When and byWhomAre Apologies Considered ? The Effects of Relationship and Victim / Observer Standing on Japanese People ’ s Forgiveness

Existing studies report on tendencies for first-party victims of a transgression to be less considerate of apologies compared to third-party observers when deciding on forgiveness. The present research investigated the conditions in which such discrepancies arise, focusing exclusively on the factor of one’s relational involvement with the transgressor. We presented Japanese participants (N = 116) with hypothetical cases of transgression, while varying on victim/observer standing, involvement with the transgressor, and the sincerity of presented apologies. As a result, the victim/observer discrepancy—where people who took the victim’s perspective rather than the observer’s perspective indicated decreased discrimination of the apology’s sincerity—emerged under high relational involvement. We further examined the role of trait evaluation of the transgressor as a mediator of the effect of apology on forgiveness. The test revealed that high involvement strengthens the mediation effect among the observers, while weakening the mediation among the victims. Based on the obtained findings, we discussed insights into how the relational factor facilitates different motives in victims and observers, leading to discrepancies in their reactions.

Following a social transgression, two factors critically determine its outcome: the account given by the transgressor and people's subsequent evaluations of the account (Hodgins & Liebeskind, 2003;Schönbach, 1990).Among the various types of accounts, an apology-where the transgressor publicly acknowledges one's responsibility for the given harm-takes the lead as a powerful catalyst for forgiveness (e.g., Darby & Schlenker, 1982;McCullough, Worthington, & Rachal, 1997;Ohbuchi, Kameda, & Agarie, 1989;Struthers, Eaton, Shirvani, Georghiou, & Edell, 2008;Zechmeister, Garcia, Romero, & Vas, 2004).Forgiveness involves a decrease in the negative intentions of retaliation or avoidance toward the transgressor along with a facilitation of conciliatory and benevolent motivations (McCullough, 2001;McCullough et al., 1997).How effective apologies are in inducing forgiveness largely depends on the manner that they are delivered and, consequently, how sincere they appear in the eyes of the recipients.The more verbally elaborated, remorseful, or spontaneous the apologies are, people will more likely perceive the transgressor as having a positive character, unlikely to repeat the wrongdoing, and hence reacceptable to the moral community (Darby & Schlenker, 1982;Gold & Weiner, 2000;Schmitt, Gollwitzer, Förster, & Montada, 2004;Weiner, Graham, Peter, & Zmuidinas, 1991).The current study focuses on this element-the manner in which apologies are given-and explores the conditions under which it influences people's decisions to forgive.Specifically, we compare the responses of victims to those of non-victimized observers of a transgression and furthermore examine how the relationship they hold with the transgressor interacts in the process of victim/observer forgiveness.

Victim and Observer's Forgiveness
The current study considers the distinction between two parties: a victim of a transgression and an observer who incurs no direct harm from the transgression.Hereafter, we refer to this distinction as the variable of standing.Of the two parties, the victims unarguably take the central role in deciding on forgiveness based on the degree that their material and psychological resources are damaged or restored (Shnabel & Nadler, 2008).But in an outbreak of a transgression, not only the victims, but observers also likely engage in punishment (Fehr & Fischbacher, 2004) and forgiveness (Brown, Wohl, & Exline, 2008)-especially when they identify with a victim or feel that the moral values they internalize are threatened (Brown et al., 2008;Carlsmith & Darley, 2008;Wenzel, Okimoto, Feather, & Platow, 2008).To date, a number of studies have directly pitted the responses of victims and observers, revealing that victims, at times, may be more inclined than third-party observers to forgive those who have hurt them (Cheung & Olson, 2013;Cooney, Allan, Allan, McKillop, & Drake, 2011;Green, Burnette, & Davis, 2008;Hashimoto & Karasawa, 2010).
Moreover, studies report on potential differences in how victims and observers consider the apology given by the transgressor.Green et al. (2008) examined people's self-recollections of either being victimized by or observing a transgression.The study suggests that the victim's forgiveness relies less on apologies compared to forgiveness expressed by observers.Risen and Gilovich (2007) pitted insincere and sincere apologies in an experimentally setup transgression.In their study, the observers distinguished the quality of the apology and expressed greater benevolence to a sincere rather than an insincere apologizer, while the victims disregarded the quality of the apology, being much forgiving toward both insincere and sincere apologizers.According to Risen and Gilovich (2007), such victim/observer incongruence in how they take into account the apology's sincerity stems from different concerns for reputation and social scripts the parties possess.

