Abstract
To shed light on the paradoxical phenomenon that third parties of interpersonal mistreatment are motivated to restore justice but often engage in unethical actions, this research differentiates between destructive and constructive punitive reactions while testing the different moderating roles of moral identity and moral thinking orientation. From two studies using different methods and samples from different cultures, we obtained consistent findings that witnessed peer-to-peer incivility triggered moral outrage, which in turn led to both types of punitive reactions. Moral identity strengthened the relationship between witnessed incivility and moral outrage, while rule-based moral thinking orientation weakened the relationship between moral outrage and destructive punitive reaction. Moral identity strengthened the relationship between moral outrage and constructive punitive reaction. The indirect effect of witnessed mistreatment on destructive punitive reaction through moral outrage was strongest among participants with higher moral identity but lower rule-based moral thinking orientation. We discuss the theoretical and practical implications of our findings.
Keywords
Studies on third-party reactions to injustice show that a third party who witnesses mistreatment (i.e., injustice) may exhibit retributive reactions to the injustice perpetrator even though the third party is in no way affected by the event (e.g., Skarlicki & Rupp, 2010; Turillo, Folger, Lavelle, Umphress, & Gee, 2002; Umphress, Simmons, Folger, Ren, & Bobocel, 2013). The deontic perspective of justice (Folger, 1998, 2001), which assumes that people are driven by a moral imperative, helps explain why third parties take action to punish the perpetrator (Folger, Cropanzano, & Goldman, 2005; Turillo et al., 2002). However, the question about how deonance influences the reactions of third parties needs more nuanced scrutiny because, ironically, punishment or retaliation of third parties may be delivered in the same unethical way as the original transgression (Rupp & Bell, 2010; Tripp, Bies, & Aquino, 2002). Is it true that people pursue punishment as an end in itself and that deonance is silent in guiding third-party behavioral reactions? Given that the deontic emotion of third parties is autonomically aroused by observed injustice and then provokes behavioral reactions (Folger et al., 2005; O’Reilly & Aquino, 2011), is it possible that deontic forces function differentially at different stages of the process? The answers to these questions can provide important insights into the role of deonance in reactions to justice/injustice.
Moral identity, defined as the extent to which possessing moral characteristics is central to a person’s self-concept (Aquino & Reed, 2002), has long contributed to the understanding of deonance and its manifestation in the individual reaction to injustice (O’Reilly & Aquino, 2011; Skarlicki & Rupp, 2010). However, there is a potential contradiction in the effect of moral identity on third parties’ unethical reactions to injustice. Specifically, will third parties with high levels of moral identity have more punitive reactions to injustice even when the act is unethical because they are more sensitive to the case of justice violation? Or will high moral identity make them less likely to engage in immoral behavior? These processes may involve two functions of moral identity: heightening sensitivity to injustice occasions and self-regulation. Previous research has either focused on only one function of moral identity (Mitchell, Vogel, & Folger, 2015) or considered only its moderating role in the direct effect of injustice on behavior reaction, which may confuse the different functions of moral identity (Barclay, Whiteside, & Aquino, 2014). To extend this line of inquiry and increase understanding of the influence of deonance, we consider the two functions of moral identity in different stages of third-party reactions (i.e., the former in the deontic emotional reaction and the latter in behavioral reaction) and simultaneously examine their moderating roles in both stages.
To clarify and extend the concept of deonance, Folger and colleagues (Folger & Glerum, 2015; Folger & Stein, 2017) interpret deonance as two “ought” forces: contributive and connotative. The former is associated with the consideration of the instrumental contributions that the behaviors can make to a morally desirable end state, while the latter refers to the perception of the extent to which the act per se is inherently right or wrong. Echoing their work, we examine the role of moral thinking orientation (rule based vs. outcome based; Lammers & Stapel, 2009) in guiding third-party behavioral reactions to witnessed mistreatment. Examining the moderating effects of moral thinking orientation can help extend the depth and breadth of deonance theory, whereas differentiating the moderating effects of moral thinking orientation and moral identity in different stages of third-party deontic reactions to injustice is theoretically important to provide a deeper understanding of the underlying process.
This research also contributes to the justice literature by differentiating two types of third-party punitive reactions. Previous arguments about the typology of third-party reactions remain inconclusive (O’Reilly & Aquino, 2011; Robinson & Bennett, 1997; Romani, Grappi, & Bagozzi, 2013), and few empirical studies have simultaneously examined different punitive reactions. We consider the punitive reaction that deviates from the moral values to harm or humiliate the perpetrator as a destructive punitive reaction (e.g., the action of committing the same mistreatment as the perpetrator based on the principle of “an eye for an eye”). The action is destructive because such punitive reactions may cause the mistreatment to spread and escalate. Answering Hershcovis and colleagues’ (2017) call to explore a broader range of third-party deontic reactions beyond those that are counterproductive, we also examine constructive punitive reactions, or reactions that conform to ethical principles, primarily aim to curb the wrongdoing, and sway the wrongdoer to acknowledge the damage caused and correct his or her behavior (e.g., requesting that the transgressor take corrective action, filing a complaint about the wrongdoing). Both destructive and constructive punitive reactions are similar in that they are motivated by a moral imperative to restore fairness by punishing the transgressor. However, they differ in terms of whether the nature of the act violates ethical rules and whether it presents a moral dilemma to the third party. Thus, examining the differential roles of moral constructs (moral identity and moral thinking orientation) in guiding the choice of destructive or constructive punitive reactions is important.
Last, we respond to Reich and Hershcovis’s (2015) call to validate the moral perspective of justice in the context of peer-to-peer incivility beyond the exclusive focus of justice research on supervisor-to-subordinate (mis)treatment (Skarlicki & Rupp, 2010; Umphress et al., 2013). Peer-to-peer incivility is a more subtle violation of interpersonal rules among coworkers and is ubiquitous in the workplace (Porath & Pearson, 2010). The pervasiveness of peer-to-peer incivility and the large number of potential witnesses warrant an examination of third-party reactions to such incivility (Pearson & Porath, 2005; Skarlicki & Kulik, 2005). From a moral perspective, this study provides further insights into the phenomenon of incivility contagion and spirals by exploring whether observers contribute to an escalatory pattern of interpersonal mistreatment when they are aroused by moral impetus but fail to morally regulate their acts. Figure 1 depicts our conceptual model. Three studies with different designs and distinct samples from different cultures (Chinese and American) demonstrate the generalizability of our findings.
Proposed Model
Theory and Hypotheses Development
In general, extant literature differentiates three motives behind people’s concern for justice. The instrumental perspective proposes that people care about fairness owing to their need to control desirable outcomes (Adams, 1965). The relational perspective centers on people’s desire to be included in social groups (Lind & Tyler, 1988). Finally, the deontic perspective suggests that people care about morality and view justice as the basis of human rights and dignity (Cropanzano, Byrne, Bobocel, & Rupp, 2001). This last perspective explains why a person reacts to unfairness experienced by other people, and these reactions are less likely driven by self-interest (Rupp & Bell, 2010). People are aware of and react to an unfair event that transgresses their moral standards of appropriate conduct (Cropanzano & Rupp, 2002). An increasing number of studies on third-party reactions to injustice (Skarlicki & Rupp, 2010; Turillo et al., 2002; Umphress et al., 2013) have shown that, when witnessing unfair events, people are motivated by moral obligation and duty to hold the perpetrators accountable for their inappropriate acts and to enforce sanctions on them (Folger et al., 2005).
