Organizational deviance is voluntary behavior that violates important organizational and societal norms and threatens the well-being of an organization (Robinson & Bennett, Reference Robinson and Bennett1995). The annual cost of workplace deviance on the American economy has been estimated to be as high as $120 billion for workplace violence (Deyle, Reference Deyle2015), $44 billion for theft (National Retail Federation, 2015), up to $200 billion for delinquent behavior (Bennett & Robinson, Reference Bennett and Robinson2000), and $550 billion in disengaged employees (Gallup, Reference Gallup2014). Observed misconduct is not rare. In recent surveys, 41% of employees in the US (Ethics Resource Center, 2014) and approximately one-third of employees worldwide (Ethics and Compliance Initiative, 2016) observed misconduct in the workplace. In response, scholars and practitioners have called for ways to reduce deviant behavior in organizations (Lawrence & Robinson, Reference Lawrence and Robinson2007).
Ethical leadership provides one such avenue to reduce deviance. Employees rely on ethical leaders to set the overall tone in dealing with ethical issues in organizations (Treviño, Hartman, & Brown, Reference Treviño, Hartman and Brown2000). Brown, Treviño, and Harrison (Reference Brown, Treviño and Harrison2005) define ethical leadership as, “the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication, reinforcement, and decision making” (120). Ethical leadership is related to a variety of important outcomes including organizational deviance and counterproductive behaviors (see reviews by Bedi, Alpaslan, & Green, Reference Bedi, Alpaslan and Green2016; Ng & Feldman, Reference Ng and Feldman2015).
To date, of the many studies published on ethical leadership and important outcomes such as deviance, none have considered supervisors’ perspectives of their ethical leadership. Self-perceptions, however, should be important because studies on other aspects of leadership have shown that leader and subordinate perceptions typically diverge and this lack of agreement has important implications for subordinate attitudes and behaviors (Fleenor, Smither, Atwater, Braddy, & Sturm, Reference Fleenor, Smither, Atwater, Braddy and Sturm2010). Furthermore, individuals often overestimate their own ethicality leading to cognitive blind spots that can contribute to unethical behavior (Bazerman & Tenbrunsel, Reference Bazerman and Tenbrunsel2011). To complicate matters, some of the qualities of ethical leaders such as integrity and honesty are difficult for followers to observe (Moore & Small, Reference Moore and Small2007). Thus, it seems likely that supervisors and subordinates will frequently have differing perceptions of the supervisor’s ethical leadership and such disagreement could have implications for subordinates’ behavior.
Disagreement in how leaders rate their own performance (e.g., style and effectiveness) relative to how they are rated by followers (especially when leaders overestimate) is associated with less favorable follower job attitudes and performance (Atwater, Roush, & Fischthal, Reference Atwater, Roush and Fischthal1995; Bass & Yammarino, Reference Bass and Yammarino1991; Yammarino & Atwater, Reference Yammarino and Atwater1997). One explanation is that individuals who have enhanced self-perceptions may pursue tasks for which they are not well-suited or may ignore risks (Taylor & Brown, Reference Taylor and Brown1988). In addition, individuals with enhanced self-perceptions are more likely to follow their own perceptions, ignoring information from others (Atwater & Yammarino, Reference Atwater, Yammarino and Ferris1997), and are more resistant to constructive feedback (Brett & Atwater, Reference Brett and Atwater2001). Also, they may not be perceived as good role models (Bandura, Reference Bandura1977, Reference Bandura1986).
In this research, we set out to answer two questions. First, what happens when subordinates and supervisors do not agree about the supervisors’ ethical leadership? We focus on the relationship between supervisor-subordinate disagreement on ethical leadership and organizational deviance because the link between follower-rated ethical leadership and deviance is already established in the literature. More specifically, ethical leadership is negatively related to deviant behaviors such as group deviance (Mayer, Aquino, Greenbaum, & Kuenzi, Reference Mayer, Aquino, Greenbaum and Kuenzi2012), unethical behavior (Mayer et al., Reference Mayer, Aquino, Greenbaum and Kuenzi2012), as well as other negative behaviors (Bedi et al., Reference Bedi, Alpaslan and Green2016). Therefore, we first examine whether supervisor-subordinate disagreement is related to unit deviance.
If disagreement on the supervisor’s ethical leadership is associated with negative outcomes like deviance, it is also important to understand what factors might contribute to disagreement between supervisors and subordinates. Therefore, in a separate analysis, we address a second question—how are supervisors’ beliefs related to disagreement? We consider how such beliefs are related to disagreement not only by affecting leaders’ perceptions of themselves, but also by influencing how supervisors are viewed by their subordinates. Specifically, we hypothesize that supervisors’ beliefs about their own relative performance as ethical leaders (a type of better-than-average [BTA] belief; Moore, Reference Moore2007) as well as beliefs that they cannot shape and improve the ethics of others (supervisor moral entity orientation; Dweck, Chiu, & Hong, Reference Dweck, Chiu and Hong1995) are associated with (dis)agreement on ethical leadership.
We had three considerations when choosing these supervisor beliefs as antecedents. First, we searched for antecedents that were relevant to our social learning theoretical framework (Bandura, Reference Bandura1986) and could potentially affect leader role modeling which is an important component of ethical leadership (Brown et al., Reference Brown, Treviño and Harrison2005). Second, we wanted to focus on ethics-related factors as opposed to more general features of personality that have been studied in previous ethical leadership research (Walumbwa & Schaubroeck, Reference Walumbwa and Schaubroeck2009). Third, we looked for antecedents that would likely contribute to disagreement on ethical leadership. We believe supervisor’s BTA and moral entity theory beliefs were logical choices based on these criteria. BTA beliefs are common in the moral domain (Tappin & McKay, Reference Tappin and McKay2017) and, like other types of self enhancement, have potentially negative implications for ethics in general (Sedikides & Gregg, Reference Sedikides and Gregg2008) and ethical leadership in particular (Messick & Bazerman, Reference Messick and Bazerman1996). BTA beliefs are often unrealistically positive (Alicke & Govorun, Reference Alicke, Govorun, Alicke, Dunning and Krueger2005), so there is great potential for them to contribute to supervisor-subordinate disagreement on ethical leadership. Such beliefs are relevant to a social learning perspective on ethical leadership (Brown et al., Reference Brown, Treviño and Harrison2005, Treviño et al., Reference Treviño, Hartman and Brown2000) because they create ethical blind spots (Tenbrunsel & Bazerman, Reference Tenbrunsel and Bazerman2011) which can hinder moral persons from engaging in the kind of proactive ethical role modeling that is necessary to be seen as ethical (as opposed to ethically neutral) leaders (Treviño et al., Reference Treviño, Hartman and Brown2000).
Likewise, a moral entity orientation has obvious implications for ethics given that it reflects a supervisor’s belief that ethics is immutable. From a social learning perspective, a moral entity orientation will impact how much (and how little) time supervisors dedicate to developing the ethical behavior of subordinates through core moral management behaviors (Treviño et al., Reference Treviño, Hartman and Brown2000) such as communicating ethical standards and ethical role modeling. Given that individuals are usually not aware of their own theories about moral mutability (Dweck et al., Reference Dweck, Chiu and Hong1995), a moral entity orientation might contribute to supervisor-subordinate disagreement by impacting subordinates’ ratings but not a supervisor’s own self-rating of ethical leadership.