The Factor of Relational Involvement
The current study sought to further broaden the understanding of the victim/observer discrepancy in reactions to apologies.We explored the situational circumstances which drive each party to either attend to or disregard the apology's sincerity.To address this issue, we considered yet another variable which characterizes how people associate with a transgression: one's relational involvement with the transgressor.Here, relational involvement refers to the magnitude in which one is proximal to the transgressor and foresees a longer duration in the relationship.The variables of relational involvement and victim/observer standing are theoretically independent; one could be victimized by a close partner or by someone they barely know, and likewise one could observe an offense committed by one's peer or by a complete stranger.
Studies on conflicts in intimate relationships have considered relational involvement as one of the key determinants of forgiveness.They report that forgiveness is more likely to occur when the victims perceive highly involved, committed relationship with the offending partner (Finkel, Rusbult, Kumashiro, & Hannon, 2002;McCullough, Rachal, Sandage, Worthington, Brown, & Hight, 1998;Rusbult, Verette, Whitney, Slovik, & Lipkus, 1991).On the other hand, studies focusing on the observers of a norm-violation propose that people could be particularly unforgiving if they perceive the violator as psychologically proximal to oneself in terms of group membership (Marques & Yzerbyt, 1988;Marques, Yzerbyt, & Leyens, 1988).Studies however have not directly pitted the effects of rela-tional involvement on forgiveness against the effects of victim/observer standing, nor have they considered such effects with regards to the role of an apology.In addressing the potential convergent effect of standing and relational involvement, we tested the following prediction: the victim/observer discrepancy-that victims disregard the apology's sincerity relative to observers (Risen & Gilovich, 2007)-will be emphasized under conditions of high relational involvement with the transgressor.In particular, when people perceive that they are more involved with the transgressor, we expected the observer's forgiveness to rely more on the apology's sincerity, whereas the victim's forgiveness would reflect less the apology's sincerity.
We predicted that relationally involved observers would pay stronger attention to a given apology.This is because perceiving the transgressor as being in one's proximity potentially leads observers to experience greater threats to the self, thus demanding them to scrutinize the transgressor's morality through an apology.One form of such threat could be physical; people would expect greater chances of future interactions with a relationally proximal transgressor, conducing to higher risks of oneself being victimized in the future.People can also perceive a more symbolic form of threat since a transgression signifies a breach into the moral values shared by oneself as well as one's community (Wenzel et al., 2008); an immorality of a relationally close member who ought to be sharing the same values would pose a great threat to the validity of one's values and the order of the community.In reaction to both of these types of threats, one would need to assess the morality of the transgressor to confirm that she shares the group's values and is unlikely to repeat the transgression.The apology functions as an optimal indicator of such moral traits (Gold & Weiner, 2000), and hence we predicted that observers would take strong regard of an apology when it is given by somebody relationally proximal to them.
Unlike the observers, we expected that the victims who are under high relational involvement would give less attention to the quality of the given apology.This is because the perceived morality of the transgressor would be less of a primary determinant of forgiveness among the victims relative to observers.Rather, the victim's decisions would depend on a complex array of factors which are specific to victims; these include, for instance, concerns for one's psychological well-being (Karremans, Van Lange, Ouwerkerk, & Kluwer, 2003), power (Shnabel & Nadler, 2008), self-esteem (Luchies, Finkel, McNulty, & Kumashiro, 2010), and reputation (Risen & Gilovich, 2007).Above all, when a conflict occurs within a close relationship, maintaining the relationship itself becomes a dominant concern for the victims.Those who perceive high physical or psychological value in the relationship strongly engage in reestablishing the damaged relationship (Burnette, McCullough, Van Tongeren, & Davis, 2012)-as signified by studies on intimate relationships reporting that people are highly forgiving toward their partner (Finkel et al., 2002;Green et al., 2008;McCullough et al., 1998;Rusbult et al., 1991).In addition, the heightened motivation of victims to forgive close others may be habitual (Risen & Gilovich, 2007), automatic (Karremans & Aarts, 2007), and possibly extrinsic (i.e., to avoid conflict escalation: Ohbuchi & Takada, 2001).Under a condition of high involvement where victims are geared toward restoring the relationship, their motivation to carefully scrutinize the transgressor's character from the apology may become less of a priority; that is, if at least some form of mitigating responses comes out from the transgressor, be it high or low in sincerity, the victims may respond benevolently.
In generating the theorization so far, we premised that the observers under high relational involvement would focus on the transgressor's traits while the victims under high involvement would be relatively less relying on trait evaluation.The current study also empirically explored on this assumption, testing the mediational role of trait evaluation.We predicted that people's trait evaluation will significantly mediate the effect of apology's sincerity on forgiveness among the observers, especially under high relational involvement; whereas, the trait ratings by victims will decreasingly mediate the effect under high compared to low involvement.

