Introduction

The wish to relieve guilt may motivate a confession, but the wish to avoid the humiliation of shame may prevent it.

Paul Ekman, Telling Lies

The issue of managerial ethical decision making and doing the right thing has gradually gained considerable attention in both academic and practitioner circles. Three of the past four themes of the Academy of Management Meetings focused on issues related to ethical decision-making: “Dare to Care” in 2010, “Green Management Matters” in 2009, and “Doing Well by Doing Good” in 2007. However, despite the focus in academic research, as well as calls for ethical decision making among businesses by public officials as diverse as President Barack Obama (Connor 2010) and Pope Benedict XVI (Butt 2009), the enactment of laws such as the Dodd Frank Wall Street Reform and Consumer Protection Act of 2010 and the Sarbannes-Oxley Act of 2002, we still face scandals such as Bernie Madoff’s ponzi scheme, financial meltdowns, and News of the World’s tapping of individuals’ phones. It thus becomes imperative to find ways in which organizations and managers can voluntarily and more effectively repair the damages that result from such actions. This will also help us understand how organizations or individuals in organizations responsible for wrongdoing and bad decision making can be best made to compensate their victims not just because of the fear of the law but also because they believe that they have done something that deserves restitution.

In this paper, we propose a new approach for investigating reparative behavior by looking at moral emotions and psychological proximity. Our study sheds some light on the differences between the effects of guilt and shame on compensation, in the context of different levels of psychological distance between transgressors and victims in business transactions. Previous research by management scholars has largely overlooked how emotions affect reparative behavior. For example, a search (as of September 2011) in 28 top management journals (e.g., Academy of Management Review, Academy of Management Journal, Journal of Management, Administrative Sciences Quarterly, and Business Ethics Quarterly) including the Journal of Business Ethics found less than thirty articles studying the effect of emotions on ethical behavior (e.g., Bodolica and Spraggon 2011; Cohen 2010; Connelly et al. 2004; Fulmer et al. 2009; Hartman 2008; Ho and Redfern 2010; Mencl and May 2009; Natale and Sora 2010; Payne et al. 2011; Schweitzer and Gibson 2008), and on compensating laid-off workers (e.g., McMahon 1999; Pfeifer 2007). More than half of these articles appear in a single source, the Journal of Business Ethics, and none of the articles dealt with the issue of how emotions and psychological proximity affect reparative behavior.

In many purely psychology and social psychology related journals, the relationships among emotions, ethical decision-making and general reparative behavior have been studied for decades (e.g., Eisenberg 2000; Maitlis and Ozcelik 2004; Tangney et al. 1996; Tangney et al. 2007). Shame and guilt are two of the emotions that are recognized as especially relevant to ethical decision-making and compensatory behavior (Eisenberg 2000; Tangney 1991; Tangney et al. 2007). While scholars agree that shame and guilt are two distinct emotions and lead to different actions, the exact effect of guilt and shame on compensating the worse-off party is an understudied area (Brown et al. 2008). Studies that have investigated these emotions and their effects on compensation found inconsistent and sometimes contradicting results (e.g., Bateman et al. 2003; Iyer et al. 2003; Iyer et al. 2007; Leach et al. 2006). While these studies and their findings are important, they do not resolve the question about what makes transgressors compensate victims of a wrongdoing. Despite evidence of their importance in social psychology, constructs such as guilt and shame and their effects on ethical decision-making have not been studied to the same degree in management and business scholarship. This becomes even more important when we realize that even well intentioned people sometimes make bad decisions, or engage in unethical acts with the desire to benefit their firms (Umphress and Bingham 2011), and organizations need to deal with the consequences of these decisions. This study offers a concrete investigation of transgressors’ compensation level based on their level of guilt and shame, and provides an explanation as to why we see these differences.

To understand ethical decisions and behavior, it is important to study the role of emotions. Emotionally motivated behaviors are ordered in the sequence of (a) an action occurring, (b) an emotion is experienced, and (c) subsequent behaviors based on the experience (e.g., Deci 1996). For example, when an individual does something that harms (or benefits) another party, the agent experiences emotions such as guilt (or fulfillment). This emotion subsequently encourages the agent to avoid (or re-engage in) such behavior, at least, in the near future and under similar situational factors.

In addition, we examine the role of psychological proximity between transgressors and their victims (Brass et al. 1998). It is not surprising that people tend to favor those whom they feel close to (e.g., family and friends) and thus offer them higher compensations (Wellman and Wortley 1989). Yet, it is not clear whether the mechanism leading to compensation (in this study, guilt and shame) works the same way when the transgressor feels close to the victim, as it does when the transgressor feels distant from the victim. While the term “proximity” can refer to more than only “psychological proximity,” for ease of discussion we use proximity to indicate psychological proximity only, unless clearly stated otherwise. Proximity here means actors’ perceived psychological closeness to others.