Overall, we draw from multiple social influence theories that have been used in prior ethical leadership research to ground our hypotheses, including social exchange theory (Blau, Reference Blau1964), social information processing theory (Salancik & Pfeffer, Reference Salancik and Pfeffer1978), and especially social learning theory (Bandura, Reference Bandura1977, Reference Bandura1986). Our study contributes to the ethical leadership, self-other agreement, leader humility, and ethical biases literatures. First, we examine whether there is a discrepancy in how supervisors rate their own ethical leadership compared to how they are rated by subordinates. Given the extensive research showing that people are not as ethical as they think they are (Tenbrunsel, Diekmann, Wade-Benzoni, & Bazerman, Reference Tenbrunsel, Diekmann, Wade-Benzoni and Bazerman2010), it is informative to examine ethical leadership from both parties’ perspectives. Second, we examine the relationship between different types of disagreement (e.g., when leader ratings are higher than follower ratings and when follower ratings are higher than leader ratings) and an important outcome, organizational deviance—a novel dependent variable to the self-other agreement literature. Third, by looking at all types of (dis)agreement, including the relatively less common leader who does not suffer from overinflated self-ratings, our study contributes to the emerging literature on leader humility (Owens, Johnson, & Mitchell, Reference Owens, Johnson and Mitchell2013). Fourth, we add to the ethical leadership literature by linking supervisors’ beliefs about ethics to (dis)agreement on ethical leadership. Furthermore, we propose that such beliefs not only have the potential to increase disagreement by affecting leaders’ perceptions of themselves, but also can affect how supervisors are seen by their subordinates. Finally, this article responds to a timely and important issue in the behavioral ethics literature. Specifically, we investigate and add to the growing literature on ethical biases by empirically testing theoretical work linking ethical bias to (un)ethical leadership (Bazerman & Tenbrunsel, Reference Bazerman and Tenbrunsel2011; Brown, Reference Brown2007; Messick & Bazerman, Reference Messick and Bazerman1996).
FOUNDATIONS OF ETHICAL LEADERSHIP
Treviño et al. (Reference Treviño, Hartman and Brown2000) proposed ethical leadership is based on two different pillars—the moral person and the moral manager. The moral person component focuses on desirable personal qualities of leaders such as being perceived as honest, fair, and trustworthy. The moral manager component focuses on the leader using transactional efforts such as rewards and punishments to reinforce desired behaviors and reduce undesired behaviors, as well as communicating about the importance of ethics. When a leader is strong in both of these areas, the leader is considered to be an ethical leader. Brown et al. (Reference Brown, Treviño and Harrison2005) describe ethical leadership as involving both personal demonstration and active promotion of ethical conduct to employees in the organization. These behaviors and messages of supervisors impact the (un)ethical behaviors of subordinates through social influence processes.
There are several social influence theories that have been used to understand how ethical leaders influence their followers, including social learning theory (SLT; Bandura, Reference Bandura1977, Reference Bandura1986), social exchange theory (SET; Blau, Reference Blau1964), and social information processing theory (SIPT; Salancik & Pfeffer, Reference Salancik and Pfeffer1978). SLT suggests that subordinates learn from behavior modeled by supervisors as well as from the ethical standards that are communicated and reinforced in the workplace. According to SET, when supervisors engage in positive and negative behaviors towards others, they create an environment where subordinates will likely reciprocate positively or negatively. SIPT posits that individuals use cues from their work environment to understand appropriate ways to behave in the organization. Based on how many, what kind of, and how consistent these cues are, they become more or less salient to subordinates and impact the norms surrounding ethics. For instance, if the supervisor consistently displays unethical behavior and rewards individuals for behaving unethically these unethical norms become salient, impacting deviant behavior in subordinates (Mayer, Kuenzi, Greenbaum, Bardes, Salvador, Reference Mayer, Kuenzi, Greenbaum, Bardes and Salvador2009; Mayer, Nurmohamed, Treviño, Shapiro, & Schminke, Reference Mayer, Nurmohamed, Treviño, Shapiro and Schminke2013). The more salient and consistent the cues are, the stronger the norms for (un)ethical behavior will be to subordinates.
SUPERVISOR-SUBORDINATE (DIS)AGREEMENT AND DEVIANCE IN WORK GROUPS
Because of the prevalence and scope of deviance in the workplace, we focus on the question of how leadership can limit or curtail costly deviant behaviors. Although deviance is an individual behavior, because of the interactions occurring within the social context of a work group, deviance can be perceived and operate as a distinct construct at the group level (see Chan, Reference Chan1998, and Kozlowski & Klein, Reference Kozlowski, Klein, Klein and Kozlowski2000, for reviews as to how group constructs emerge). As many organizations utilize teams and/or departments to function, it is important to examine how the leadership influences group behaviors such as deviance. Therefore, consistent with previous theory and research examining leadership, specifically ethical leadership and group deviance (e.g., Brown & Treviño, Reference Brown and Treviño2006; Mayer et al., Reference Mayer, Kuenzi, Greenbaum, Bardes and Salvador2009; Robison & O’Leary-Kelly, Reference Robinson and O’Leary-Kelly1998) we examine employee deviance in the work group as our outcome of interest. We define group organizational deviance as voluntary behavior by members of a work group that violates the norms and threatens the well-being of the group (Robinson & Bennett, Reference Robinson and Bennett1995, Reference Robinson and Bennett1997).
The relationships between ethical leadership and important outcomes in organizations such as deviance are well documented (Bedi et al., Reference Bedi, Alpaslan and Green2016; Ng & Feldman, Reference Ng and Feldman2015). Although many studies have found that ethical leadership is negatively related to (un)ethical behaviors (Mayer et al., Reference Mayer, Kuenzi, Greenbaum, Bardes and Salvador2009; Ng & Feldman, Reference Ng and Feldman2015), we believe this relationship is more complex than has been previously examined. Research on leadership and self-other agreement has found that supervisor and subordinate perceptions of the supervisor often disagree (e.g., Dansereau, Graen, & Haga, Reference Dansereau, Graen and Haga1975; Harris & Schaubroeck, Reference Harris and Schaubroeck1988). This disagreement is related to various employee behaviors including performance (Atwater, Ostroff, Yammarino, & Fleenor, Reference Atwater, Ostroff, Yammarino and Fleenor1998), fairness (Jeffcoat, Reference Jeffcoat2000), job satisfaction, and commitment (Szell & Henderson, Reference Szell and Henderson1997). Supervisor-subordinate disagreement is also more likely to occur when examining attributes that are difficult to observe like honesty and ethics (Alicke, Reference Alicke1993; Brown & Mitchell, Reference Brown and Mitchell2010; Moore & Small, Reference Moore and Small2007; Morgan, Reference Morgan1993). Thus, it is important to examine perceptions of ethical leadership from both supervisors and subordinates to assess the implications of the degree and direction of supervisor-subordinate agreement on organization deviance. Specifically, we explore two types of disagreement (overestimation and underestimation) and two types of agreement (agreement on stronger ethical leadership and agreement on weaker ethical leadership) (see Figure 1). Our primary interest is on what happens when supervisors and subordinates disagree on the ethical leadership of the supervisor, but in this research, we explore both agreement and disagreement.
Figure 1: Types of Supervisor-Subordinate (Dis)Agreement with Corresponding Ethical Leader
Again, we focus on the unit level because members of the same unit are exposed to similar cues from their environment leading to consistent behaviors from the group (i.e., group norms; Salancik & Pfeffer, Reference Salancik and Pfeffer1978). In addition, when individuals experience consequences in a group setting, the outcomes observed by members can affect the behavior of the group as a whole (Bandura, Reference Bandura1986).
Disagreement and Organizational Deviance
First, we examine what happens when supervisors and subordinates disagree on their perceptions of the supervisor’s ethical leadership. We do not consider the sources of (dis)agreement, but will address them in our antecedent hypotheses (see Hypotheses 3 and 4). Disagreement occurs when supervisors either overestimate or underestimate their ethical leadership in relation to their subordinates. Although supervisors and subordinates are expected to differ in their perceptions of the supervisor’s ethical leadership, we are not proposing one perception is more valid than the other. Rather, we are simply interested in the type and level of (dis)agreement between the supervisors and subordinates and how that disagreement relates to deviance in the work unit. Therefore, consistent with previous research (Atwater & Yammarino, Reference Atwater, Yammarino and Ferris1997; Yammarino & Atwater, Reference Yammarino and Atwater1997), overestimation is operationalized as the supervisor having a higher perception of his/her own ethical leadership than the perception of his/her subordinates. On the other hand, underestimation is operationalized as when the supervisor has lower perceptions of his/her own ethical leadership than the perception of his/her subordinates. We hypothesize that the type of disagreement influences whether there is a positive or negative relationship with organizational deviance.