The current research focuses on a third party’s response to peer-to-peer incivility. Workplace incivility refers to low-intensity deviant behaviors that violate the social norm of respect, though incivility may or may not be intended to harm the target, such as making demeaning or derogatory remarks or paying little attention to others’ statements (Anderson & Pearson, 1999). Because incivility is characterized by its low intensity, it is not often explicitly prohibited by policies. As a result, peer-to-peer incivility is a pervasive and increasing phenomenon in the workplace, causing serious harm to both employees and organizations (Pearson & Porath, 2005). As the number of possible third parties for every uncivil encounter can be large (Skarlicki & Kulik, 2005), it is worthwhile to consider how they react to peer-to-peer incivility.
Two Forms of Punitive Reactions
When people observe another’s unfair treatment, they may be motivated to respond actively to punish the offenders (Tripp, Bies, & Aquino, 2007). Research has discussed the typology of punitive responses to injustice but without standardized conceptualization. As the same reaction may be caused by different motivations (Robinson & Bennett, 1997) and the consequences of punitive actions are sometimes beyond the actor’s control, it is important to differentiate different punitive responses according to the nature of the act per se rather than its motivation and consequence. By referencing the conceptualization of destructive and constructive action in the marketing literature (Romani et al., 2013), the discussion on destructive and constructive deviance (Warren, 2003), and the literature on retributive justice (justice by unilaterally inflicting harm or humiliation on offenders) and restorative justice (justice by seeking reaffirmation of some values; Okimoto & Wenzel, 2008; Wenzel, Okimoto, & Cameron, 2012), we distinguish two types of punitive reactions (destructive and constructive) in the context of peer-to-peer incivility and emphasize their difference in whether the act per se violates or conforms to commonly held ethical values.
A destructive punitive reaction is taken to harm or humiliate the perpetrator, and the action per se deviates from the moral principles of how individuals should be treated, such as sabotage (Skarlicki, van Haarsveld, & Walker, 2008), theft, aggression (O’Leary-Kelly, Griffin, & Glew, 1996), and counterproductive work behavior (Fox, Spector, & Miles, 2001). These punitive reactions may not communicate to the perpetrator that his or her behavior is inappropriate, but they will interrupt the peer relations and deteriorate the situation by allowing the uncivil behavior to spread in the workplace. The uncivil interaction is usually “tit for tat” (Anderson & Pearson, 1999; Taylor & Kluemper, 2012), and people are likely to try to get even with a perpetrator by using similar methods or ways that may cause an equal amount of harm to the wrongdoer (Tripp et al., 2002). Therefore, we operationalize destructive punitive reaction in our research context as punishing the perpetrator with the same uncivil treatment.
By contrast, we define constructive punitive reaction as the ethical action taken to warn transgressors to stop the wrongdoing and to elicit corrective action. Constructive punitive reaction does not violate basic moral principles but reaffirms the values and norms of social interaction (Wenzel et al., 2012), such as suggesting reprimanding the unfair behavior, confronting the perpetrator, encouraging the offender to acknowledge his or her wrongdoing, requesting the transgressor to take corrective action (e.g., apology), or filing a complaint about the wrongdoing (Hershcovis et al., 2017; Skarlicki & Rupp, 2010; Umphress et al., 2013). We argue that to restore justice, people will tend to take action, destructive and/or constructive, to punish the transgressor when they observe peer-to-peer incivility.
Hypothesis 1: People will have (a) destructive punitive reaction and/or (b) constructive punitive reaction when they observe peer-to-peer incivility.
According to the deontic model of injustice, although third parties are not directly affected by the uncivil treatment, they may have emotional responses based on moral imperatives (Reich & Hershcovis, 2015), termed “moral anger” or “moral outrage” (Bies & Tripp, 2002). Third parties’ moral outrage is automatically aroused, even though they may not closely identify with the victims or are not harmed by the event themselves (Skarlicki & Rupp, 2010). A great deal of research has shown that behavioral reaction to injustice functions through the impetus of emotion (Folger et al., 2005). When something unfair has happened, third parties almost simultaneously experience moral outrage (O’Reilly & Aquino, 2011), and this moral outrage serves as the basis of punitive reactions (Folger & Glerum, 2015). Therefore, we hypothesize the mediating role of moral outrage in the destructive and constructive punitive reaction to observed peer-to-peer incivility.
Hypothesis 2: Third parties’ observation of peer-to-peer incivility is positively related to their moral outrage.
Hypothesis 3: Third parties’ moral outrage mediates the relationship between observed peer-to-peer incivility and their (a) destructive punitive reaction or (b) constructive punitive reaction.
Moderating Effects of Moral Identity
Moral identity reflects the centrality and importance of moral characteristics in one’s self-concept (Aquino & Reed, 2002). It influences people’s interpretation of social information, affect, and cognitive responses to social cues related to moral issues (Aquino & Reed, 2002). When observing interpersonal mistreatment, people with a strong moral identity are more sensitive to such violations of moral norms and principles (Barclay et al., 2014), and their moral intuition is easily elicited (O’Reilly & Aquino, 2011). High moral identity causes people to experience a strong sense of moral obligation to react to injustice and thus to have more deontic responses to experienced or observed interpersonal mistreatment (Skarlicki & Rupp, 2010; Skarlicki et al., 2008). Given that moral outrage is a deontic response automatically aroused by observed incivility, moral identity should strengthen the relationship between observed incivility and moral outrage.
Hypothesis 4: Third parties’ moral identity moderates the relationship between observed peer-to-peer incivility and their moral outrage such that the relationship is stronger when their moral identity is high rather than low.
Experiencing moral outrage aroused by observed interpersonal mistreatment does not necessarily lead to the same reactions (O’Reilly & Aquino, 2011). As discussed previously, third parties may be inspired to punish the perpetrator and regard this action as a “moral remedy” (Folger et al., 2005), but the punitive reaction may range from being destructive to being constructive. Moral identity has a self-regulation function in the context of emotion-driven behavior (Mitchell et al., 2015). It makes people have a strong commitment to their moral self and motivates them to regulate their moral behavior to maintain self-consistency and the integrity of self-conception (Aquino & Reed, 2002). High moral identity people are motivated to engage in ethical actions that protect the welfare of others, whereas low moral identity people feel less obligation to hold themselves to a moral benchmark, to go out of their way to help others, or to uphold moral values such as justice (for a review, see Jennings, Mitchell, & Hannah, 2015). Thus, third parties’ moral identity will strengthen the relationship between their moral outrage and constructive punitive reaction. Destructive punitive reaction driven by moral outrage is inconsistent with their sense of self; therefore, those with a strong moral identity will experience discomfort and tension that motivate them to reduce the desire to act on their impetus of emotion and thus have less destructive punitive reactions (Mitchell et al., 2015).
Hypothesis 5a: Third parties’ moral identity moderates the relationship between their moral outrage and destructive punitive reaction such that the relationship is weaker when their moral identity is high.
Hypothesis 5b: Third parties’ moral identity moderates the relationship between their moral outrage and constructive punitive reaction such that the relationship is stronger when their moral identity is high.