In the next section, we put forth a series of hypothesis by reviewing literature on shame and guilt. Then, we expand on the concept of psychological proximity and its implications to develop hypotheses about its interaction with shame and guilt in determining levels of compensation to the victims of a wrongdoing. We then outline our methodology and explain our results, and discuss their implications for both theory and practice.

Literature Review

This study investigates the effect of moral emotions on compensatory decision-making and compares these effects across different levels of perceived psychological distance. Moral emotions are the link between internal standards and behaviors and decisions that are morally acceptable to the agent (Tangney et al. 2007). These self-conscious emotions are not based purely on internal validations of self, but are also based on agents’ understanding of how they have been evaluated by others (Leary 2007). From the pool of moral emotions that can determine behavior, we have chosen the two most referenced emotions, guilt, and shame (Tangney et al. 2007). We argue that these two emotions, while often used interchangeably by managers (Leith and Baumeister 1998; Tangney 1991), have different effects on decisions. Several studies have investigated the psychological differences between shame and guilt (Brown et al. 2008; Leary 2007; Leith and Baumeister 1998; Lindsay-Hartz 1984; Orth et al. 2010; Tangney 1991; Tangney et al. 2007); however, their differences in affecting ethical decision-making remain substantially under investigated (Brown et al. 2008). Furthermore, their effect is unknown when the transgressed are at different levels of psychological distance from the perpetrator. To explore the effect of these two emotions on compensation, we first introduce the two concepts and describe how they are different from each other.

Shame and Guilt

Many people can recall moments of feeling guilty or ashamed distinctively, but they might also use the two terms interchangeably, as these emotions have many similar characteristics (Leith and Baumeister 1998; Tangney 1991). Both guilt and shame are considered negative emotions (Leary 2007; Tangney 1991; Tangney et al. 1998; Tangney et al. 2007), but they differ in their relationships to morality. Smith et al. (2002) suggested that shame is relevant to both moral and non-moral related transgressions, but guilt is only a result of moral transgressions. Both emotions are cognitive responses as one perceives being viewed by others following a behavior or decision (Leith and Baumeister 1998; Tangney et al. 2007).

Each of these emotions has distinctive causes and effects on the actor and people around them. Guilt is defined as an agitation-based emotion of regretting a wrong action or decision (Ferguson and Stegge 1998). Based on this definition, guilty transgressors perceive the wrongdoing aspect of their actions or decisions, assume responsibility for them, and desire to find a way to either undo the wrong or punish themselves (Eisenberg 2000; Lindsay-Hartz 1984). In this way, the transgressor is capable of putting him or herself in victims’ shoes and seeing the issue from their perspective (Leith and Baumeister 1998; Tangney 1991); hence these emotions are called cognitive responses to a behavior. Since guilty people can feel the pain as the victims feel, they are motivated to take other-oriented actions, such as apologizing to and compensating the victim, in order to eliminate the negative feeling. Guilty people perceive their behavior, but not themselves, as the target of criticism (Tangney 1991); therefore they can correct their wrongdoings without feeling humiliated or threatened.

Shame, on the other hand, is a dejection-based emotion of condemning one’s entire self (Ferguson and Stegge 1998; Gramzow and Tangney 1992; Leary 2007; Tangney et al. 1996). When ashamed, the transgressor feels like the target or even victim of criticism (Tangney et al. 1996). People feel ashamed when they feel they are bad people (Leary 2007). Feeling ashamed, the transgressors devalue themselves and fear contempt from others (Ferguson and Stegge 1998). Emotions of shame are more painful than those of guilt, because ashamed people perceive an attack to their self and identity (Lindsay-Hartz 1984; Tangney 1991; Tangney et al. 2007). Because of the threat to the self and fear of scorn, ashamed people feel like the target of criticism and usually respond by self-oriented actions, such as avoiding others and hiding away (Orth et al. 2006; Orth et al. 2010; Tangney 1991).

Ashamed people perceive that blame is focused on themselves, but guilty people perceive it to be focused on their behavior (Leary 2007; Tangney et al. 2007). Ashamed individuals are more likely to blame others or get angry at them for creating shame-eliciting situations, and are not likely to undertake any measure to change the wrongdoing (Leith and Baumeister 1998; Tangney 1995). Guilt, in contrast, being based on the ability to take other people’s perspectives, a cognitive ability that did not result from shame, is more likely to activate senses and behaviors that can benefit others (Leith and Baumeister 1998). For example, de Hooge et al. (2007) found that ashamed people are not motivated to cooperate. However, they found that individuals, who would usually act on self-interest, are more likely to cooperate when they feel guilty. Unlike shame-prone individuals, guilt-prone individuals are more likely to take responsibility for their actions and correct their wrongdoings (Tangney et al. 2007).