Hypotheses and Present Research
To summarize, the current study explored people's reactions to an apology based on an interaction of three factors: the standing of whether one is an observer or the victim, ones relational involvement with the transgressor, and the sincerity of the apology they perceive.We specifically hypothesized as follows.
H 1 .Victims who have low relational involvement with the transgressor will respond with stronger forgiveness to a sincere apology than an insincere apology, while this distinction of apology sincerity will be less salient when victims are highly involved with the transgressor.
H 2 .Observers who have high relational involvement with the transgressor will respond with stronger forgiveness to a sincere apology than an insincere apology, while this distinction of apology sincerity will be less salient when observers are less involved with the transgressor.
We presented participants with vignettes depicting mundane transgressions, and tapped on their levels of forgiveness aimed at the transgressor.We additionally measured how they evaluated the transgressor's traits to test for possible mediations of the effect of apology sincerity on forgiveness.

Method Participants and Design
One-hundred and sixteen Japanese university students (65 male, 50 female, and 2 unstated) participated in the experiment; their age ranged from 18 to 26 years old with a mean of 20.74 (SD = 1.36).They were randomly assigned to one of eight conditions in the 2 (victim/observer standing) × 2 (relational involvement) × 2 (apology sincerity) design.Undergraduate research assistants individually distributed the questionnaire packets to their peers to collect data.

Materials
Questionnaire packets contained two vignettes describing different hypothetical transgressions within a university setting.We randomized the order of the vignettes across participants.While all vignettes adopted a first person writing style, the readers took either the perspective of the victim or the observer in each situation.We utilized relational involvement based on factors such as closeness or duration of relationships with the depicted transgressor, and apology sincerity through factors such as verbal elaboration, expression of remorse, and spontaneousness i .
Please see the Appendix for the sample vignettes in each condition.

Injury Vignette
One type of vignette described a situation where the transgressor accidentally bumps into and injures a classmate.
In this vignette, we manipulated the sincerity of the transgressor's apology by varying the degree of its verbal elaboration (e.g., Darby & Schlenker, 1982).We manipulated relational involvement by describing the transgressor in the high-involvement conditions as a friend belonging to the same university, while omitting such text in the low-involvement conditions.