Moderating Role of Moral Thinking Orientation
In addition to moral identity, we examine the role of moral thinking orientation in guiding a third party’s behavioral reaction to peer-to-peer incivility. Moral thinking orientation provides the perspective from which people engage in moral reasoning; that is, whether an act is morally right depends on its conformity to moral norms (i.e., rule-based moral thinking orientation) or its consequences (i.e., outcome-based moral thinking orientation; Cornelisson, Bashshur, Rode, & Menestrel, 2013). It does not represent the level of moral identity, as people with high moral identity can hold a rule-based or an outcome-based moral thinking orientation. Deonance has a bearing on duty and ought, and the ought forces may come from the rightness of the behavior or the contribution of the behavior to a morally desired state or outcome (Folger & Glerum, 2015; Folger & Stein, 2017). Moral thinking orientation can help us better understand how the relative strength of the two types of ought forces work to influence third parties’ behavioral reactions to injustice.
The focal tenet of rule-based thinking is the concept of duty—that is, morally mandated actions or prohibitions (Tanner, Medin, & Iliev, 2008). Rule-based moral thinking leads people to regard moral values as absolute and unconditional and thus protects them from trade-offs (Ritow & Baron, 1999; Tetlock, Kristel, Elson, & Lerner, 2000). It drives people to stick with moral rules when making decisions and to exhibit high moral behavior consistency across different situations to maintain their self-concept (Baron & Sparanca, 1997; Cornelisson et al., 2013; Tanner et al., 2008). By contrast, people with outcome-based moral thinking (i.e., low in rule-based thinking orientation) tend to appraise the consequences of behavior alternatives, which allows them to be more flexible when encountering a moral dilemma (e.g., Cornelisson et al., 2013; Lammers & Stapel, 2009). These people do not pursue moral perfection, and their attention can easily shift from cultivating a moral image to pursuing other values or outcomes, resulting in the engagement in some immoral behavior (Cornelisson et al., 2013; Tanner et al., 2008).
We argue that third parties’ rule-based moral thinking orientation weakens the relationship between moral outrage and destructive punitive reaction. People with a rule-based moral thinking orientation will adhere to the same strict moral standard to regulate their own behavior (Beauchamp, 2001), believing that everyone should be treated according to the same rules (Lammers & Stapel, 2009). Because morally consistent behavior is important for maintaining their self-concept (Baron & Sparanca, 1997), they will prohibit themselves from inflicting the same uncivil behavior on others. Conversely, people with outcome-based moral thinking orientation are flexible in making decisions and lax in following moral rules (Cornelisson et al., 2013). These individuals believe that sometimes people should be able to work around the rules and make exceptions when necessary (Lammers & Stapel, 2009). In our research context, for angry third parties with an outcome-based moral thinking orientation, the moral imperative may give license to punitive reaction, even if the act itself is not ethical (Folger et al., 2005). Therefore, they are more likely to have destructive punitive reactions.
At the same time, third parties’ rule-based moral thinking orientation may strengthen the relationship between their moral outrage and constructive punitive reaction when observing peer-to-peer incivility. As constructive punitive reaction aims to stop wrongdoing and encourage corrective action without breaking moral rules, rule-based moral thinking orientation does not exert inhibitive moral self-regulation on such action. Rather, this punitive action may become a morally mandated action for rule-based moral thinking people, as they are strongly motivated to protect the moral rules from being violated (Ritow & Baron, 1999; Tetlock et al., 2000).
Hypothesis 6a: Third parties’ moral thinking orientation moderates the relationship between their moral outrage and destructive punitive reaction such that the relationship is weaker when their rule-based moral thinking orientation is high.
Hypothesis 6b: Third parties’ moral thinking orientation moderates the relationship between their moral outrage and constructive punitive reaction such that the relationship is stronger when their rule-based moral thinking orientation is high.
We do not propose a moderating role of moral thinking orientations in the first stage of the indirect relationships (i.e., the relationship between observed peer-to-peer incivility and moral outrage) because, given our research focus, neither the manipulation nor the measurement of peer-to-peer incivility in our studies involves any information about consequences; thus, we do not propose that moral thinking orientation will moderate third parties’ emotional reaction to observed incivility.
In summary, we predict that third parties’ moral identity will make them more sensitive to the occurrence of norm violation in the context of peer-to-peer incivility and that their moral outrage will be more easily elicited. In turn, their moral identity and rule-based moral thinking orientation will motivate them to take action to punish the perpetrator in a more constructive and ethical way. Therefore, the indirect relationship between observed peer-to-peer incivility and constructive punitive reaction will be strongest when third parties’ moral identity and rule-based moral thinking orientation are both high.
Regarding the indirect relationship between observed incivility and third parties’ destructive punitive reaction through their moral outrage, third parties’ moral identity may amplify the first-stage relationship and alleviate the second stage. Therefore, the relative magnitude of the two competing forces will determine the direction of the influence of moral identity on this indirect relationship. While moral identity taps into whether possessing a set of moral qualities is personally important, it does not explicitly explain why people want to be moral or how to be moral (Monin & Jordan, 2009). Barclay et al. (2014) suggest that although moral identity inspires an impetus for action in response to injustice, it lacks the power to guide the direction of the response; thus, another normative framework is required. They argue that the role of moral identity in heightening sensitivity to injustice and triggering reaction is more powerful than its role in inhibiting the destructive punitive reaction. This proposition may hold particularly true in our research context in which the destructive punitive reaction is only a slight deviance from moral norms and may not cause strong discomfort and tension even for high moral identity people. Therefore, this indirect relationship will be stronger when moral identity is high. As we also argue that third parties’ rule-based moral thinking orientation weakens the relationship between their moral outrage and destructive punitive reaction, we integrate the influence of both moral thinking orientation and moral identity and propose the following:
Hypothesis 7a: The indirect effect of observed peer-to-peer incivility on third parties’ destructive punitive reaction through their moral outrage is strongest when their moral identity is high and rule-based moral thinking orientation is low.
Hypothesis 7b: The indirect effect of observed peer-to-peer incivility on third parties’ constructive punitive reaction through their moral outrage is strongest when their moral identity is high and rule-based moral thinking orientation is high.
Study 1
Participants and Procedures
The participants in our study consisted of 145 Chinese working adults who were recruited online by Sojump, an organization in China that helps researchers recruit participants for academic survey projects. All the participants had >6 months of working experience, and 52.6% were women. Most participants were between the ages of 23 and 37 years (76.5%), and 64.8% had completed university.
We manipulated the independent variable (peer-to-peer incivility) by asking all the participants to read a fictitious scenario describing an encounter that they had witnessed between colleagues (P as the perpetrator and V as the victim; Hershcovis et al., 2017; Reich & Hershcovis, 2015; Tripp et al., 2002). The participants were randomly assigned to one of the two experimental conditions (incivility and nonincivility). They all read the following identical information across the two conditions: “Two coworkers, P and V, have been preparing for parallel presentations in front of an audience of 40 people, including coworkers, their boss, their boss’s boss, and managers from other departments, all of whom comprise the New Products Division of their company. The preparation requires a full week, and the feedback from colleagues will be part of the final evaluation of this presentation.” The participants in the incivility condition continued to read the following: “V makes the first presentation. During the presentation, P challenges V’s ideas at several points and uses some derogatory words such as ‘nonsense’ and ‘pointless.’” Conversely, the participants in the nonincivility condition read the following: “V makes the first presentation. During the presentation, P does not interfere and listens carefully to V’s ideas.” After reading the scenarios, the participants completed the measures of manipulation check, moral outrage, destructive and constructive punitive reaction, moral identity, and moral thinking orientation.