Overestimation
The first type of disagreement is when subordinates rate the supervisor lower in ethical leadership than the supervisor rates him or herself, or overestimation by the supervisor. In this situation, the leader believes she or he is displaying strong ethical leadership, but subordinates are not perceiving such behaviors to the same degree. Generally speaking, overestimators are more likely to be content with their current performance and less likely to seek out feedback from others (Atwater & Yammarino, Reference Atwater, Yammarino and Ferris1997). These supervisors perceive themselves as stronger ethical leaders than they are actually seen by their subordinates, and this may come across as acting morally superior. Overestimators can suffer from an illusion of optimism, misdiagnose their own strengths and weaknesses, and fail to invest time in improving themselves because they see their own performance as acceptable (Fleenor et al., Reference Fleenor, Smither, Atwater, Braddy and Sturm2010).
For example, if supervisors talk confidently about ethics, but subordinates do not see them as strong ethical leaders, then according to SLT, they will not be seen as credible ethical role models (Bandura, Reference Bandura1977, Reference Bandura1986). When subordinates perceive the supervisor is acting inconsistently, they are likely to mimic any perceived negative behaviors. Even worse, due to the norms of negative reciprocity of SET (Cropanzano & Mitchell, Reference Cropanzano and Mitchell2005), subordinates are more likely to engage in deviant behaviors to retaliate when they perceive supervisors are acting ethically superior towards them (Colvin, Block, & Funder, Reference Colvin, Block and Funder1995). Drawing on SIPT, when leaders are seen as inconsistent, research shows that actions “speak louder than words” and subordinates are more likely to engage in negative behaviors (Greenbaum, Mawritz, Piccolo, Reference Greenbaum, Mawritz and Piccolo2015). Overall, social influence theories point to a positive relationship between overestimation of ethical leadership and group deviance.
Underestimation
Research on self-other agreement indicates that there are instances when a subordinate rates his or her supervisor higher on various aspects of leadership than the supervisor rates him- or herself (Atwater & Yammarino, Reference Atwater and Yammarino1992). This type of disagreement is underestimation by the supervisor. Although previous research has not specifically looked at underestimation on ethical leadership, we believe some leaders will underestimate their level of ethical leadership relative to how they are rated by subordinates. Underestimators likely receive feedback from others that they are strong ethical leaders, and yet they see themselves as having room to improve their ethical leadership. Supervisors who underestimate can be thought of as humble leaders. “Humility is grounded in a self-view of accepting that something is greater than the self and manifests in self-awareness, openness to feedback, appreciation of others, low self-focus, and self-transcendent pursuit” (Ou, Tsui, Kinicki, Waldman, Xiao, & Song, Reference Ou, Tsui, Kinicki, Waldman, Xiao and Song2014: 38). Humility is important for leaders because it can enhance their effectiveness and even neutralize the damaging effects of undesirable personality traits (Owens, Wallace, & Walderman, Reference Owens, Wallace and Waldman2015).
SLT suggests that humble supervisors are seen as attractive and credible role models; subordinates will mimic the behaviors of the supervisor and avoid deviant actions (Bandura, Reference Bandura1977, Reference Bandura1986). Drawing on SET, humble leaders are also perceived as likable and trustworthy so subordinates feel obligated to respond with positive reciprocity, meaning they are less likely to engage in behaviors that hurt the supervisor or the unit (Blau, Reference Blau1964; Nielsen, Marrone, & Slay, Reference Nielsen, Marrone and Slay2010). Overall, the salient cues in the environment from humble leaders will be ethically positive, with norms that deviant behavior is not tolerated.
Leader humility is related to many positive outcomes (Owens et al., Reference Owens, Wallace and Waldman2015). Moshavi, Brown, and Dodd (Reference Moshavi, Brown and Dodd2003) theorized that leaders who underestimate their potential can be perceived by subordinates as showing high levels of concern for others. This concern is thought to motivate subordinates because it can be viewed as altruistic leadership or leaders that gain the trust and respect of followers (Kanungo & Mendonca, Reference Kanungo and Mendonca1996). Sosik (Reference Sosik2001) examined affective components of the leader-follower relationship and found that self-aware managers and underestimators were associated with more positive affective and performance outcomes than those who overestimated. Research on agreement between leaders and followers indicates that leaders who overestimate themselves on important leader behaviors are rated as less effective by, and generate more negative affective reactions from, their subordinates compared to leaders who underestimate (Atwater et al., Reference Atwater, Roush and Fischthal1995; Brett & Atwater, Reference Brett and Atwater2001; Moshavi et al., Reference Moshavi, Brown and Dodd2003; Van Velsor, Taylor, & Leslie, Reference Van Velsor, Taylor and Leslie1993).
Overall, social influence theories and previous research lead us to predict that when supervisors underestimate their ethical leadership there will be less deviance in the work group while overestimation is associated with more examples of deviant behavior. Thus, we hypothesize:
Hypothesis 1: Underestimation of ethical leadership (i.e., subordinate ratings are higher than supervisor self-ratings) will be related to less frequent unit organizational deviance than overestimation (i.e., subordinate ratings are lower than supervisor ratings).
Agreement and Organizational Deviance
Next, we examine cases where supervisors and subordinates agree on their perceptions of the supervisors’ ethical leadership and their relationship to organizational deviance. We expect the two cases of agreement, agreement on stronger ethical leadership and agreement on weaker ethical leadership, will be straightforward and similar to those found in previous research examining a main effect of ethical leadership on deviance (e.g., Kacmar, Bachrach, Harris, & Zivnuska, Reference Kacmar, Bachrach, Harris and Zivnuska2011; Mayer et al., Reference Mayer, Aquino, Greenbaum and Kuenzi2012; Neves & Story, Reference Neeves and Story2015). Consistent with previous research on leader self-other agreement (Atwater et al., Reference Atwater, Ostroff, Yammarino and Fleenor1998), we predict that agreement on perceptions of ethical leadership is relevant for outcomes such as workplace deviance.
Agreement on Weaker Ethical Leadership
Agreement on weaker ethical leadership situations occurs when supervisors and subordinates both agree that the supervisor is weaker on ethical leadership. This type of agreement impacts subordinates’ deviant behavior via three distinct paths. First, drawing on SLT, when supervisors perceive themselves as having low-levels of ethical leadership, they believe that they are falling short as ethical role models. Agreement on weaker ethical leadership means that supervisors are not strongly sending and subordinates are not clearly receiving messages that ethical behavior is as (or more) important than other outcomes such as achieving bottom-line outcomes (Greenbaum, Mawritz, & Eissa, Reference Greenbaum, Mawritz and Eissa2012). For example, weaker ethical leaders might not actively discipline employees who engage in unethical behavior, model strong ethical behavior, or emphasize the importance of maintaining high ethical standards; this can contribute to increased deviance in the work unit.
Furthermore, both supervisors and subordinates recognize that weaker ethical leaders have failed to demonstrate that they are trustworthy and fair which, according to SET theory, can foster negative interactions in the workplace. Norms of reciprocity suggest that subordinates will respond to poor ethical leadership by exhibiting negative behaviors, including deviant behaviors (Cropanzano & Mitchell, Reference Cropanzano and Mitchell2005). Applying SIPT, when supervisors and subordinates agree that ethical leadership is absent from the workplace, strong cues develop that ethical conduct is not valued or expected in the work environment. These negative cues are adopted by the unit, leading to higher levels of deviance exhibited by subordinates (Mayer, Kuenzi, & Greenbaum, Reference Mayer, Kuenzi and Greenbaum2010). Overall, when both supervisors and subordinates agree that weaker ethical leadership is present, there will be a stronger, positive relationship with organizational deviance.