Workgroup Vignette
The other vignette consisted of a group work setting and described a group member who failed to do the share of the work.The victim in this situation was a member of the group who had to do extra work to make up, but Effect of Relationship on Victim/Observer Forgiveness nevertheless ends up with a poor evaluation from the supervisor.In this vignette, the sincerity manipulation involved the factor of spontaneousness (e.g., Risen & Gilovich, 2007;Weiner et al., 1991) as well as verbal elaboration and remorsefulness.We manipulated relational involvement by varying the expected duration of the relationship, describing the workgroup as lasting a year (high involvement) or being one-time (low involvement).

Questionnaire
Following each vignette, participants responded to the forgiveness and trait evaluation items.

Forgiveness
We employed the Transgression-Related Interpersonal Motivation scale (McCullough & Hoyt, 2002;McCullough et al., 1997).Being one of the most widely used in research on forgiveness, the 19-item scale which we adopted from McCullough and Hoyt (2002) consisted of five items assessing the extent to which one feels vengeful toward a conflicting individual (e.g., "I wish that something bad would happen to him/her"), seven items on the extent to which one feels like avoiding the individual (e.g., "I keep as much distance between us as possible"), and seven items on the extent to which one intends to act benevolently toward the individual (e.g., "Despite what he/she did, I want us to have a positive relationship again").After translating the original items into Japanese, a bilingual colleague unfamiliar with the research objective retranslated them into English.The reverse-translated items were then matched in meaning with the original scale.In the process, we modified wordings for two items so they could be applied to the observer conditions in our study; specifically, we changed the original item of "Although he/she hurt me, I put the hurts aside so we could resume our relationship" to "Although X hurt or brought anger with me, I put them aside so we could resume our relationship," and "I forgive him/her for what he/she did to me" to "I forgive X for what X did."Participants responded to the items on a 7-point scale, anchored with 1 (Disagree), 4 (Cannot Decide), and 7 (Agree).In parallel with the existing studies (Cheung & Olson, 2013;Green et al., 2008), we created composite indices of forgiveness by averaging the 19 items with revenge and avoidance items reversecoded.The 19 items showed high internal consistency for each of the injury (α = .95)and workgroup (α = .92)vignettes ii .

Trait Evaluation of the Transgressor
We administered participants with items from the "personal friendliness" dimension of Hayashi's (1979) measure of interpersonal trait perception.We used this measure since it is the most widely used generalized measure on trait perception in Japan.We asked participants to rate the impression of the transgressor on 7-point bipolar scales with pairs of trait words (i.e., Japanese equivalent of disagreeable/agreeable, not-rude/rude, unlikable/likable, unfriendly/friendly, mean/kind, unfavorable/favorable) as endpoints of the scale, anchored on either end with very and on the midpoint with not sure.After uniformly coding the items so that higher scores indicate positive evaluation of the target, we averaged the scores to use in further analyses (α = .88for injury vignette, α = .82for workgroup vignette).

Effect of Conditions on Forgiveness
Table 1 displays the means and standard deviations for forgiveness.We conducted a mixed-design analysis of variance including the type of vignette (injury/workgroup) as a within-participant independent variable and the variables of standing (observer/victim), relational involvement (high/low), and apology (insincere/sincere) as between-participant independent variables iii .The analysis indicated main effects of victim/observer standing (F(1, 108) = 2.98, p = .09,η 2 p = .03)and apology type (F(1, 108) = 47.03,p < .001,η 2 p = .30),qualified by a three-way interaction of standing, involvement, and apology (F(1, 108) = 5.18, p = .02,η 2 p = .05;see Figure 1).A further fourway interaction of standing, involvement, apology, and vignette type did not emerge (F < .56,p = .46,η 2 p = .01).Effect of Relationship on Victim/Observer Forgiveness Post-hoc analyses revealed that in the observer condition, the apology type affected forgiveness in both the lowinvolvement (M insincere = 4.27, M sincere = 5.19; p = .01,g = 2.29) and the high-involvement (M insincere = 3.95, M sincere = 5.16; p < .001,g = 2.20) conditions.Thus, the observers discriminated the quality of the apology-being more forgiving toward a sincere as opposed to an insincere apologizer-regardless of how involved they were with the transgressor.
In contrast, among the victims, the type of apology made a difference in the levels of forgiveness under low involvement (M insincere = 3.57, M sincere = 5.21; p < .001,g = 2.99) but not under high involvement (M insincere = 4.11, M sincere = 4.53; p = .22,g = .76).In line with our expectation, the victims were indiscriminate toward the quality of an apology when they were highly associated with the transgressor.Alternate mean comparisons indicated that victims in the high-involvement condition who received an insincere apology were marginally more forgiving than their counterparts in the low-involvement condition (p = .08,g = .99);whereas, the victims in the sincere apology condition who had high involvement expressed weaker forgiveness than their counterparts who were low in involvement (p = .04,g = 1.25).