Measures
As the participants in this study were all Chinese, all the measures were translated into Chinese via the back-translation technique (Brislin, 1986). Unless otherwise noted, the measures were rated on 7-point Likert scales (1 = strongly disagree, 7 = strongly agree).
Manipulation check
We checked the manipulation of incivility using Colquitt’s (2001) four-item scale. A sample item is “P has refrained from improper remarks or comments” (reverse scored). Cronbach’s alpha was .87.
Moral outrage
The participants were asked to report on a 7-point scale (1 = not at all, 7 = very much) the intensity of their emotional state after reading about the interpersonal encounter, using Lotz, Okimoto, Schlösser, and Fetchenhauer’s (2011) five items. Emotions reflecting moral outrage toward the perpetrator include anger, shock, hostility, distress, and aggravation. The reliability of this scale was satisfactory (α = .96).
Destructive punitive reaction
We measured participants’ intention to punish the perpetrator by using the same uncivil treatment. We used seven items adapted from studies on workplace incivility (Cortina, Magley, Williams, & Langhout, 2001; Lim, Cortina, & Magley, 2008) and tailored them to fit our research context. A sample item is “During P’s presentation, I would make demeaning or derogatory remarks.” Cronbach’s alpha was .94.
Constructive punitive reaction
We used a six-item scale consisting of four items adapted from Skarlicki and Rupp (2010) and two items adapted from Umphress et al. (2013) that fit our definition of constructive punitive reaction. The items are as follows: “I would warn coworkers that P is unfair,” “P should be reprimanded for the way he treated V,” “I would like to sign a letter stating that P did the wrong thing and should be corrected in the future,” “I would be likely to write a letter to log a complaint against P’s uncivil behavior,” “I would like to take corrective action in response to P,” and “I would like to submit a negative comment in response to P’s behavior.” Cronbach’s alpha was .95.
Moral identity
We measured moral identity using Aquino and Reed’s (2002) five-item scale. The participants evaluated the extent to which having some characteristics (e.g., caring, fair, friendly) was important to them. A sample item is “Having these characteristics is an important part of my sense of self.” Cronbach’s alpha was .70.
Moral thinking orientation
We used Lammers and Stapel’s (2009) 12-item scale to assess moral thinking orientation. A sample item is “I believe it is important for everyone to follow the rules.” A higher score represents a rule-based moral thinking orientation, and a lower score indicates an outcome-based moral thinking orientation. We summed the 12 items to form the variable score (α = .70).
Control variables
As negative affectivity may influence self-reported negative features of events and negative affective reactions (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003), we controlled for this variable in our analysis models to lessen the concern with common method bias. We measured negative affectivity using the scale of MacKinnon et al. (1999). Participants were asked to indicate the extent of their general feelings in terms of five negative emotions (i.e., being afraid, upset, nervous, scared, and distressed). Cronbach’s alpha of this scale was .92.
Results
Manipulation check
After we controlled for negative affectivity, participants in the incivility condition reported a higher score of manipulation check than their counterparts in the nonincivility condition, M = 4.98 (SD = 1.25) vs. M = 1.46 (SD = 0.67), F(1, 142) = 461.08, p < .001. The scenario successfully manipulated participants’ perceptions of incivility.
Descriptive analysis and confirmatory factor analysis
Table 1 shows the means, standard deviations, and intercorrelations among the study variables. To evaluate the discriminant validity of the study variables (moral outrage, destructive and constructive punitive reaction, moral identity, moral thinking orientation, and negative affectivity), we conducted a confirmatory factor analysis (CFA) using AMOS. To ensure appropriate assessment of the models with sufficient parameter:sample size ratios (Beauducel & Wittmann, 2005), we used three-item parcels for all variables by randomly assigning the original items to one of the three parcels for each construct. A six-factor measurement model provided satisfactory model fit (χ2 = 293.98, df = 120; comparative fit index [CFI] = .93; incremental fit index [IFI] = .93; standardized root mean square residual [SRMR] = .065). The model fit of the six-factor model was better than that of all the other alternative measurement models, including the five-factor model combining destructive and constructive punitive reactions (χ2 = 375.72, df = 125; CFI = .90; IFI = .90; SRMR = .071) and the five-factor model combining moral identity and moral thinking orientation (χ2 = 408.85, df = 125; CFI = .89; IFI = .89; SRMR = .100). The one-factor model yielded a poor fit (χ2 = 1288.77, df = 136; CFI = .53; IFI = .54; SRMR is not available because of lack of model fit). Chi-square difference tests also showed that the six-factor model had significantly smaller chi-square values than all the other alternative models. Therefore, the study variables are conceptually distinctive.
Means, Standard Deviations, and Correlations for Variables: Studies 1 and 2
Note: Study 1: N = 145, correlations shown below the diagonal. Study 2: N = 161, correlations shown above the diagonal. Incivility: 1 = incivility, 0 = nonincivility. Interaction products were created after variables were mean centered. MID = moral identity; MTO = moral thinking orientation.
p < .05.
p < .01.
Hypotheses testing
The results of the general linear model show that the effects of incivility on destructive punitive reaction, F(1, 142) = 69.36, p < .001, constructive punitive reaction, F(1, 142) = 152.19, p < .001, and moral outrage, F(1, 142) = 158.26, p < .001, were all significant, after we controlled for negative affectivity. The participants in the incivility condition reported higher destructive punitive reaction (3.91 vs. 1.93), constructive punitive reaction (4.76 vs. 1.90), and moral outrage (5.11 vs. 1.78) than their counterparts in the nonincivility condition. Thus, Hypotheses 1a, 1b, and 2 were supported.
We used Hayes’s (2013) PROCESS macro to test the indirect relationships between incivility and punitive reactions through moral outrage (PROCESS Model No. 4), as well as the model (PROCESS Model No. 64) that simultaneously incorporated all hypothesized moderations, including the moderating effect of moral identity in the first stage and the moderating effects of moral identity and moral thinking orientation in the second stage of the indirect relationships. We mean centered the moderators and the mediator first to alleviate collinearity among predictors and to enhance interpretation of the regression results (Aiken & West, 1991; Iacobucci, Schneider, Popovich, & Bakamitsos, 2016). We used the bootstrapping procedures that Preacher, Rucker, and Hayes (2007) outlined to assess the relationships.
As Model 1 in Table 2 shows, the relationship between incivility and moral outrage was significant (b = 3.22, SE = .17, p < .001). The effects of moral outrage on destructive punitive reaction (b = 0.56, SE = .09, p < .001) and constructive punitive reaction (b = 0.58, SE = .07, p < .001) were both significant. The bootstrapping analysis showed that the incivility–moral outrage–destructive punitive reaction indirect effect was significant (estimate = 1.80, SE = .37), as the 95% confidence interval (CI) did not contain zero [1.05, 2.47]. The indirect effect for constructive punitive reaction was also significant (M = 1.86, SE = .35, 95% CI [1.21, 2.60]). Therefore, Hypotheses 3a and 3b were both supported.
Hypothesis Testing Results: Study 1
Note: Standard errors are shown in parentheses; p values are shown in brackets. Moral outrage, moral identity, and rule-based moral orientation were mean centered.
p < .05.
p < .01.
p < .001.