Agreement on Stronger Ethical Leadership
In contrast, in situations where both subordinates and supervisors agree on higher ratings of ethical leadership, there should be less deviant behavior based on the same reasoning as above. According to SLT, social learning requires supervisors to actively role model, communicate, and enforce ethical standards. If supervisors make these ethical leader behaviors salient, then subordinates will get the message that deviant and other negative behaviors will not be tolerated (Brown & Treviño, Reference Brown and Treviño2006). Further, ethical leaders create positive exchanges with subordinates that are characterized by perceptions of mutual trust and fairness (Potipiroon & Fellows, Reference Potipiroon and Fellows2016). Positive reciprocity by subordinates may include refraining from deviant behavior that will hurt the organization (SET; Blau, Reference Blau1964). Drawing on SIPT, when supervisors demonstrate that they are ethical persons and actively manage the ethical environment, the salient norms of the group will not promote deviant behavior and subordinates are less likely to engage in acts of deviance (Mayer et al., Reference Mayer, Kuenzi and Greenbaum2010).
In sum, when supervisors and subordinates agree regarding the ethical leadership of the supervisor, less sensemaking is needed for subordinates to know the standards for ethical behavior as there is a consistent interpretation of the ethical norms in the work environment. However, we suggest that there are different types of agreement for ethical leadership of the supervisor between supervisors and subordinates and these will impact deviance differently. We posit when supervisors and subordinates agree that ethical leadership is weaker, there will be more deviant behavior in the work group. On the other hand, when supervisors and subordinates agree that ethical leadership is stronger, there will be less deviant behavior in the work group. Specifically, we hypothesize:
Hypothesis 2: Supervisor-subordinate agreement on weaker ethical leadership will be related to more frequent unit deviance than agreement on stronger ethical leadership.
ANTECEDENTS OF SUPERVISOR-SUBORDINATE (DIS)AGREEMENT
Because we propose disagreement (specifically overestimation) is related to negative outcomes in organizations, we now turn our attention to what may influence supervisor-subordinate disagreement on perceptions of ethical leadership. Social influence processes affect how messages and cues are interpreted by subordinates (Ruben & Kim, Reference Ruben and Kim1975; Ruben & Stewart, Reference Ruben and Stewart2016). Individuals do not respond directly to work environments; rather, they engage in cognitive sensemaking processes involving first perception and then interpretation of cues in the work environment (Campbell, Dunnette, Lawler, & Weick, Reference Campbell, Dunnette, Lawler and Weick1970). These cues are not just words, but also actions of the supervisor, non-verbal messages, and context. These cues sent to subordinates can be intentionally created or produced accidently (Ruben & Kim, Reference Ruben and Kim1975; Ruben & Stewart, Reference Ruben and Stewart2016). However, supervisors not only create cues in the environment, but also impact how cues are interpreted as well as their salience in the environment through social influence processes.
We believe that supervisors’ beliefs about their own ethics (specifically, that they are better-than-average ethical leaders) as well as the ethics of others (that ethics is a fixed character trait that cannot be changed) are key to understanding disagreement on ethical leadership. In this study we examine two antecedents, supervisors’ better-than-average beliefs and moral entity perspective, which we hypothesize impact the ethical behaviors and cues supervisors display as well as their salience to subordinates. We draw from social learning theory to explain how these supervisor beliefs impact disagreement on perceptions of ethical leadership between supervisors and subordinates.
Supervisor Better-Than-Average (BTA) Beliefs
Bazerman and Tenbrunsel (Reference Bazerman and Tenbrunsel2011) provide convincing evidence that individuals are often not as ethical as they think they are. They suggest that most individuals do not want to be unethical, but individuals either overestimate their ability to do the right thing or act in some unethical manner without awareness. It is easy for individuals to fall prey to biases such as self-enhancement (overly positive view of self to maintain self-esteem) and motivated blindness (seeing what we want to see and not seeing contradictory information) when it comes to issues related to ethics. If supervisors cannot overcome their blind spots and recognize their ethical deficiencies, supervisor-subordinate agreement will be difficult to attain.
One way to examine supervisors’ inflated sense of their ethical leadership is the better-than-average effect. The BTA effect is the tendency for people to favorably compare themselves to others—typically the “average” person. Although these BTA beliefs have been well established (Alicke & Govorun, Reference Alicke, Govorun, Alicke, Dunning and Krueger2005; Taylor & Brown, Reference Taylor and Brown1988), there are different explanations as to how BTA beliefs develop. Some research suggests that self-enhancing motivations play an important role in favorable better-than-average comparisons (Alicke & Govorun, Reference Alicke, Govorun, Alicke, Dunning and Krueger2005). For example, compared to negative information (such as poor performance outcomes), positive information about ourselves is more easily remembered, seen as more important, and more frequently attributed to our own efforts (see Fiske & Taylor, Reference Fiske and Taylor1991 for a review). In this way, individuals’ bad qualities, behaviors, and performance are overlooked, leading to inflated self-perceptions. Other researchers argue that the BTA phenomenon occurs because people know more about themselves and less about others (Moore, Reference Moore2007; Moore & Small, Reference Moore and Small2007). The BTA effect is also more likely to occur when individuals compare themselves to others on skills or attributes that are difficult to observe such as honesty or ethics (Moore & Small, Reference Moore and Small2007). For example, most people know themselves to be generally honest, but the information they have about others’ honesty is more limited.
The literature on self-other agreement is unequivocal that self-enhancing bias beliefs produce inflated self-ratings of leadership that can lead to disagreement (Harris & Schaubroeck, Reference Harris and Schaubroeck1988). When applied specifically to ethical leadership, ethically overconfident leaders believe in their own goodness and assume that this goodness is readily apparent to others. In turn, this impacts the cues and messages that supervisors provide to subordinates.
From a social learning perspective, we propose that when supervisors believe their ethical leadership is better than average, they might not consistently engage in active moral management such as role modeling, talking about ethics, and rewarding/punishing (un)ethical behavior. Supervisors may believe they are strong ethical leaders, but without demonstrating it they will not be perceived as such by subordinates (Dunning, Heath, & Suls, Reference Dunning, Heath and Suls2004; Treviño, Brown, & Hartman, Reference Treviño, Brown and Hartman2003). Strong BTA beliefs can also affect the attractiveness of the supervisor which is critical to being seen as an ethical role model by subordinates. Leaders who believe they are well above average can suffer from a high degree of self-admiration, arrogance, or aloofness which will not be received positively by others (Hogan, Curphy, & Hogan, Reference Hogan, Curphy and Hogan1994; Judge, LePine, & Rich, Reference Judge, LePine and Rich2006). Thus, we predict that supervisors with stronger BTA beliefs will rate themselves more positively as ethical leaders; however, such beliefs will not necessarily translate into stronger subordinates’ ratings of ethical leadership. This will contribute to supervisor-subordinate disagreement. Specifically, we hypothesize:
Hypothesis 3: Supervisor better-than-average (BTA) beliefs for ethical leadership are negatively related to supervisor-subordinate agreement on ethical leadership ratings.
Supervisor Moral Entity Theorist
Next, we examine the supervisors’ beliefs about the malleability of subordinate’s ethicality. According to implicit person theory, individuals have beliefs as to whether personal attributes such as morality and intelligence are fixed or malleable (Dweck et al., Reference Dweck, Chiu and Hong1995). Individuals who believe that traits such as morality are malleable are called incremental theorists. Individuals who believe traits of individuals do not change are entity theorists. We hypothesize that supervisors who believe the ethicality of others cannot be changed, or moral entity theorists, will be viewed as weaker ethical leaders by subordinates based on the cues and messages they provide to subordinates.
Having beliefs that individuals cannot change regarding their ethics impacts not only the types of cues and messages individuals are given regarding the importance of ethics, but also the type of interactions supervisors have with different subordinates. For example, entity theorists are less likely to coach employees as well as recognize positive changes in their performance (Heslin, Latham, & Vandewalle, Reference Heslin, Latham and Vandewalle2005; Heslin, Vandewalle, & Latham, Reference Heslin, Vandewalle and Latham2006; Heslin & VandeWalle, 2008). Supervisors with a moral entity orientation also will invest less time fostering ethical behavior of subordinates by setting and communicating ethical standards because they do not believe it would change the behavior of subordinates. When supervisors spend less time talking about how to be ethical or helping subordinates become more ethical, they will come across as weaker moral managers (Treviño et al., Reference Treviño, Hartman and Brown2000). From a social learning perspective, such supervisors are not strong ethical role models.
