actor’s behavior that ignore the specific features of the mistreatment event. This is consistent with Jones and Skarlicki’s (2005) finding that a person’s reputation for unfairness colored perceptions of subsequent behaviors such that even after controlling for the actual behavior of an actor, one’s reputation for unfairness led to lower perceptions of interactional justice. Conversely, perceptions that entities are fair may (unduly) increase perceptions of fairness of their behaviors and impair the recognition of mistreatment perpetrated by such sources. This suggests that third parties can rely on their stable perceptions of involved parties to formulate mistreatment perceptions, potentially without attending to characteristics of the event, and that prior information about actors can create perceptual biases that affect subsequent information processing.

We also recommend expanding research on the outcomes of vicarious mistreatment. In addition to confirming the proposition that System I and System II are linked to different outcome domains, research should broaden the scope of examined consequences to include the ways in which vicarious mistreatment may affect off-the-job outcomes. Given that boundaries constructed around the work and nonwork domains are permeable, experiences at work can have cross-domain effects. Vicarious workplace mistreatment may be one such experience that can spill over into the nonwork domain through both System I (e.g., work-to-family depletion through negative mood spillover; Williams & Alliger, 1994) and System II (e.g., work-to-family spillover decreasing family satisfaction; Liu, Kwan, Lee, & Hui, 2013). A consideration of the cross-domain effects of vicarious mistreatment should be integrated into future studies.

Extending Potential Moderating Variables

We also advocate for future research that explores additional moderating variables that may affect third-party outcomes of vicarious mistreatment. Namely, though studies often do not consider the channel through which vicarious mistreatment is disseminated, bystanders can learn of vicarious mistreatment directly (i.e., witnessing the event firsthand) or indirectly (i.e., secondhand knowledge of mistreatment’s occurrence through word of mouth, social media, etc.), and it is conceivable that these sources have a differential impact on third parties. In particular, direct observations may evoke stronger emotional reactions (e.g., stronger empathetic concern resulting from witnessing the victim’s harm firsthand) than indirect knowledge through the activation of System I. Indirect accounts of vicarious mistreatment may conversely elicit more deliberative processing since the provoking event is more distal and since recounts of mistreatment introduce a greater amount of information to process. That is, in addition to characteristics of the mistreatment event, perpetrator, and victim, these cases require third parties attend to information about the source through which they learned about vicarious mistreatment (e.g., whether the indirect source is trustworthy) in the deliberation process.

Moreover, direct and indirect vicarious mistreatment may also differ in the number of third parties that may become aware of the mistreatment event. Indirect recounts, particularly those disseminated through social media, can reach a greater number of employees and customers than would likely be directly exposed to vicarious mistreatment and, thus, may be more damaging to organizations. Future research should examine whether the number of bystanders affected by vicarious mistreatment, the magnitude of those effects, and the processing system activated vary as a function of the direct or indirect mode through which mistreatment was diffused to third parties.

Additionally, research has examined relatively few third-party dispositional moderators, and we propose that third-party personality may affect bystander responses to vicarious mistreatment. In particular, conscientiousness may predict bystanders’ willingness to intervene, given that this trait has been linked to many positive workplace behaviors, including helping behaviors directed at coworkers and the organization (Organ & Ryan, 1995). Likewise, third parties high in need for cognition may be more likely to engage in System II processing strategies and actively acquire and evaluate information (Cacioppo, Petty, Feinstein, & Jarvis, 1996), which may lead to more intervention. In contrast, neurotic individuals may be more likely to react with strong negative emotions and to process vicarious mistreatment using System I. Thus, whereas individuals high in conscientiousness and need for cognition may respond to vicarious mistreatment in constructive ways, third parties high on neuroticism may use responses that are counterproductive. Other third-party dispositional variables (e.g., agreeableness, authoritarianism, social dominance orientation) should also be explored.

Methodological Recommendations

Before concluding, it is worthwhile to discuss some potential research design and methodological limitations in the existing literature that may undermine current conclusions and future progress. As proposed by the dual process model, outcomes are expected to develop in different time frames depending on the system activated, and third-party responses to vicarious mistreatment may change as System II processing unfolds. Despite the proposition that some outcomes may be delayed or show variation across time, the majority of the reviewed studies rely on survey designs that measure variables at a single time point and/or measure outcomes close in time to the third party’s exposure to vicarious mistreatment. Future work should match the research design to the specific research purposes, and when the interest is in System II outcomes or how System II can influence System I processing, repeated-measures or time-lagged designs may be advantageous. Within-person, repeated-measures designs can also confirm the expectation that System I outcomes develop more rapidly than System II outcomes, as forwarded by the proposed dual process model, and can sequence the reactions that develop more proximally and distally in the vicarious mistreatment process.

Confirming the notion that vicarious mistreatment can activate both System I and System II processing and that each processing mode can provoke differential outcomes may also require moving beyond traditional, self-report studies and toward methods that can better examine processing patterns (e.g., functional MRI studies). Recent evidence from the field of social cognitive neuroscience indicates that each form of processing activates unique brain areas, with System I activating the amygdala, basal ganglia, and lateral temporal cortex (regions associated with conditioning and implicit learning) and System II activating the anterior cingulate cortex, prefrontal cortex, and the medial-temporal lobe (regions linked to explicit learning and executive control; as reviewed in Evans, 2008). Examining activation in these brain areas can assess how vicarious mistreatment is processed under different conditions (i.e., what triggers System I and System II activation) and substantiate the claim that some processing of vicarious mistreatment happens rapidly and some is more delayed.

Finally, the vicarious mistreatment literature is limited by the quality and type of mistreatment measures used. First, measures of vicarious mistreatment do not always clearly delineate between personally experienced mistreatment and observations of mistreatment targeting another party. That is, some vicarious mistreatment measures assess the general occurrence of mistreatment and could be endorsed because of personal or vicarious experiences. Such measures hinder the ability to draw conclusions about the true impact of being a nontargeted observer to mistreatment. Second, many studies use direct measures of vicarious mistreatment that require third parties to label their vicarious experiences as mistreatment, and these measures are associated with more reluctance to endorse items in comparison to behavioral measures of mistreatment (see Ilies, Hauserman, Schwochau, & Stibal, 2003, for a discussion of direct vs. behavioral measures). More problematic than the general limitations associated with direct measures of vicarious mistreatment is the use of disparate measurement strategies when comparing the effects of personal and vicarious mistreatment. Of the four reviewed studies that concluded personal mistreatment is more harmful than vicarious mistreatment, three used direct measures of vicarious mistreatment but behavioral measures of personal mistreatment (Cooper-Thomas et al., 2014; Hansen et al., 2006; Hauge et al., 2007). Thus, it is unclear whether differences between personal and vicarious mistreatment are a product of true differences or measurement artifacts.