Model 2 in Table 2 tested all hypothesized moderations simultaneously (Model 2a for destructive punitive reaction and Model 2b for constructive punitive reaction). The interaction effect of incivility and moral identity on moral outrage was significant (b = 0.46, SE = .17, p < .01). The interaction, as depicted in Figure 2a, indicated that the relationship between the incivility condition and moral outrage was stronger when moral identity was high (1 SD above the mean; b = 3.66, SE = .24, p < .001) rather than low (1 SD below the mean; b = 2.76, SE = .24, p < .001), in support of Hypothesis 4.
Moderating Effect of Moral Identity and Moral Thinking Orientation: Study 1
The interaction between moral outrage and moral identity was significant for constructive punitive reaction (b = 0.09, SE = .04, p = .026) but not for destructive punitive reaction (b =.05, SE = .05, ns), in support of Hypothesis 5b but not Hypothesis 5a. Figure 2b shows that the relationship between moral outrage and constructive punitive reaction was stronger when moral identity was high (b = 0.66, SE = .06, t = 11.77, p < .001) rather than low (b = 0.48, SE = .10, t = 4.63, p < .001).
For destructive punitive reaction, the interaction between moral outrage and moral thinking orientation was significant (b = −0.19, SE = .08, p < .05). Figure 2c shows that the relationship between moral outrage and destructive punitive reaction was stronger when rule-based moral thinking orientation was low (b = 0.64, SE = .10, t = 6.4, p < .001) rather than high (b = 0.43, SE = .10, t = 4.3, p < .001), in support of Hypothesis 6a. For the model with constructive punitive reaction as the dependent variable, the interaction effect of moral outrage and moral thinking orientation was not significant (b = −0.10, SE = .07, ns). Therefore, Hypothesis 6b was not supported.
We compared the bootstrapping results for the indirect effects across different levels of moral identity and moral thinking orientation (see Table 3). The strongest indirect effect of incivility on destructive punitive reaction through moral outrage occurred when moral identity was high and rule-based moral thinking orientation was low (M = 2.52, SE = .51, 95% CI [1.54, 3.53]). Tests of difference showed that the indirect effect in this condition was significantly stronger than those in the other three conditions (difference = 0.90, 1.46, 0.75; 95% CI [0.04, 1.79], [0.23, 2.52], [0.05, 1.35], respectively). Therefore, Hypothesis 7a was supported. However, Hypothesis 7b was not supported. The indirect relationship for constructive punitive reaction was not strongest when both moral identity and rule-based moral thinking orientation were high. We show that this indirect effect was stronger when moral identity was high, no matter whether moral thinking orientation was low or high (M = 2.61 and 2.21 for the two conditions with high moral identity vs. M = 1.48 and 1.17 for the two conditions with low moral identity). 1
Bootstrapping Results for Simple Indirect Effects: Studies 1–3
Note: High/low = 1 SD above/below the mean. MID = moral identification; MTO = moral thinking orientation.
p < .10.
p < .05.
p < .01.
p < .001.
Discussion
Consistent with the proposed hypotheses, when participants read a scenario involving incivility toward a peer, they experienced moral outrage more intensively, which in turn led to higher intentions for a punitive reaction, both destructive and constructive. Moral identity strengthened the relationship between incivility and moral outrage, while a rule-based moral thinking orientation weakened the relationship between moral outrage and destructive punitive reaction. As a result, the indirect effect of incivility on destructive punitive reaction through moral outrage was strongest when moral identity was strong but moral thinking orientation was more outcome based. Moral identity strengthened the relationship between moral outrage and constructive punitive reaction, but moral thinking orientation did not influence this relationship.
Study 2
In Study 1, participants were asked to read a fictitious scenario and to imagine their reaction. This may lead to concerns about whether the findings reflect the reality of reactions to incivility in the real workplace. Therefore, in Study 2, we integrated a critical incident technique in our survey design by asking participants to recall a recent incident in their actual working experience, and we measured their actual emotional and behavioral reactions. Research on justice recommends the use of the critical incident technique to effectively examine the nature of a specific mistreatment event and witnesses’ reactions (Mitchell et al., 2015). In addition, we used a sample from the United States, a culture distinct from China’s (Hofstede, 1980). Consistent results will serve as evidence of the external validity of Study 1 and generalizability of our findings.
Participants and Procedures
We recruited participants from the United States through Qualtrics, which provides professional online sampling and survey services for researchers. All the participants needed to meet the criteria of having >6 months’ working experience and being >18 years of age. Overall, 161 working adults (49.1% male, 50.9% female) from various age groups participated in the study. Most (62.1%) had completed university or higher.
The participants were randomly assigned to the incivility or nonincivility condition. They were asked to think of a recent incident in which they witnessed an uncivil or civil interpersonal encounter between two of their coworkers in their current or most recent company. In the incivility group, participants were instructed that uncivil interactions could involve rude comments, thoughtless acts, or negative gestures but not intended aggression and physical assaults. In the nonincivility group, participants were asked to recall an event in which one employee treated the other neutrally without uncivil comments and behavior. To ensure that the incidents that participants recalled were specific, clear, and vivid, they answered a series of open-ended questions about the incident, including time, venue, context, process, consequence, and their feelings about it. Those who failed to describe the incident clearly were not included in the data. Thereafter, the participants responded to scale questions on the study variables.
Measures
We assessed all the study variables (i.e., manipulation check, moral outrage, constructive punitive reaction, moral identity, moral thinking orientation, and negative affectivity) except destructive punitive reaction with the same measurement items as in Study 1. The only difference was that the measurement was administered in English. Cronbach’s alphas for these constructs were .98, .92, .95, .83, .75, and .95, respectively. In Study 1, to fit the specific research scenario, we revised Lim and colleagues’ (2008) scale of incivility to measure participants’ destructive punitive reaction. In Study 2, we used its original items. Participants assessed whether they engaged in the behavior described in the items toward the colleague whom they identified as the actor of the uncivil or civil interpersonal treatment. A sample item rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree) is “I put this colleague down or was condescending to him/her.” The reliability was satisfactory (α = .90).
Results
Manipulation check
The manipulation was effective. Participants in the incivility condition reported higher scores for the manipulation check items, M = 4.56 (SD = 1.75), than those in the nonincivility group, M = 0.86 (SD = 1.64), F(1, 158) = 178.51, p < .001, after we controlled for negative affectivity.
Descriptive statistics and CFA
Table 1 also shows the means, standard deviations, and intercorrelations among the study variables of Study 2. We conducted a CFA with three-item parcels for all variables to assess the discriminative validity of the study variables. Consistent with Study 1, the six-factor model fit the data (χ2 = 317.12, df = 120; CFI = .92; IFI = .92; SRMR = .080) better than all the other alternative models, including the five-factor model combining destructive and constructive punitive reactions (χ2 = 401.59, df = 125; CFI = .89; IFI = .89; SRMR = .081) and the five-factor model combining moral identity and moral thinking orientation (χ2 = 421.07, df = 125; CFI = .88; IFI = .88; SRMR = .096). The one-factor model yielded a poor fit. Chi-square difference tests also showed that the six-factor model had significantly smaller chi-square values than all the other alternative models. Therefore, the study variables are conceptually distinctive.
Hypotheses testing
We used the same methods and procedures as in Study 1 to test the hypotheses. The results of a general linear model indicated that participants who recalled an event involving peer-to-peer incivility reported higher destructive punitive reaction, 3.30 vs. 1.79, F(1, 158) = 43.50, p < .001, constructive punitive reaction, 4.34 vs. 1.88, F(1, 158) = 79.49, p < .001, and moral outrage, 4.71 vs. 1.78, F(1, 158) = 161.91, p < .001, than those in the nonincivility group. Thus, Hypotheses 1a, 1b, and 2 were again supported.