Additionally, supervisors with a moral entity orientation attribute wrongdoing to personal qualities of the transgressor. According to Heslin et al. (Reference Heslin, Latham and Vandewalle2005), managers with an entity theory are disinclined to alter their perceptions of an employee’s performance even when presented with opposite information. This happens because entity theorists believe that an individual’s attributes are innate and unalterable. In situations like this, supervisors who are moral entity theorists will have less positive exchanges on matters related to employee misconduct because they do not invest in helping subordinates develop and improve (Dweck, Reference Dweck1999). Dweck et al. (Reference Dweck, Chiu and Hong1995) also suggest that entity theorists will relate to others differently because they view the behavior of others as a result of enduring, static personal qualities. Supervisors with an entity theory perspective are also more likely to provide biased performance appraisals based on preconceived beliefs about individuals (Heslin et al., Reference Heslin, Latham and Vandewalle2005). Subordinates will see these biased appraisals as unfair, such that supervisors with a moral entity orientation will be perceived as having less integrity (i.e., they will come across as weaker ethical leaders) than they perceive themselves. These undesirable attributes will make supervisors less attractive and, therefore, less effective as role models, according to SLT.
Overall, we propose that supervisor moral entity orientation will be negatively related to supervisor-subordinate agreement. Supervisors with a moral entity orientation believe that the morality of their subordinates is fixed, so these supervisors will be less likely to actively manage ethics, including ethical role modeling, which is important from a social learning perspective on ethical leadership. A moral entity orientation will also contribute to a leader being seen as a weaker moral person (e.g., seen as having less integrity); however, moral entity orientation should be unrelated to supervisors’ self-ratings of ethical leadership because individuals are usually not aware of their own theories about moral mutability (Dweck et al., Reference Dweck, Chiu and Hong1995). Thus, we hypothesize:
Hypothesis 4: Supervisor moral entity orientation is negatively related to supervisor-subordinate agreement on ethical leadership.
SAMPLE AND PROCEDURE
We collected data from 343 units from different organizations in the southeast US including technology, government, insurance, financial, food service, retail, manufacturing, and medical organizations. After securing IRB approval, we asked 343 business students at a large university in the southeastern United States to hand-deliver survey packets to five subordinates and the supervisor of a smaller unit in an organization that each identified. Unit members were told that the students would get extra credit if they completed the surveys and were offered an aggregate summary of the results. Students were given detailed instructions and a checklist to turn in showing they had permission to hand out the surveys to employees during their shift. Respondents were provided with detailed instructions and were assured confidentiality of their responses. A postage-paid envelope was included in the packet to return the survey anonymously.
We received a total of 1525 subordinates’ responses and 334 supervisors’ responses. Previous research suggests that three member responses per unit is a sufficient number to aggregate measures to the group level (Colquitt, Noe, & Jackson, Reference Colquitt, Noe and Jackson2002; Richardson & Vandenberg, Reference Richardson and Vandenberg2005; Schneider, White, & Paul, Reference Schneider, White and Paul1998; Tracey & Tews, Reference Tracey and Tews2005). Forty-five supervisors had fewer than three subordinate responses and were therefore deleted from this sample. The researchers examined the remaining surveys to make sure they had been completed, looked for careless responses (see Meade & Craig, Reference Meade and Craig2012), and inspected the writing on the surveys to see if they had been completed by different people. We eliminated another six departments because they had surveys that were incomplete.
Thus, our final sample consisted of 1399 subordinates and 283 supervisors from 283 departments (M = 4.94 subordinates per department). Subordinate respondents were 41.8% male and 61.2% Caucasian (12.6% Hispanic and 11.6% African-American). They averaged 30.3 years of age with 4.2 years of experience in the organization and 3.4 years in the unit. Supervisor respondents were 57% male and 72.2% Caucasian (7.1% Hispanic and 7.7% African-American). They averaged 39.2 years of age with 9.2 years working in the organization and 7.6 years in the unit.
The subordinate survey contained measures of supervisor ethical leadership, group deviance, frequency of interaction with supervisor, and demographic questions. The supervisor survey contained measures of self-ratings of ethical leadership, better-than-average beliefs, supervisor moral entity orientation, group deviance, and demographic questions.
Measures
Except where specified below, responses for all items were made on a five-point scale (1 = strongly disagree, 5 = strongly agree).
Ethical Leadership
Supervisor ethical leadership was rated by both supervisors and subordinates. Supervisor self-ratings of ethical leadership were measured with the 10-item ethical leadership scale developed by Brown et al. (Reference Brown, Treviño and Harrison2005). Supervisors rated the extent to which they agreed with statements such as “I talk about the importance of ethics,” and “I set an example of how to do things the right way in terms of ethics.” Subordinate perceptions of ethical leadership was measured using the same 10-item scale rated by subordinates in which they rated their supervisor (supervisor α = .90, subordinate α = .97); for example, “My supervisor talks about the importance of ethics.”
Better-Than-Average (BTA) Beliefs
BTA beliefs were rated by the supervisor and based on work by Moore (Reference Moore2007). Supervisors were asked to provide a percentage that describes how their ethical leadership related to that of other managers. The item used is “I believe that my ethical leadership is stronger than the ethical leadership of __% of the other managers I know.”
Supervisor Moral Entity Orientation
Supervisors’ rated their own moral entity orientation using the 3-item implicit theory scale developed by Dweck et al. (Reference Dweck, Chiu and Hong1995). Supervisors rated the extent to which they agreed with statements such as “A person’s moral character is something that is very basic about them and it can’t be changed much,” and “Whether a person is responsible and sincere or not is deeply ingrained in their personality. It cannot be changed very much” (α = .81).
Organizational Deviance
Deviant behavior of the unit was assessed using Bennett and Robinson’s (2000) 12-item Organizational Deviance Scale (ODS). Both employees and the supervisor of the unit rated the extent to which employees, as a whole, engaged in various deviant behaviors within the past year on a seven-point response format (1 = never, 2 = once, 3 = a few times, 4 = several times, 5 = monthly, 6 = weekly, 7 = daily). These were combined into one score for the unit. Example behaviors included in the scale were “taking property from work without permission” and “coming in late to work without permission” (α = .94).
Controls
Observational opportunities impact self-other agreement in that individuals must have enough chances to interact with the focal individual to form a reliable impression (Harris & Schaubroeck, Reference Harris and Schaubroeck1988). Therefore, frequency of interaction between supervisor and subordinate was measured and controlled for with a 3-item scale we developed. The items include “I frequently interact with my supervisor,” “I frequently witness my supervisor interacting with other employees,” and “There are frequent opportunities to observe my supervisor’s behavior” (α = .93).
Data Aggregation
The unit of analysis in this study is the work unit. Thus, we aggregated subordinate responses on ethical leadership and subordinate and supervisor-rated organizational deviance to the unit level. The referent for our deviance measure was the work group (i.e., we did not ask individuals to report on the frequency of their own deviant behavior, but on deviance they observed in the group as a whole). Ethical leadership was measured at the individual level but aggregated to the group level because individuals in the same work unit are exposed to stimuli from the leader that impact the entire unit, not just one individual. In other words, we treat each of these constructs as a “shared unit property” (see Kozlowski and Klein, Reference Kozlowski, Klein, Klein and Kozlowski2000, for a review of how these group-level constructs emerge).