Therefore, it seems that the guilt and shame produced by transgressions are not equally responsible for the compensatory behavior that follows. Guilty individuals would try to correct the transgression, and either accept retribution from the victim or give out compensation. We expect guilty people to take some measures to compensate their victims. Ashamed transgressors, on the other hand, would be more likely to feel like they want to hide from everyone and as a result, would be less likely than transgressors who consider themselves guilty to compensate their victims (Leith and Baumeister 1998; Tangney 1995). Therefore, we propose that guilty people would be more willing than ashamed people to compensate victims of their transgression.

H1

Guilt is more likely than shame to lead to compensating victims.

Both emotions, as discussed at the beginning of this section, are based on discrepancies between one’s behavior and other people’s expectations. Since people evaluate themselves from the perspectives of significant others (e.g., friends, relatives, spouses, and children) as well as everybody else around them, the level of guilt and shame in people should be a product of strength and weakness of ties with the victim. In the next section, we discuss how psychological proximity affects the level of guilt, shame, as well as compensation.

Psychological Proximity

Based on the theory of self-discrepancies (Higgins 1987), there is an actual self and an ought self, and they are different from each other most of the time. Actual self refers to what we are and how we behave. Ought self is derived from other people’s, especially family and friends’, expectations of us. When there are inconsistencies between what we actually do (actual self) and what we are expected to do (ought self), we may feel guilty or ashamed. For example, we are expected to look out for our family members. If someone secretly puts his or her family at risk for his or her own financial gains, this person will feel guilty and/or ashamed in front of the family. These assumptions are consistent with the theory of planned action (Fishbein 1967; Fishbein and Ajzen 1975).

The theory of planned action (Fishbein 1967; Fishbein and Ajzen 1975) suggests that individuals’ actions are affected by normative beliefs and subjective norms. Normative beliefs refer to behaviors that are acceptable by people close to us—collectively categorized as “significant others”—such as parents, siblings, spouse, friends, and family. For example, being honest to each other is a normative belief among family members and close friends. Subjective norms are derived from norms in the broader society, and include people with looser ties to us than significant others. These people, in addition to significant others, could include community, acquaintances, and anyone related to us based on the context, location and time circumstances. Both normative beliefs and subjective norms play an important role in construing our ought self. What our significant others expect and what the general society expects us to do are both parts of our obligations.

At the center of this discussion is whether people closer to us (e.g., significant others) have more impact on our ought selves than strangers. If so, when significant others are involved, discrepancies resulting from the violation of normative standards have stronger influence on emotions compared to when strangers are involved. While both the theory of planned action and theory of self-discrepancy give equal weight to significant others and to general society, it seems intuitive that people closer to us might play a more important role in influencing our behavior. First, significant others are fewer in number compared to the broader community and society where individuals live, but each category has an almost equal weight in the theories. Therefore, any individual significant other being a friend, relative, teacher, or colleague, has more influence over our behavior and what is accepted from us than an individual member of the broader society. Second, significant others are also members of the broader society. Their influence in setting or defining acceptable rules of subjective norms is minor, but it still exists. Therefore, significant others have a larger influence on us and our behavior than others.

The importance of significant others is manifested as perceived closeness to significant others, termed “psychological distance” or inversely put “psychological proximity”. Psychological distance can also be used to describe actors’ subjective experience of objects, events, or other people in terms of their proximity in time and location (Trope and Liberman 2010). The point of reference for psychological distance is self, and distance is measured based on perceived proximity to self (Trope and Liberman 2010). For instance, the physical distance between two people on a crowded bus is an absolute value, such as one foot. However, one person might perceive this distance as too close and the other might think this is an acceptable distance. In terms of relationships, psychological proximity (or psychological distance) toward another person is one’s subjective experience of the perceived closeness (or distance) to that person. This latter definition is what we have used in our study.

The idea of psychological proximity has been used in research on management and ethical decision-making (Brass et al. 1998; Flannery and May 2000; Jing et al. 2009; Jones 1991; Mencl and May 2009; Novicevic et al. 2008). For example, Jones (1991) proposed an issue-contingent model that includes six constructs: (1) magnitude of consequences, (2) social consensus, (3) probability of effect, (4) temporal immediacy, (5) proximity, and (6) concentration of effect. In making ethical decisions, Jones argued that “intuitively, people care more about other people who are close to them (socially, culturally, psychologically, or physically) than they do for people who are distant” (Jones 1991, p. 376). Perpetrators’ ethical actions are contingent on how close they perceive their victims to be; the closer the victim is perceived, the more likely the decision will be ethical.

Brass et al. (1998) used social network theory to explain the importance of proximity in ethical behavior. Our social ties, they argued, are grouped as strong ties and weak ties. Close relationships such as those we have with our in-groups (e.g., family and close friends) are stronger ties, and relationships with acquaintances and people we only know (i.e., out-groups) are weak ties (Granovetter 1973). Brass et al. (1998) also proposed that stronger ties diminish the effect of situational and normative constructs.