Table 4 presents the results of testing the indirect relationship between incivility and punitive reactions through moral outrage (Model 1), as well as the models simultaneously incorporating all hypothesized moderating effects of moral identity and moral thinking orientation (Model 2a for destructive punitive reaction and Model 2b for constructive punitive reaction). Incivility was positively related to moral outrage (b = 2.73, SE = .21, p < .001), and moral outrage was positively and significantly related to both destructive (b = 0.44, SE = .07, p < .001) and constructive (b = 0.65, SE = .08, p < .001) punitive reactions. The bootstrapping estimate of the indirect effect of incivility on destructive punitive reaction through moral outrage was significant (M = 1.21, SE = .23, 95% CI [0.80, 1.74]). We found similar results for the indirect relationship for constructive punitive reaction (M = 1.78, SE = .25, 95% CI [1.32, 2.30]). As in Study 1, Hypotheses 3a and 3b were both supported.
Hypothesis Testing Results: Study 2
Note: Standard errors are shown in parentheses; p values are shown in brackets. Moral outrage, moral identity, and rule-based moral orientation were mean centered.
p < .05.
p < .01.
p < .001.
Consistent with Study 1, the results revealed a significant interaction effect of incivility and moral identity on moral outrage (b = 0.50, SE = .17, p < .01), in support of Hypothesis 4. As Figure 3a depicts, the relationship between incivility and moral outrage was stronger when moral identity was high (b = 3.31, SE = .28, p < .001) rather than low (b = 2.11, SE = .29, p < .001).
Moderating Effect of Moral Identity and Moral Thinking Orientation: Study 2
However, the moderating effects of moral identity on the relationship between moral outrage and punitive reactions were not significant (for destructive punitive reaction: b = 0.002, SE = .04, ns; for constructive punitive reaction: b = 0.02, SE = .05, ns). Thus, Hypotheses 5a and 5b were not supported.
The moderating effect of moral thinking orientation on the relationship between moral outrage and destructive punitive reaction was significant (b = −0.15, SE = .06, p < .05). As Figure 3b illustrates, this relationship was stronger when rule-based moral thinking orientation was low rather than high (b = 0.47, SE = .09, p < .001, vs. b = 0.25, SE = .09, p < .01). Thus, consistent with Study 1, Hypothesis 6a was supported. For constructive punitive reaction, the interaction between moral outrage and moral thinking orientation was not significant (b = −0.04, SE = .07, ns). Thus, Hypothesis 6b was not supported.
As Table 3 shows, Hypothesis 7a was supported. The strongest indirect effect of incivility on destructive punitive reaction through moral outrage occurred among those who possessed high moral identity and a low rule-based moral thinking orientation (M = 1.56, SE = .33, 95% CI [0.96, 2.26]). Tests of difference showed that the indirect effect in this condition was stronger than that in the other conditions (difference = 1.05 and 0.74, 95% CI [0.24, 1.86] and [0.23, 1.25], respectively), except the low moral identity, low rule-based thinking orientation condition (difference = 0.58, 95% CI [–0.15, 1.34]).
As in Study 1, the simple indirect effects comparison did not show support for Hypothesis 7b. The indirect relationship between incivility and constructive punitive reaction through moral outrage was stronger when moral identity was high (M = 2.09, SE = .43, and M = 1.88, SE = .41, for low and high rule-based moral thinking, respectively) rather than low (M = 1.26, SE = .28; M = 1.12, SE = .30). However, the strongest indirect effect did not occur when moral identity and rule-based moral thinking orientation were both high. 2
Discussion
Based on a critical incident technique and a U.S. sample, Study 2 obtained largely consistent results with Study 1. Third parties’ moral outrage mediated the relationships between observed incivility and punitive reactions (destructive and constructive). Third parties’ moral identity moderated the first stage of the indirect relationship (i.e., between observed incivility and moral outrage) but did not influence the second stage. Rule-based thinking orientation weakened the relationship between moral outrage and destructive punitive reaction. Although the findings of Studies 1 and 2 are consistent, the two studies have some limitations. First, they cannot capture employees’ reactions to currently occurring incivility; participants’ reactions were based on a scenario that they read in Study 1 and incidents that they recalled in Study 2. Second, the data in both studies were cross-sectional, which prevents us from evaluating how the effects of observed peer-to-peer incivility unfold over time and also leads to common method variance concerns. Therefore, we conducted Study 3 to overcome these limitations and further test our hypotheses.
Study 3
Participants and Procedures
Previous studies on workplace incivility (e.g., Rosen, Koopman, Grabriel, & Johnson, 2016) suggest that incivility varies on a day-to-day basis. Therefore, Study 3 used a daily investigation to test our hypotheses. In this study, participants were asked to evaluate the incivility encountered by coworkers on the survey days and assess their actual reactions as third parties. They completed measurements of our key variables at different points in time.
Data collection involved two phases. During the first phase, we sent a recruitment email to MBA students enrolled in an organizational behavior course at a university in southern China. We invited those who fit the following requirements to participate in the study: (1) have a full-time job, (2) are able to observe interactions among coworkers, and (3) have regular working daytime hours (e.g., from 8–10
Eighty-one participants completed the sign-up survey and were invited to begin the second phase. In this phase, participants completed two online surveys each day for 10 consecutive workdays. The Time 1 survey, which took place from 12:00
Of the 81 participants who responded to the sign-up survey, two did not fill out the daily surveys in the second phase, and three did not provide valid responses (e.g., missing the survey for more than 7 days, providing the same answer to all questions). Thus, our final sample included 76 participants (93.8% response rate) who provided 759 daily-level data entries. Nearly all participants in the final sample responded to the surveys for all 10 days.
The final sample consisted of 53 women (70%) and 23 men (30%). The participants fell into different age groups, and most (47%) were between 28 and 32 years of age. Most (74%) had obtained a bachelor’s degree or above. They worked in various types of organizations and had different tenures.
Measures
The measurement items of all study variables, except the independent variable (observed peer-to-peer incivility), were the same as in Study 2. In contrast with the first two studies, we measured observed peer-to-peer incivility in this study using Lim and colleagues’ (2008) seven-item scale of incivility. The daily Time 1 survey instructed participants to rate on a 7-point scale (1 = not at all, 7 = very much) the extent to which one colleague demonstrated the behaviors described in the statements to another colleague, based on their observations since they began working on that day. The daily Time 2 survey asked participants to rate the extent to which they had engaged in or were planning on engaging in punitive behaviors (destructive and constructive) toward the perpetrator of the incivility that they evaluated at Time 1 on the same day. Cronbach’s alphas for the scales were all >.70 (see Table 5).
Means, Standard Deviations, and Correlations for Variables: Study 3
Note: Level 1, n = 759. Level 2, n = 76. Gender was coded as male = 1 and female = 0. Age, education, and tenure were coded as the higher the score, the older the participant, the higher the education received, and the longer tenure. Coefficient alphas are on the diagonal.
p < .05.
p < .01.