We assessed the degree of agreement for ethical leadership and organizational deviance by calculating the r wg statistic (George & James, Reference George and James1993) as well as the intraclass correlations ICC(1) and ICC(2). These statistics indicate the extent of agreement within units, the degree of inter-rater reliability, and the stability of unit-level means, respectively. The mean r wg statistic met acceptable levels (Lance, Butts, & Michaels, Reference Lance, Butts and Michaels2006) for ethical leadership (r wg = .96), frequency of interaction with supervisor (r wg = .88), and organizational deviance (r wg = .97). The ICC(1) for ethical leadership was .34. and also .34 for organizational deviance. The ICC(2) for ethical leadership was .70 and .64 for organizational deviance. As all of the ICC values were statistically significant and the r wg values were high indicating that there was high within-group agreement, there is support for aggregation.
RESULTS
Descriptive Statistics
The means, standard deviations, reliabilities, and intercorrelations among the key variables are presented in Table 1. As predicted, organizational deviance is negatively correlated with supervisor perceptions of ethical leadership (r = -.22) and subordinate perceptions of ethical leadership (r = -.38). In addition, the mean value for BTA beliefs is almost 70%, suggesting most leaders in our study thought their ethical leadership was better-than-average, which is to be expected. Leaders, in particular, often see themselves as above average, especially on issues related to ethics and leadership (Bazerman & Tenbrunsel, Reference Bazerman and Tenbrunsel2011; Messick & Bazerman, Reference Messick and Bazerman1996). Approximately 80% of the supervisors in our sample rated themselves higher on ethical leadership than they were rated by their subordinates.
Table 1: Means, Standard Deviations, and Correlations Among Key Measures
Note. N = 283. M = mean. SD = standard deviation. Values on the diagonal represent coefficient alphas. *p < .05, **p < .01, ***p < .001
Validity of Measures
We conducted confirmatory factor analysis using MPlus 8 to ensure all the variables were distinct constructs. First, we examined the fit of a four-factor model (supervisor ethical leadership rated by subordinates, self-rated ethical leadership, supervisor moral entity orientation, and deviance). This model had an adequate fit with the data (χ2 = 2866.84, df = 856, p < .001; RMSEA = .07, CFI = .92). We compared the fit of the four-factor model with a three-factor model (combined subordinate-rated and self-rated ethical leadership, supervisor moral entity orientation, and deviance) and a one-factor model (all combined). Neither the three-factor model (χ2 = 3930.51, df = 859, p < .001; RMSEA = .11, CFI = .64) nor the one-factor model (χ2 = 6228.02, df = 862, p < .001; RMSEA = .14, CFI = .37) fit the data well. The four-factor model fit the data better than either the three-factor model (Δ χ 2 = 1063.67, df = 3, p ≤ .001) or the one-factor model (Δ χ 2 = 3361.18, df = 6, p ≤ .001) suggesting that subordinate-rated and supervisor-rated ethical leadership, supervisor moral entity orientation, and unit deviance are distinct constructs. We do not include BTA in this model because it is assessed by a single item and, as such, should not be included in the CFA.
Hypothesis Testing
First, we found that there was a significant difference between supervisor and subordinate rated ethical leadership. The mean level of supervisor self-rated ethical leadership (M = 4.27) was higher than follower-rated ethical leadership (M = 3.80), t(283) = 13.64, p < .001 (two-tailed); Cohen’s d = .91.
Next, we used polynomial regression and response surface methodology to examine Hypotheses 1 and 2 (Edwards, Reference Edwards1993, Reference Edwards, Drasgow and Schmitt2001; Edwards & Parry, Reference Edwards and Parry1993) to look at levels of (dis)agreement. Polynomial regression allows the impact of fit/agreement at different levels of ethical leadership and the impact of misfit/(dis)agreement resulting from higher supervisor or subordinate ethical leadership perceptions. Following Edwards and Parry (Reference Edwards and Parry1993), this framework consists of analyzing particular features of the surfaces corresponding to quadratic equations to examine the impact of (dis)agreement or (mis)fit on unit deviance in Figure 2.
Figure 2: Supervisor-Subordinate Agreement on Ethical Leadership and Organizational Deviance
We start by estimating the following equation:
$$Z = {b_0} + {b_1}X + {b_2}Y + {b_3}{X^2} + {b_4}XY + {b_5}{Y^2} + e,$$where X represents the subordinates’ rating of ethical leadership, Y is the corresponding supervisor’s self-rating of ethical leadership, and Z is unit deviance. We follow Edwards and Parry (Reference Edwards and Parry1993) by examining particular features of the surfaces (i.e., slope and curvature) that result from the quadratic equation to analyze the results (see Table 2).
Table 2: Results of Polynomial Regressions of Organizational Deviance on Supervisor-Subordinate Agreement on Ethical Leadership
Note. Following Edwards and Rothbard (1999), columns labeled X, Y, X2, XY, Y2 reflect unstandardized regression coefficients with all predictors entered simultaneously. The column R2 indicates the variance explained by the predictors. X = employee-rated ethical leadership, Y = supervisor self-rated ethical leadership. *p < .05, **p < .01, *** p < .001.
Hypothesis 1 predicts that supervisor underestimation of ethical leadership will be related to less frequent unit organizational deviance than overestimation. To test this, we examined the X = -Y line, which extends from the left-most corner of the “floor” (marked C) in Figure 2 to the right-most corner (marked D). Results indicate the slope (represented by b3-b4+b5) and the curvature (represented by b1-b2) are both significant and negative (-1.19, p < .05 and -.18, p < .05, respectively), supporting Hypothesis 1 (See Table 2). Furthermore, a visual inspection of Figure 2 shows that levels of deviance are higher when leaders overestimate at lower levels of ethical leadership (e.g., when a leader rates him or herself a 3 and a follower rates him or her a 1) compared to higher levels of ethical leadership (e.g., when a leader rates him or herself a 5 and a follower rates him or her a 3). This is not surprising given that lower levels of ethical leadership are associated with more frequent occurrences of deviance.
Hypothesis 2 predicts agreement at lower-levels of subordinate and supervisor ratings of ethical leadership will be related to more organizational deviance than agreement at higher levels of ethical leadership. To test this, we examine the surface along a line that reflects equal values for X and Y (the X = Y line). In Figure 2, this is the line that extends from the nearest corner (marked A) to the furthest corner (marked B) of the X, Y plane. Hypothesis 2 predicts that the shape of the surface along this X = Y line will decrease moving from the nearest corner of the surface (A) to the furthest corner (B). Results indicate that the slope (represented by b1 + b2) is significant and negative (-2.73, p < .01) supporting Hypothesis 2 (see Table 2).
Hypotheses 3 and 4 focus on antecedents of supervisor-subordinate agreement on ethical leadership. In order to test these hypotheses, we followed the approach outlined by Edwards (Reference Edwards1995) for studying congruence as a dependent variable, which we outline below in five steps. First, we conducted a multivariate regression analysis so that both dependent variables—the subordinate and supervisor rating components of agreement—were tested simultaneously. The omnibus test from this analysis is significant (Wilks’ Λ = .56, F[8, 554] = 218.70, p < .001) indicating that the set of antecedent variables is related to subordinate and supervisor ratings of ethical leadership considered jointly.
Second, we examined each individual antecedent to determine if it had an equal but opposite effect on the joint supervisor-subordinate ratings. Equal but opposite effects are possible when the Wilks’ Λ for an antecedent is non-significant, but the regression coefficient for the antecedent is significant. Neither of our antecedents met these criteria. Therefore, we ruled out the possibility that equal but opposite effects were occurring.
In the third step, the relationships between each antecedent and both types of ratings are examined individually, so that we can better understand the nature of the potential (dis)agreement. To do this, we checked to see if the regression coefficients for the antecedents differed for overestimators (i.e., supervisor ratings > subordinate ratings) versus underestimators (i.e., supervisor ratings < subordinate ratings). We created a dummy variable and reran our analyses by adding this dummy variable to our model as well as interaction terms for each antecedent and this dummy variable. None of these interactions are significant, meaning that there are no differences in antecedent effects between overestimators and underestimators.
Fourth, we plotted the regression lines for supervisor- and subordinate-rated ethical leadership from the previous steps (using unstandardized regression coefficients) on the same set of axes. Ratings of ethical leadership are on the y-axis and values of predictors are on the x-axis: BTA beliefs (Figure 3) and supervisor moral entity orientation (Figure 4).