Mencl and May (2009) tested the influence of psychological and physical proximity in interaction with magnitude of consequences on ethical decision-making. They defined psychological proximity as agents’ perceived closeness to others, and physical proximity as the physical distance to a potential victim. They suggested that decision-makers’ decisions about employees are affected by the interaction between their information about the consequences of their decision and how close they feel to the employee (Mencl and May 2009). They did not, however, find the two types of proximity to have significantly different effects on ethical decision-making.

The research reviewed so far indicates that emotions are created by peoples’ actual behavior violating their set of expected behavior (Higgins 1987). So the set of standards—whether set by subjective norms, normative behavior, ought self, ideal self or actual self—determine peoples’ actual behavior as well as their intended behavior (Higgins 1987). Emotions derived from the inconsistencies between standards and behaviors are products of peoples’ relationships with others (Tangney 1995). Baumeister et al. (1994) also posited that the more people feel concerned about one another, the more likely they would feel guilty if they did something wrong. This indicates that guilt arises from within relationships, as transgressions against people important to the wrongdoer creates more guilt (Baumeister et al. 1994). Therefore, we propose here that as the psychological distance between perpetrator and victim reduces, failure in matching standards with behavior would result in more negative emotions of guilt and shame.

H2a

The closer the perpetrators perceive their psychological proximity to their victims, the guiltier they feel.

H2b

The closer the perpetrators perceive their psychological proximity to their victims, the more ashamed they feel.

Although there has been limited attention on the relationship between psychological proximity and compensation, there is a strong body of literature about advantages people enjoy from relationships and ties (Chand and Ghorbani 2011; Chow and Ng 2004; Cohen et al. 2006; Ibarra 1995; Neyer and Lang 2003; Rowley 1997; Westphal and Milton 2000). For example, theories of kinship (Korchmaros and Kenny 2006; Neyer and Lang 2003) and network theories (e.g., Chand and Ghorbani 2011; Granovetter 1973; Jing et al. 2009; Stam and Elfring 2008; Zhang et al. 2008) suggest that stronger ties (e.g., family and close friends) are more helpful during difficult times than weaker ties (e.g., acquaintances). Korchmaros and Kenny (2006) also suggested that family relationships and perceived closeness could predict helping behavior. Together, these studies indicate a potential connection between proximity and favors. When one is closer to another and has stronger tie with the person, he or she is more likely to act in ways which will benefit the other party (Wellman and Wortley 1989).

While there is a lack of direct empirical support from the literature because of the dearth of studies done in this regard, we consider compensation after wrongdoing as a benefit to the victim, and based on our deductions from kinship and network theories, expect proximity to affect the potential for compensation.

H2c

The closer the perpetrators perceive their psychological proximity to their victims, the more likely they are to compensate them (victims) for the wrongdoing.

Interaction of Shame and Guilt with Proximity

Other than the direct impact on compensation, proximity may also interact with shame and guilt in determining the level of compensation. As indicated earlier, there is a dearth of empirical evidence on behavioral consequences of shame and guilt with respect to compensating others (Brown et al. 2008). Nonetheless, studies on compensation of out-groups provide some indirect evidence of the role of proximity.

In one study, Iyer et al. (2003) investigated the effect of group-level guilt and shame on the tendency to compensate out-groups, and they concluded that “White guilt,” felt by European Americans, could predict support of this group for programs such as affirmative action that attempt to compensate African Americans (i.e., out-group members) for historical racial discrimination. However, this study does not distinguish between the effects of guilt vis-à-vis shame, because shame was included as an item measuring guilt. In another study, Iyer et al. (2007) investigated the effect of guilt and shame separately and did not find guilt to predict any action or behavior. The results of this study were somewhat contradicted by Brown et al. (2008) who also examined group-level shame and guilt, and compensation of out-groups for historical wrongdoings. They found guilt to have a longitudinal causal effect on compensating out-groups for historical wrongdoing, but did not find shame to have a similar effect. While there are differences in the outcomes of these studies, one clear theme that we can distil is that guilt works the same when it comes to out-groups and in-groups; therefore, we do not expect a moderating effect of proximity on the relationship between guilt and compensation.

H3a

Guilt’s effect on compensation is the same across different level of psychological proximity.

However, shame seems to have a stronger predictive power on compensation when the victims are out-group members. One common theme in all the above studies (Brown et al. 2008; Iyer et al. 2003) is that they follow Lickel et al. (2006) to suggest that shame determines compensation because the group tries to remove the negative image from the in-group and restore its own good reputation. For example, Brown et al. (2008) noticed that historical group shame could predict the tendency to compensate out-groups, but this effect was mediated by the intention to improve or maintain in-group reputation. Iyer et al. (2007) found that American and British peoples’ shame of their countries’ occupation of Iraq predicted reparative behavior (i.e., support for withdrawal of foreign troops from Iraq). While these studies focus on group level shame and guilt, we argue that the same logic should be applicable, to some extent at least, to individuals as well.