Analytical Approach
We used multilevel modeling with Mplus 6.12 (Muthén & Muthén, 2012), given the nested nature of the data (daily events nested within subject). First, we conducted a multilevel CFA to test the construct validity. As in Studies 1 and 2, we created three-item parcels for each construct. We included observed incivility, outrage, and the two types of punitive reactions at the within-subject level (Level 1) and personal attributes, including moral thinking orientation, moral identity, and negative affectivity, at the between-subjects level (Level 2). The CFA results indicated that the model fit the data well (χ2 = 155.63, df = 72; CFI = .98; TLI = .97; SRMR = .030 for within-subject level and .025 for between-subjects level). The model fit of this model was better, and the chi-square value was significantly smaller than that in the other alternative measurement models, indicating that the constructs are conceptually distinct.
We used a simultaneous path analysis approach (Preacher & Hayes, 2004; Preacher, Zyphur, & Zhang, 2010) to test the indirect effects and the model incorporating all the moderated mediation relationships, and we adopted the Monte Carlo bootstrapping method with 100,000 replications to derive CIs (Preacher et al., 2010). To reduce any potential bias due to conflated estimates of the between- and within-subject-level components of effects and better interpret our results (MacKinnon, 2008; Ryu, 2015; Zhang, Zyphur, & Preacher, 2009), we group mean centered the Level 1 variable (i.e., observed incivility) and included the group means as predictors in the equations for the dependent variables. We grand mean centered the Level 2 variables.
Results
Table 5 reports the means, standard deviations, and correlations. As the demographic variables were correlated with the study variables, we included them in subsequent analyses. 3 We first conducted a model to test the indirect effects of observed incivility on the two types of punitive reactions through moral outrage, without considering the moderators (see Model 1 in Table 6). Consistent with our hypotheses, observed incivility was positively related to moral outrage (γ = .37, SE = .06, p < .001), and moral outrage was positively associated with both destructive punitive reaction (γ = .21, SE = .05, p < .001) and constructive punitive reaction (γ = .15, SE = .06, p < .01). The results of Monte Carlo bootstrapping provided support for the significant indirect relationships between observed incivility and destructive punitive reaction (M = .08, 95% CI [0.04, 0.13]) and constructive punitive reaction (M = .06, 95% CI [0.01, 0.11]) through moral outrage. Therefore, Hypotheses 1–3 were supported, consistent with Studies 1 and 2.
Hypothesis Testing Results: Study 3
Note: Standard errors are in parentheses; p values are in brackets. Incivility was group mean centered. Moral identity and rule-based moral orientation were grand mean centered.
p < .05.
p < .01.
p < .001.
We then included moral thinking orientation and moral identity in the model to test the hypotheses on the moderated mediation relationships. We allowed moral identity to predict the random slope for the effect of observed incivility on moral outrage and allowed moral identity and moral thinking orientation to predict the random slopes for the effects of moral outrage on destructive and constructive punitive reactions. We also allowed moral identity to predict moral outrage and the two punitive reactions, and we allowed moral thinking orientation to predict the two punitive reactions. As Model 2 in Table 6 shows, moral identity moderated the relationship between observed incivility and moral outrage (γ = .16, SE = .07, p < .05). Figure 4a depicts the pattern of the moderation. The relationship between observed incivility and moral outrage was stronger when moral identity was high (γ = .51, SE = .08, p < .001) rather than low (γ = .26, SE = .08, p < .01). Thus, Hypothesis 4 was again supported. As in Study 2, Hypotheses 5a and 5b were not supported. The interaction effects of moral identity and moral outrage on destructive punitive reaction (γ = .04, SE = .06, ns) and constructive punitive reaction (γ = .01, SE = .06, ns) were not significant.
Moderating Effect of Moral Identity and Moral Thinking Orientation: Study 3
In support of Hypothesis 6a, moral thinking orientation moderated the relationship between moral outrage and destructive punitive reaction (γ = −.22, SE = .05, p < .001). The relationship was significant when rule-based moral thinking orientation was low (γ = .42, SE = .08, p < .001) but not significant when it was high (γ = .06, SE = .06, ns). Figure 4b illustrates the moderation relationship. Regarding constructive punitive reaction, the interaction between moral thinking orientation and moral outrage was nonsignificant (γ = −.11, SE = .09, ns), and Hypothesis 6b was not supported. These results are consistent with those in Studies 1 and 2.
As Table 3 shows, consistent with Studies 1 and 2, Hypothesis 7a was supported, but Hypothesis 7b was not. The indirect relationship between observed incivility and destructive punitive reaction through moral outrage was significant when moral identification was high and rule-based moral thinking orientation was low (estimate = 0.23, SE = .07, 95% CI [0.10, 0.39]). The indirect effect in this condition was stronger than those in the other conditions (Difference 1 = 0.13, SE = .07, p = .073; Difference 2 = 0.22, SE = .08, p < .01; Difference 3 = 0.18, SE = .05, p < .001). However, the strongest indirect effect for constructive punitive reaction did not occur when both moral identification and rule-based thinking orientation were high. Thus, Hypothesis 7b was not supported. 4
General Discussion
Across three studies, we found that moral identity strengthened the relationship between observed incivility and moral outrage and that rule-based moral thinking orientation weakened the relationship between moral outrage and destructive punitive reaction. We found the strongest indirect effect between incivility and destructive punitive reaction among participants with high moral identity and an outcome-based moral thinking orientation. Our results show that third parties’ moral identity and moral thinking orientation exert moderating influences in different stages of their reaction to observed peer-to-peer incivility. Moral identity can heighten people’s sensitivity to injustice and ignite outrage when they observe peer-to-peer incivility, while their moral thinking orientation has more power in regulating their behavioral reactions.
In addition, third parties’ moral identity and moral thinking orientation interact with moral outrage contingent on the type of punitive reaction. Across three studies, we found that third parties’ moral thinking orientation weakened the relationship between their moral outrage and destructive punitive reaction but did not moderate the relationship between their moral outrage and constructive punitive reaction. Third parties’ moral identity strengthened the relationship between their moral outrage and constructive punitive reaction in Study 1 but failed to weaken the relationship between their moral outrage and destructive punitive reaction in all three studies.
Theoretical Implications
This research makes several theoretical contributions, particularly with regard to the role of morality in the deontic perspective of justice (Folger et al., 2005). First, we underscore the importance of one morality-related construct—namely, moral thinking orientation—in regulating third-party punitive reaction to injustice. Our research answers recent calls for studies to extend the depth and breadth of deonance (Folger & Glerum, 2015; Folger & Stein, 2017). Folger and colleagues (Folger & Glerum, 2015; Folger & Stein, 2017) clarify that the deonance approach is not a normative theory and that deontic responses are driven by contributive or connotative ought forces in specific situations. Moral thinking orientation can reflect the relative strength of the ought forces, and our results provide one plausible explanation for the paradoxical phenomenon that deontic reaction to injustice is sometimes delivered as an unethical act. That is, this phenomenon is more likely to occur among people with an outcome-based moral thinking orientation. Outcome-based moral thinking may have a stronger contributive ought force that dictates how the response contributes to a morally desired end state (e.g., the perpetrator is punished) and a weaker connotative ought force that considers the rightness of the act itself. The role of moral thinking orientation deserves further investigation in research on deonance.