Figure 3: Better-Than-Average Beliefs and Supervisor-Subordinate Agreement on Ethical Leadership
Note. Unstandardized coefficients were used to plot regression lines. Line shown is when Moral Entity Orientation is at its mean value.
Figure 4: Supervisor Moral Entity Orientation and Supervisor-Subordinate Agreement on Ethical Leadership
Note. Unstandardized coefficients were used to plot regression lines. Line shown is when better-than-average beliefs are at their mean value.
Finally, we examined the pattern of congruence between the lines. The results indicate that as BTA beliefs increase, supervisor-subordinate agreement on ethical leadership decreases (Figure 4). More specifically, BTA beliefs are positively related to supervisor self-rated ethical leadership (see Table 3) but unrelated to subordinate ratings of ethical leadership moving supervisors and subordinates further apart as BTA beliefs increased. Thus, Hypothesis 3 is supported.
Table 3: Relationships Between Better-Than-Average Beliefs and IPT-Moral on Supervisor Self-Ratings and Subordinate Ratings of Ethical Leadership
Note. N = 283; Standardized regression coefficients (Betas) are shown. *p < .05.
In terms of Hypothesis 4, as supervisor moral entity orientation increased, supervisor-rated ethical leadership remained unchanged while subordinate-rated ethical leadership decreased (see Figure 4). Thus, Hypothesis 4 is supported.
DISCUSSION
In this research, we studied antecedents (better-than-average beliefs, supervisor moral entity orientation) and an outcome (unit-level deviance) of supervisor-subordinate disagreement on ethical leadership. We found that supervisors, on average, rate themselves higher on ethical leadership than employees rate them. Also, we hypothesized and found that deviance is higher when subordinates and supervisors agree the supervisor is weaker on ethical leadership as well as when subordinate ratings of ethical leadership are lower than supervisor self-ratings (overestimation). Deviance is lower when supervisors underestimate their ethical leadership compared to subordinates and when subordinates and supervisors agree the supervisor is stronger on ethical leadership. As for antecedents of (dis)agreement, we found that better-than-average beliefs and supervisor moral entity orientation are negatively related with agreement on ethical leadership. Overall, our findings are consistent with social learning and other influence perspectives on ethical leadership (e.g., Brown et al., Reference Brown, Treviño and Harrison2005).
Theoretical Implications
The results of this research have implications for multiple literatures. First, we extend recent work on ethical biases, particularly blind spots (Tenbrunsel et al., Reference Tenbrunsel, Diekmann, Wade-Benzoni and Bazerman2010), to ethical leadership by exploring whether leaders believe they are more ethical than they really are (i.e., compared to how they are really seen by others). We found that, on average, supervisors overestimate their own ethical leadership relative to how they are rated by their subordinates. This is important because previous research on self-other agreement (Atwater et al., 1997; Harris & Schaubroeck, Reference Harris and Schaubroeck1988) and ethical bias (Bazerman & Tenbrusnel, 2011) has not explicitly examined (dis)agreement among supervisors and subordinates on ethical leadership. Our findings demonstrate that examining both leader and follower perceptions can complement the traditional follower-centered approach used in most ethical leadership research.
We also contribute to the ethical leadership and deviance literatures by highlighting the relationship between (dis)agreement and deviance. Prior research has shown that ethical leadership (as rated by subordinates) is negatively related to deviance (Mayer et al., Reference Mayer, Kuenzi, Greenbaum, Bardes and Salvador2009; Mayer et al., Reference Mayer, Nurmohamed, Treviño, Shapiro and Schminke2013). Our results indicate this relationship is more complicated when we take into account both the supervisors’ and subordinates’ perceptions of ethical leadership. When supervisors and subordinates agree on the supervisors’ ethical leadership, the results mimic the main effects of subordinate perceptions of ethical leadership on deviance. However, when we look at disagreement, the results are more nuanced with overestimation associated with greater deviance, and underestimation related to less deviance. We attribute these findings, in part, to leader humility.
According to Owens and colleagues (2013) accurate self-perception (i.e., self-other perceptions are in agreement) is a core dimension of humility. Our findings, however, show that accuracy does not always correlate with humble, effective leadership; such as when leaders and followers agree that the leader is a weaker ethical leader. Furthermore, we found that some inaccurate perceptions are beneficial; specifically, underestimation which is associated with lower levels of deviance (see Figure 3). Our research adds to the emerging literature on humility (Owens et al., Reference Owens, Johnson and Mitchell2013; Owens et al., Reference Owens, Wallace and Waldman2015; Owens & Hekman, Reference Owens and Hekman2012) and suggests future research is needed to clarify the relationships between humility, (dis)agreement, and important outcomes like deviance.
Given the potential for negative outcomes such as deviance, we examined antecedents linking supervisor beliefs about ethics (better-than-average beliefs and moral entity perspective) to (dis)agreement on the supervisor’s ethical leadership. These antecedents new to the self-other agreement literature were chosen, in part, because of their relevance to a social learning perspective on ethical leadership. In line with our predictions, we found both better-than-average beliefs and moral entity orientation were negatively related to agreement on ethical leadership. However, the form of disagreement for each antecedent differed. Disagreement based on BTA beliefs was driven by supervisor self-ratings (BTA beliefs were unrelated to follower perceptions); the opposite was true for moral entity orientation, which was related to follower perceptions but not supervisor self-perceptions.
It is possible that this pattern of results can be explained by the shared foci of the beliefs and ratings. We found that supervisors’ beliefs about their own ethics impacted how they rated themselves as ethical leaders (i.e., both have a common focus on how supervisors see themselves). Supervisors’ beliefs about the ethics of others were related to how the supervisors were rated by subordinates on ethical leadership (i.e., beliefs about others can affect how supervisors behave toward and are ultimately perceived by others). Our findings suggest that supervisor beliefs can produce disagreement by influencing follower perceptions. This is a novel contribution to the literature on antecedents of (dis)agreement because previous research has focused on identifying relationships between leader individual differences (e.g., personality) and leader self-ratings as well as follower individual differences and follower-ratings (Atwater & Yammarino, Reference Atwater, Yammarino and Ferris1997). Future research should explore other antecedents and identify the different ways they can contribute to disagreement on ethical leadership. For example, there might be follower individual differences that are related to (dis)agreement on ethical leadership and deviant behaviors (e.g., followers who are complicit in workplace deviance might be subject to a type of motivated reasoning that makes it harder for them to see ethical leadership behaviors).
Finally, our study addresses the paucity of self-other agreement research in the area of ethics, and in particular, ethics and leadership (Brown & Mitchell, Reference Brown and Mitchell2010). We found that different types of agreement and disagreement on ethical leadership were related to deviance. The findings from the broader self-other agreement literature suggest supervisor-subordinate (dis)agreement on other phenomenon might be relevant. Overall, this study represents a first step toward answering these types of questions.
Practical Implications
There are also important practical implications of the present work. In our sample, almost 80% of supervisors rated themselves as more ethical than their subordinates rated them, which suggests that overestimating personal ethical leadership might be an epidemic. In this study we found that overestimation is related to greater levels of workplace deviance. Overestimation of ethical leadership can contribute to a negative relationship with subordinates, interfere with the benefits of being an ethical leader, and inhibit self-improvement of supervisors. These results are in line with research in other areas. For instance, subordinates receiving feedback from leaders who are overestimators are more likely to experience feelings of anger and discouragement (Brett & Atwater, Reference Brett and Atwater2001). Also, the relationship between transformational leadership and important outcomes (e.g., leader’s performance evaluations and promotion potential) is strongest when leader self-ratings are most closely related to followers’ ratings and weakest when leader ratings exceed follower ratings (Atwater & Yammarino, Reference Atwater and Yammarino1992). Organizations should seek to identify supervisors who overestimate their ethical leadership and utilize self-awareness training (e.g., as part of training to develop leader humility; Owens et al., Reference Owens, Wallace and Waldman2015) to help leaders become more aware of how others perceive their actions as well as to understand the importance of these perceptions on important organizational outcomes and subordinate behaviors. However, leaders who overestimate might fail to see the benefits of such training, especially if they see themselves as strong ethical leaders already. Future research should look at the impact of self-other (dis)agreement, particularly overestimation, on training effectiveness.