Individuals’ shame is related to their personal image. Once individuals feel that their image has been tarnished, it is important for them to restore their reputations. However, transgressions do not easily ruin one’s reputation in the eyes of significant others. Those with close psychological proximity have a thorough understanding of one another, so reputations are relatively stable. On the other hand, there are usually fewer interactions with individuals who are psychologically distant, and each individual interaction, good or bad, has a stronger impact on one’s reputation. Therefore, endorsing or engaging in compensatory actions can dramatically improve reputations in the eyes of people who are psychologically distant. As such, we suggest that perpetrators’ shame for wrong doing predicts their compensatory action more when victims are psychologically distant.

H3b

Psychological distance positively moderates shame’s effect on compensation, such that the higher distance, the stronger effect of shame on compensation.

Interactions between the two moral emotions and psychological proximity are different. When individuals feel guilty about transgressions, they want to compensate the victims regardless of their psychological proximity. In contrast, individuals who feel ashamed of their transgressions are more likely to compensate people who are more psychologically distant, than those who are psychologically close.

Methodology

We designed a within-subject scenario study to investigate our hypotheses. Participants read a story about a manager who was facing a decision to compensate for his or her wrongdoing toward different victims. We manipulated the type of victims along the dimension of psychological proximity, and tested whether the level of shame and guilt influenced the ethical decisions regarding compensations made toward victims at different levels of psychological proximity. Types of victims included family members, friends, strangers with concrete descriptions, and strangers with abstract descriptions. Construal Level Theory suggests people feel psychologically closer when the description of the target is concrete than when it is abstract (Trope et al. 2007), and thus creating two types of strangers leads to further examination of the role of psychological distance.

Participants

Two hundred and sixty-one undergraduate business students from a large west coast university participated in exchange for partial course credit. There were 145 females and 116 males. Their mean age was 21.17 years (SD = 2.53).

Procedure

Participants were asked to imagine that they were managers and read about an ethical dilemma they must solve. In the scenario, the manager had sold garden-level housing units to different people. Later, the manager receives a report from an architect suggesting if a flood occurs there is a significant possibility of massive damage to the units. The manager had not done anything to make these homes flood proof. One year later a flood occurs in the area and most units are flooded. However, lawyers are in disagreement as to whether the manager is legally responsible to compensate homeowners. Participants were then asked to indicate how much they would like to compensate each of the following owners: (1) their parents, (2) close family members or relatives other than their parents, (3) close friends, (4) a retiree who is looking for a low maintenance investment that could generate rental income for him or her, (5) a middle-income couple who are expecting their first child and in need of a bigger and more affordable house, (6) a stranger they met at a social occasion, and (7) a total stranger they know nothing about. Participants were also asked to report how ashamed and guilty they felt facing each of the above owners before they made the decision to compensate. They then answered questions on control variables and filled in their demographic information.

Measurements

Shame and Guilt

Shame and guilt were measured with items drawn from previous studies (e.g., Dahl et al. 2005; Harder and Zalma 1990). We drew selected items from previous studies to create the shame and guilt scale for the current research because (1) existing scales focus on shame- and guilt-proneness instead of feelings after transgressions, and (2) we were concerned the survey would be too long for our participants if we used a 16-item scale. We asked a five person focus group to choose items from Harder and Zalma’s (1990) 16-item pool whose meanings most resembled feelings of guilt and shame. Three items for shame and four items for guilt, which were thought to relate to corresponding constructs where selected. Consultation with three experts in scale development suggested those items have high face validity. The three items measuring shame are ashamed, embarrassed, and foolish; and the four items measuring guilt are guilty, remorseful, worried about upsetting someone, and worried about hurting someone. The internal consistency reliabilities, as indicated by Cronbach’s alpha, were above 0.80 for shame and 0.83 for guilt on all seven target owners.

Compensation

Participants indicated the portion of the total damage they would be willing to compensate to each owner on a 5-point Likert scale (1 = 0 %, 2 = 25 %, 3 = 50 %, 4 = 75 %, and 5 = 100 %).

Proximity

Participants indicated how close they felt to each owner on a 5-point Likert scale (1 = not close at all, 5 = very close). Proximity was used as manipulation check.

Internal Locus of Control

Locus of control was measured by an eight-item sub-scale of internal locus of control from Levenson and Miller (1976) and included as a control variable in line with previous research (Tangney et al. 1996; Tracy and Robins 2006; Treviño 1986). Participants reported their locus of control by rating a statement about themselves on a 5-point Likert scale (1 = strongly disagree to 5 = strongly agree). The internal consistency reliability, as indicated by Cronbach’s alpha, was 0.75. The average score of the eight items was calculated and used as a control variable in regressions (Detert et al. 2008; Treviño 1986).

Religiosity

Participants indicated how religious they were on a 5-point Likert scale (1 = not religious at all, 5 = very religious). This variable was used as a control variable in regressions in line with related previous research (de St Aubin 1996; Quiles and Bybee 1997; Tangney 1992).