Second, to better understand the concept of deonance and its manifestation in third parties’ reaction to incivility, our research differentiates the functions of moral factors in different stages of the reaction process. Justice research has often examined the effect of moral identity, with some studies focusing on its function of heightening sensitivity in the case of moral violation (Barclay et al., 2014; O’Reilly & Aquino, 2011) and others highlighting its role in regulating behavioral reaction (Mitchell et al., 2015). Studies investigating only one of its two functions (e.g., Mitchell et al., 2015) make the role of the other function inconspicuous in that context. Researchers have also usually examined moral identity’s moderating effects on the direct relationship between mistreatment and behavioral reaction (Barclay et al., 2014), which makes us unable to untangle the two roles of moral identity. In this research, we examined the two functions of moral identity in different stages of the process simultaneously. Our results indicate that in the context of peer-to-peer incivility, moral identity is more influential in strengthening the relationship between incivility and moral outrage but less significant in inhibiting destructive punitive reaction driven by moral outrage. It is possible that the destructive punitive reaction operationalized in our research (i.e., uncivil treatment) is only a subtle deviance from the moral standard; thus, the discomfort caused by this act for high moral identity people is not strong enough to trigger self-regulation. The self-regulation function of moral identity may be more powerful when the destructive punitive reaction violates the moral norms more obviously and intensively, such as sabotage or aggressive deviance.
Third, our findings regarding moral thinking orientation lend support to the existence of moral self-regulation in people’s reaction to injustice. Studies have mostly deemed a third party’s deontic responses to injustice as relatively automatic and effortless (Folger et al., 2005; Muller & Skitka, 2006). More recent research suggests that a more elaborate cognitive process is essential to gauge third parties’ actions, and many factors may take on moderating roles in the cognitive process (O’Reilly & Aquino, 2011; Rupp & Bell, 2010). Our findings on the moderating role of moral thinking orientation also reflect the inhibitive function of moral self-regulation that keeps people from violating their own moral rules. To date, research has made only limited attempts to explore the moral self-regulation process in third parties’ reactions to injustice. More studies on this topic are required.
Fourth, this research differentiated between destructive and constructive punitive reactions and found that the moderating roles of moral factors depends on the type of reactions. All three studies showed that moral thinking orientation moderates the influence of moral outrage on destructive punitive reaction but not on constructive punitive reaction. One plausible explanation is that the self-regulation function of moral thinking regulation is operative only when a moral dilemma is present in the case of destructive punitive reaction. It also may be the case that a rule-based moral thinking orientation is more likely to exert inhibitive influence on unethical behavior (destructive punitive reaction) rather than promote ethical behavior (constructive punitive reaction). These explanations deserve further scrutiny. In contrast with moral thinking orientation, moral identity enhanced the relationship between moral outrage and constructive punitive reaction in Study 1. Future research should explore other personal or contextual factors that increase the likelihood of adopting a constructive punitive reaction, such as explicit ethical codes or ethical training in the workplace.
Finally, our findings extend the research on the deontic model of interpersonal justice that primarily focuses on overt supervisor-to-subordinate interpersonal mistreatment in the context of peer-to-peer incivility. Although incivility is characterized by low-intensity deviance and slight violations of social norms, our research found that it can easily ignite an observer’s moral outrage, which in turn leads to punitive reactions. Our findings also suggest that a destructive punitive reaction that is only slightly deviant can present an ethical dilemma to the third party and arouse the inhibitive effect of rule-based moral thinking orientation. Peer-to-peer incivility is an important extension of the deontic perspective of justice (Reich & Hershcovis, 2015), and we call for further research on this prevalent phenomenon from the moral or deontic perspective.
Limitations and Future Research Directions
Despite this study’s contributions, some limitations should be noted. First, we tested our hypotheses using self-reported data, raising a potential concern about common method variance. As common method variance cannot contribute to interaction relationships artificially (Evans, 1985), it is less likely to explain the differential moderating effects of moral thinking orientation for different outcomes in all three studies. Nevertheless, it would be worthwhile to replicate our findings by collecting data from multiple sources.
Second, moral identity has two dimensions (Aquino & Reed, 2002); we measured only the internalization dimension of moral identity in this study. Future studies could improve the research design by controlling for the symbolization dimension (the public aspect of moral identity).
Third, the online recruitment method used in Studies 1 and 2 may raise concerns about the generalizability of our findings. However, as compared with recruiting participants from specific organizations, the online recruitment method can better ensure sample diversity (Gosling, Vazire, Srivastava, & John, 2004), which is appropriate for the hypotheses tested in our research.
Fourth, the correlations between some key variables (e.g., destructive and constructive punitive reactions) were quite high. A possible explanation is that the destructive punitive reaction that we operationalized in the context of peer-to-peer incivility is a low-intensity deviant behavior that only slightly violates ethical norms. The distinction between destructive and constructive punitive reaction might not be very salient. Although the CFA results suggest that they are conceptually distinct constructs, it would be instructive to develop the measures of the constructs with better content and discriminant validity, which would also be useful for future research on this topic.
Finally, other reactions fall into the category of destructive punitive reaction, including aggressiveness and sabotage. It would be valuable for future studies to evaluate our findings by examining other forms of destructive and constructive punitive reactions.
Practical Implications
Our research provides several practical implications for organizations and practitioners in managing uncivil events among their employees. First, peer-to-peer incivility is pervasive but detrimental in the workplace (Cortina et al., 2001; Pearson & Porath, 2005). For any single event of workplace incivility, there are many third parties who potentially have destructive punitive reactions, which bring social and economic costs to the organizations (O’Reilly, Aquino, & Skarlicki, 2016). Organizations should specifically consider intervening in uncivil interactions among peers to curb the escalation of incivility. A zero-tolerance policy should be incorporated to prevent the occurrence of incivility (Hoffman & Chunta, 2015). In addition, managers are role models who can demonstrate organizations’ commitment to the established zero-tolerance policy. To do so, they should develop open and trusting communication channels for reporting incivility and hold perpetrators accountable for their actions.
Second, our findings provide evidence that moral identity can strengthen the relationship between observed incivility and constructive punitive reaction through moral outrage. Previous research suggests that cues in the social environment that make moral concern salient (e.g., slogans, posters, material symbols) or corporate social responsibility programs can activate employees’ moral identity (Mayer, Aquino, Greenbaum, & Kuenzi, 2012; O’Reilly et al., 2016). Therefore, managers can make better use of these workplace cues and corporate programs to induce moral identity in their employees. By applying appropriate human resource procedures, managers can also carefully recruit and select employees with high levels of moral identity, who can serve as ethical role models to others.
Third, given the finding that rule-based moral thinking orientation can weaken the relationship between moral outrage and destructive punitive reaction, managers should find ways to direct employees’ rule-based moral thinking orientation and thus prevent third parties’ destructive punitive reactions. Managers should emphasize the importance of consistently adhering to moral rules. They can involve employees in the conversation about moral issues and reduce the likelihood of cases in which employees compromise moral principles for desired outcomes. An organizational culture of morality can be reinforced by using posters, meeting ground rules, or specific policies regarding civility and treating others with dignity (Miner et al., 2018).
Conclusion
Extending the deontic perspective of justice research to the context of peer-to-peer incivility, we found that observed peer-to-peer incivility ignites a third party’s moral outrage, which in turn leads to punitive reaction to the transgressor. Moral identity strengthens the relationship between peer-to-peer incivility and moral outrage, while a rule-based moral thinking orientation weakens the relationship between moral outrage and destructive punitive reaction. This research deepens understanding of the roles of moral factors in the justice process and calls attention to moral self-regulation.