Supervisors also need coaching on how to seek feedback from subordinates given that negative feedback seeking is related to perceived effectiveness. Yet, research suggests that individuals are more likely to seek feedback from their superiors than from their peers or subordinates, because feedback from superiors is more instrumental in obtaining success (Ashford & Tsui, Reference Ashford and Tsui1991). Nevertheless, providing feedback to leaders regarding how they are perceived by their followers enhances subsequent leader-follower agreement (Atwater, Waldman, Atwater, & Cartier, Reference Atwater, Waldman, Atwater and Cartier2000). Obtaining this feedback from subordinates can be problematic in that subordinates are reluctant to share with someone whom they perceive to be less ethical. For instance, Atwater et al. (Reference Atwater, Waldman, Atwater and Cartier2000) found that when leaders received negative feedback they were more cynical and had less commitment to subordinates. Leaders may also discount subordinate feedback because they think subordinates do not have sufficient information or understanding to provide feedback or their feedback can be politically motivated (Ashford, Blatt, & Vandewalle, Reference Ashford, Blatt and Vandewalle2003). Again, organizations could provide training for both subordinates and leaders on how to effectively provide and use 360-degree feedback.
Last, but not least, our results show that having a supervisory moral entity orientation is associated with greater disagreement on ethical leadership (specifically due to lower follower ratings of ethical leadership). This has potential implications for leader selection and development. Organizations should take care in selecting individuals with strong moral entity beliefs and placing them in positions of leadership. Would-be ethical leaders must believe that they are capable of being effective moral managers who have the power to influence the ethical conduct of those they lead. Training, education, and other leader development activities can help leaders understand that individuals’ ethics can be influenced, particularly through the demonstration of strong ethical leader behavior which has been shown effective across many studies in promoting ethical, and reducing unethical, conduct of subordinates (Ng & Feldman, Reference Ng and Feldman2015).
Limitations and Future Directions
Although there are a number of strengths of the current research, there are also important limitations. First, although we draw on social influence theories, we do not directly measure the mechanisms the theories propose. For instance, we draw on SLT (Bandura, Reference Bandura1977, Reference Bandura1986) but we did not directly measure any role-modeling behaviors. Our results are consistent with SLT; however, we cannot state conclusively that our results are due to social learning processes without directly measuring the processes. Further, the importance of each mechanism might vary depending on the antecedents and outcomes in question. For example, social learning seems more important to understanding the antecedents of supervisor-subordinate (dis)agreement while social exchange might provide a more powerful explanation of linking such (dis)agreement to outcomes. Future research should examine the different roles and differential effects of social influence theories of ethical leadership on different antecedents and outcomes of (dis)agreement.
A second limitation is that we only examined a single dependent variable—organizational deviance. Recent research has expanded the nomological network of ethical leadership, relating it to organizational citizenship behaviors, voice, and other negative outcomes such as relationship conflict and unethical behavior (Ng & Feldman, Reference Ng and Feldman2015). Thus, in future research, it would be interesting to examine additional outcomes of (dis)agreement, especially positive ones such as prosocial or whistle-blowing behavior. Furthermore, researchers might also consider whether (dis)agreement on ethical leadership influences (dis)agreement on supervisor and subordinate perceptions on outcomes. For example, perhaps disagreement on ethical leadership explains why senior managers and rank and file employees have differing views about key dimensions of organizational ethics (Treviño, Weaver, & Brown, Reference Treviño, Weaver and Brown2008).
A third limitation is that all data, although drawn from multiple sources, are cross-sectional, so we cannot establish causality. Although previous research has demonstrated a causal relationship between follower-rated ethical leadership and deviance, future research in a longitudinal setting is necessary in order to confirm our findings.
Fourth, this study looked at supervisors and their immediate subordinates. Thus, we are unable to account for differences in levels of management that might influence our findings. We know that follower perceptions of leaders vary based on distance between the leader and follower (Antonakis & Atwater, Reference Antonakis and Atwater2002). While we controlled for frequency of interaction, would we find the same level of overestimation as well as the same relationships between disagreement, antecedents, and outcomes if we surveyed executives and more distant employees (i.e., those who are not direct reports)? Along this line, previous research has found a discrepancy between how top managers and lower-level employees view their organization’s ethical performance with more favorable views prevailing at the top and more negative views residing at the bottom (Treviño et al., Reference Treviño, Weaver and Brown2008). Thus, we expect that disagreement in leader-subordinate perceptions of executive ethical leadership will be greater for executive and lower-level employees compared to executives and middle- or senior-level employees.
Finally, we only considered supervisor-subordinate (dis)agreement on perceptions of the supervisor; however, disagreement among other raters (e.g., supervisors, subordinates, superiors, and peers) might also be important in terms of predicting relevant outcomes. Future research is necessary to determine how much (dis)agreement exists between other rating sources. Does (dis)agreement among rating groups share similar antecedents and consequences? At the executive level, ethical leadership researchers might consider studying (dis)agreement among an organization’s internal and external constituencies in general, or specific stakeholder groups in particular (Brown & Mitchell, Reference Brown and Mitchell2010). This type of research might provide new insights into corporate social responsibility and descriptive stakeholder theory and research.
While we examined two antecedents of supervisor-subordinate ethical leadership disagreement, there is more work to be done to add to this stream of research. An additional supervisor variable would be to examine how subordinate perceptions of behavioral integrity (Simons, Reference Simons2002)—the supervisor enacts the same values he or she espouses—are associated with perceptions of and (dis)agreement on ethical leadership. Future research could also examine subordinate individual differences as antecedents. For instance, morally disengaged subordinates or those with low moral identity may perceive the supervisor as less ethical because they are less likely to recognize the leader’s ethical behaviors. Furthermore, we collected moral entity beliefs only from the supervisor, but it could also be valuable to see if subordinate moral entity beliefs impact their perceptions of ethical leadership and (dis)agreement.
It would also be interesting to investigate the relative impact of different types of (dis)agreement on unit deviance across settings (e.g., across different types of organizations and industries). For instance, we found that both agreement on stronger ethical leadership and underestimation are related to lower levels of deviance. Are they always important? When and why might one be more influential than the other? Similarly, we found that disagreement on weaker ethical leadership and overestimation are related to higher levels of deviance. Again, under what circumstances might one be more strongly related to deviance than the other? Conducting an experiment to control and compare across the groups would shed more light on the issue and provide better guidance to organizations.
Finally, another avenue would be to examine contextual variables such as national culture to see how they impact (dis)agreement. Leaders from individualistic cultures are more likely to have inflated self-ratings (see Fleenor et al., Reference Fleenor, Smither, Atwater, Braddy and Sturm2010). Thus, the importance of being a humble leader might vary depending on whether a culture more strongly values individualism or collectivism (Hofstede, Reference Hofstede1980). Humility might be essential for leading in collectivist cultures, and thus produce more positive outcomes when it is present (and more negative outcomes when it is absent).
CONCLUSION
The present research adds to the ethical leadership literature by examining a type of leader ethical bias: specifically, the phenomenon that ‘we are not as ethical as we think we are’ (Tenbrunsel et al., Reference Tenbrunsel, Diekmann, Wade-Benzoni and Bazerman2010) and how this is associated with deviance in organizations. To date, research has focused only on subordinate perceptions of ethical leadership, yet supervisors and subordinates differ in their perceptions of leadership. Our results indicate that when supervisors overestimate their ethical leadership compared to subordinates, it is related to higher deviant behavior in organizations. By examining factors that affect the (dis)agreement between subordinates and supervisors, we begin to shed light on ways to foster greater consistency between supervisor and subordinate perceptions of ethical leadership, and ultimately reduce workplace deviance.