Results

Manipulation Check

A pilot study showed the seven types of owners gathered around three clusters based on the psychological distance participants felt toward them. Based on Construal Level Theory and the results from the pilot study, we categorized the seven owners into three groups: (1) in-groups including parents, close family members or relatives other than parents, and close friends, (2) strangers with concrete descriptions (strangerCON) including retiree and middle class couple, and (3) strangers with abstract descriptions (strangerABS) including a stranger they meet at a social occasion, and a total stranger they knew nothing about. The average proximity scores were calculated for each group. In order to confirm the three groups were different on psychological distance, we first ran a repeated measures analysis of variance (ANOVA) and found a significant difference in proximity among the three groups (F(2, 259) = 1186.73, p < 0.01). Two post hoc paired-samples t tests showed participants felt closer to their in-groups than to strangerCON (t(260) = 33.18, p < 0.01), and they felt closer to strangerCON than to strangerABS (t(260) = 10.34, p < 0.01). These results indicated that participants felt closest to their in-groups, followed by strangerCON, and felt most distant to strangerABS. The results supported our logic of grouping, and thus all key variables were averaged within each group. We conducted the following analyses within each group first and then examined how the hypothesized regression model differed across the three groups. Means, standard deviations, and correlations of key variables can be found in Table 1.

Table 1 Means, standard deviations, and correlations of key variables
Full size table

Regressions

In order to test Hypothesis 1, we examined whether and how shame and guilt predict ethical decision-making (compensation) within each group. We ran a multiple regression to investigate the proposed model in each group separately. Age, gender, degree of religiosity, and internal locus of control were entered as control variables. For in-groups, guilt (β = 0.44, p < 0.01), but not shame (β = −0.06, p = ns), predicted compensation. For strangerCON, both shame (β = 0.30, p < 0.01) and guilt (β = 0.30, p < 0.01) predicted compensation. For strangerABS, both shame (β = 0.25, p < 0.01) and guilt (β = 0.27, p < 0.01) predicted compensation. The results partially supported Hypothesis 1 as guilt was a stronger predictor than shame for in-groups, but both emotions predicted compensation equally well for strangers. Detailed regression results are presented in Table 2.

Table 2 Regressions across three groups
Full size table

ANOVAs and t Tests

In order to test whether proximity affects the level of shame, guilt, and compensation (H2a-c), we ran three repeated measures ANOVAs with group as a within-subject factor. Results indicated a significant difference on shame across the three groups (F(2, 259) = 185.41, p < 0.01). Two post hoc paired-sample t tests showed that participants indicated higher levels of shame to their in-groups than to strangerCON (t(260) = 8.85, p < 0.01), and higher levels of shame to strangerCON than to strangerABS (t(260) = 13.87, p < 0.01). Analyses on guilt and compensation showed similar patterns. Participants felt most guilty and offered the highest compensation to their in-groups, followed by strangerCON, and felt least guilty and offered lowest compensation to strangerABS. Consistent with our predictions, lower levels of psychological proximity lead to higher levels of shame, guilt, and compensation. Results of repeated measures ANOVAs and paired-sample t tests for each variable can be found in Table 3.

Table 3 Results of repeated measures ANOVA and paired-sample t test on key variables
Full size table

Moderation Analyses

To test Hypothesis 3, we examined whether the regression model differs across the three groups. We first compared the in-group with both stranger groups to determine whether there is an overall effect of grouping between in-groups and strangers. Following guidelines for testing moderation using ordinary least squares (OLS) regression in within-subject designs (Judd et al. 2001), we found a marginally significant difference in the regression coefficient of shame on compensation among the three groups (t = 1.69, p < 0.10), but no difference in the regression coefficient of guilt on compensation (t = −1.20, p = ns). In addition, guilt (t = 4.64, p < 0.01) mediated the effect of grouping of proximity on compensation.

Next, we ran two post hoc regressions to further investigate the moderating effect of proximity on the relationship between shame and compensation. Comparing in-group and strangerCON, we found evidence that the effect of shame on compensation was moderated by proximity (t = −2.44, p < 0.05). Comparing strangerCON and strangerABS, we found shame was not moderated by proximity (t = 0.50, p = ns). Overall the results suggest that when facing in-groups, the more guilty people feel, the more likely they are to offer higher levels of compensation to victims. However, the level of shame does not affect the decision in the same way as that of guilt does. When facing strangers, regardless of how concretely or abstractly they are described, people make decisions based on both how ashamed and guilty they feel toward the strangers. Regression results are depicted in Fig. 1.

Fig. 1
figure 1

Regression of shame and guilt on compensation among the three groups

Full size image

Taken together, our results supported our hypothesis that the degree to which perpetrators feel guilty or ashamed, and their subsequent decisions to compensate victims, are functions of perceived psychological distance to the victim. The closer people feel toward the target, the more likely they will feel ashamed and guilty about the past harmful behavior, and therefore offer higher compensation. Moreover, the effect of ethical emotions on decision-making varies when the victim is an in-group member versus when the victim is a stranger. When the victim is an in-group member, guilt is a predictor of compensation while shame is not. When the victim is a stranger, both shame and guilt predict compensation equally well.

Discussion

We tested the impact of individuals’ perceptions of psychological distance on the level of guilt and shame, and on their tendency to compensate the victim of wrongdoing. There were three groups representing three different levels of psychological proximity on a continuum from very close relationship to the parents to total strangers that we know absolutely nothing about and have no prior relationship with.

Regression analyses (Table 2) indicate partial support for our first hypothesis that guilt is a better predictor for compensation than shame. Guilt predicted compensation in all groups, whereas shame only predicted compensation to strangers. This is consistent with previous predictions that guilt is more responsible for reparative actions than shame (Lindsay-Hartz 1984; Orth et al. 2010; Smith et al. 2002; Tangney 1991). Shame was not predictive of in-group compensation, which is consistent with the literature on shame. Tangney (1992) had posited that ashamed individuals are unlikely to rectify their wrongdoings. When feeling ashamed, the perpetrator is not thinking of going and amending the wrongdoing, but thinking of hiding from others (Lindsay-Hartz 1984). In-groups are comprised of parents, other close relatives, and close friends, and it is difficult to avoid them. Therefore feelings of shame, while they may cause individuals to want to hide from in-groups, may not in reality be acted upon since it is difficult to do this in case of in-groups. This might explain why shame does not predict in-group reparative behavior. This is not to say that individuals do not experience shame when transgressing people close to them. In fact, levels of shame were higher when victims were members of in-groups, versus out-group victims. It is possible that individuals expect people who are close to them to be more forgiving and as such do not think shame is a reason to compensate them. Another possible explanation is that individuals see compensation to in-group members as an obligation and nothing to do with their level of shame.

We also found support for our second set of hypotheses. We posited that psychological proximity can influence the level of shame, guilt and compensation, in that the closer individuals feel to their victims, the more likely they would be to feel guilty and ashamed, and offer higher compensation as a result. Participants reported the highest levels of guilt, shame, and compensation when the victim was an in-group member, followed by when the victim was a stranger with a concrete description, and the lowest levels of guilt, shame, and compensation resulted when the victim was an abstractly described stranger.

In the third set of hypotheses, we predicted how psychological proximity interacts with shame and guilt in individuals’ decisions to compensate victims of transgression. While the levels of guilt and compensation increased as perpetrators felt closer to the victims, the interaction of proximity and guilt in predicting the level of compensation were not significant. However, we found some support for the mediating effect of guilt between perceived psychological proximity and the level of compensation.

This latter finding is perhaps not too surprising because guilt is the emotion affects individuals through empathy (Leith and Baumeister 1998; Tangney et al. 2007). Empathetic people are capable of taking the perspective of others (Mehrabian and Epstein 1972). Empathy is a source of altruistic behavior as people who help others out of empathy actually relieve themselves of negative feelings (Batson et al. 1981; Detert et al. 2008). Consistent with this logic, when perpetrators feel close to victims of their wrongdoings, they feel pain as the victim does, regret their behavior, and want to do things to relieve the victims’ pain to relieve the perpetrators’ own painful feelings. Guilt is associated with other-oriented empathy (Tangney 1991; Tangney et al. 2007).

Mencl and May (2009) also suggested that empathy increases as proximity decreases. The more individuals can relate to others and become familiar with them, the more they can empathize with them (Brass et al. 1998). Mikulincer et al. (2005) posited that individuals distance themselves from the suffering of others to reduce empathy and the possibility of altruistic help. The closer individuals feel to their victims the more empathetic and guiltier they feel for transgressing them. Higher levels of guilt also indicate higher levels of compensation. As such, guilt could be a good mediator for the effect of proximity on the level of compensation.

We found partial support for hypothesis 3b, which proposes that proximity moderates the effect of shame on levels of compensation. Our findings suggest that psychological proximity moderates the effect of shame as a categorical variable instead of a continuous variable. We found dissimilarities in compensation offered to in-group members versus strangers. However, despite our findings (H2) that people perceived different levels of shame, compensated varying amounts, and felt like they were at different psychological distances toward the two stranger groups, there was no difference in the predictive power of shame on compensation between the two stranger groups. When perpetrators decided on the level of compensation to offer in-group members, shame was not a determinant of the amount. On the other hand, when deciding on the level of compensation to offer strangers, shame was a significant predictor of compensation.

This finding indicates that ashamed transgressors categorized their psychological distances to their victims based on victims’ statuses as either in-group or out-group members. This type of dyadic categorization and moderation is consistent with previous studies about group-level tendencies to compensate out-groups (Berscheid et al. 1968; Berscheid and Walster 1967; Brown et al. 2008). Shame is caused by violations of character (Keltner and Buswell 1997) and is related to individuals’ identities. Transgressors try to compensate victims to reduce their feelings of shame and restore their reputations. Therefore, shame can transform perpetrators’ image and gives them a lower evaluation of themselves; an image different and less worthy of one to self (Lindsay-Hartz 1984).

Perpetrators do not defend their images and reputations in front of in-group members to the same extent as they do in front of out-group members, since in-groups are expected to accept each other even when it is difficult. In addition, it is hard to imagine a damaged in-group image resulting from one mistake, when in-group members know one another so well. Therefore, compensating an in-group member is not likely to change how the in-group members perceive the transgressor if the damage was caused by a one off event. This could explain why shame was not a significant predictor of in-group compensation. If we look closer at shame, it is associated with running away and hiding from others (Tangney 1991; Tangney et al. 2007; Tangney et al. 1992). Transgressors might still feel ashamed to face members of their in-groups, as indicated by the first two sets of supported hypotheses. However, results do not indicate that shame is the reason for transgressors to compensate the in-group.

In contrast, transgressors try to protect and restore their own images in front of out-group members. When perpetrators feel ashamed and condemn themselves, they still want to keep a positive image in front of other people who do not know them well. After all, strangers often judge each other based on one action or behavior. So it is important for perpetrators to make the wrong right in the eyes of strangers, in order to restore a good image. In addition, a single act such as compensation can make a large difference in their image with strangers, but will probably not make as much of a difference with the in-group. By offering compensation, transgressors send out the message that they take responsibility for their behavior. This positive image can help restore reputations that were damaged during transgressions. Therefore, when transgressors feel distant to their victims and think their images are vulnerable, they need to take actions to protect or improve their image from victims’ point of view.

Implications and Directions for Future Studies

Our findings indicate that while guilt increases as proximity decreases, guilt can predict transgressors’ tendencies to compensate victims. This is an important finding as it indicates that increasing the level of guilt in managers and entrepreneurs could improve their ethical behavior. We suggested that guilt works through empathy and empathy is also linked with psychological proximity. We also posited that empathy increases as concrete knowledge of victim increases. While our suggestions needs further testing, these findings indicate that when people provide more information to firms, they seem more concrete to managers. As the level of abstraction decreases, managers are likely to feel more empathetic. Higher levels of empathy and more concrete information about clients could insure better treatment by managers.

This better treatment could occur in two ways. The first is that managers who feel more empathetic toward clients may avoid decisions that cause transgressions. For example, managers might develop preemptive guilt through interactions with the board of directors when the board emphasizes the reality of the people who can be hurt by these decisions. The second way that clients could be protected against wrongdoing is through compensation. Managers who feel closer to their clients are more likely to compensate them in the event of any transgression. Therefore, the frequency of compensation might increase when clients and customers have the opportunity to meet face-to-face and get to know one another better. Both of these mechanisms provide opportunities for further research.

Another area that would benefit from further study is the implication for online businesses where almost all customers are known only at an abstract level. Abstract relationships in an online environment are less prone to being influenced by feelings of guilt. Based on our findings, we expect that the possibility and level of compensation would diminish in e-commerce situations compared to brick-and-mortar businesses. However, shame is more likely than guilt to determine compensation in this case. Bringing issues of wrong doing to the public could increase victims’ chances of being compensated. Perhaps this explains why blogs and online customer opinions are increasingly important. From the point of view of online customers, it is important that they make their voices heard in an organized and systematic manner regarding corporate wrongdoing, since this would create more shame for the concerned organization and increase the chances of victims of corporate fraud being compensated.

One final area for future research is to combine this line of research with antecedents of guilt and shame. Personality traits, values, culture, guilt-proneness, and shame-proneness have all been shown to have some relationship with feelings of guilt or shame. Culture, values, and personality possibly influence levels of guilt and shame and could change individuals’ perceptions of psychological proximity (e.g., some cultures value family more than others, while in some other cultures societal or out-group collectivism could be higher). These antecedents could help explain how the level of guilt and shame can be increased or decreased. For example, how does guilt-proneness lead to the level of guilt when levels of concrete knowledge and psychological proximity are controlled? Do psychological proximity and guilt-proneness interact in the same ways as guilt and proximity?

Conclusion

Our purpose was to provide an explanation as to how and why people attempt to compensate others for their unethical decisions. In this study, we investigated the effect of guilt, shame, their interaction with the perceived psychological proximity to the victim in affecting the transgressor’s reparative behavior. We found that guilt increases as proximity decreases, and that guilt can predict transgressors’ tendencies to compensate victims. Shame interacts with psychological proximity to predict compensation, whereas guilt mediates the relationship between psychological proximity and compensation. Through this study, we increased our knowledge about the effects of guilt, shame, and psychological proximity on decisions leading to reparative behavior. We hope that this research can help with improving the frequency of moral reparative actions, including righting previously committed wrongs. While we do recognize that there are always anomalies and exceptions (for example, Madoff personally knew many of his victims), we believe that this study has helped us gain a better understanding of how organizations and decision makers in organizations can better compensate their victims and right their previously committed wrongs.