Testing the Role of Trait Evaluation
Given that discrepancies existed between the observers and victims in their response to an apology, we next examined whether the pattern could be explained by how people took into account the trait evaluation of the transgressor.We explored the role of trait evaluation as a mediator of the effect of apology sincerity on forgiveness, and specifically tested if such mediation is moderated by the variables of standing and involvement.As a preliminary analysis on whether victims and observers differed in how they evaluated the transgressor, we first ran a mixed four-way ANOVA on trait evaluation with standing, involvement, apology, and vignette as independent variables.
The general effect of apology's sincerity on trait evaluation was evident (F(1, 108) = 56.69,p < .001,η 2 p = .34),such that an insincere apologizer gained more negative evaluations than a sincere apologizer.However, we found no signs of main nor interaction effects with the factor of victim/observer standing (Fs < .2.08, ps > .15,η 2 p s < .02).According to these results, victims and observers in our experiment did not differ in how they formed the impressions of the transgressor based on the quality of the apology they received.
We next tested if the victims and observers differed in how trait evaluation associated with forgiveness.We regressed forgiveness on standing, involvement, apology sincerity, trait evaluation, and all of the possible interactions of the variables.Upon entering into the analysis, we mean-aggregated forgiveness and trait evaluation scores across the vignettes, dummy-coded the experimental conditions (0/1 with 0 for observers, low-involvement, and low-sincerity conditions), and mean-centered the independent variables.The results for this regression (adjusted R 2 = .49)indicated main effects of apology (β = .27,p = .006)and trait evaluation (β = .44,p < .001),interactions of involvement × apology (β = -.20,p = .04)and involvement × trait evaluation (β = -.25, p = .03),and most importantly an interaction of standing × involvement × trait evaluation (β = -.35,p = .003).Simple slope analyses for this three-way interaction indicated that among the observers, the trait evaluation more strongly predicted their forgiveness when they were highly involved (b = 1.33 p < .001)than weakly involved (b = -.25, p = .43)with the transgressor; the victims showed a reversed pattern in that their trait ratings became less predictive of their forgiveness when they were highly involved (b = .51,p = .15)as opposed to weakly involved (b = .77,p < .001)with the transgressor.
The results so far suggest that the effect of the mediator (trait evaluation) on forgiveness is moderated by standing and involvement.We directly tested this moderated mediation using the PROCESS macro for SPSS (Hayes, 2013).In the model, forgiveness was the dependent variable, apology sincerity was the independent variable, trait evaluation was the mediator, and standing and involvement multiplicatively moderated the path from the mediator to the dependent variable.As expected, the indirect effect of apology sincerity on forgiveness via trait evaluation was moderated by an interaction of standing and involvement (t(107) = 4.34, p < .001).According to the 95% bias-corrected confidence intervals (CI) of (conditional) indirect effects based on 2,000 bootstrap resample estimation, trait evaluation significantly mediated the effect of apology on forgiveness among observers in the high-involvement condition (b = .76,95% CI [.30, 1.20]) and among victims in the low-involvement condition (b = 1.04, 95% CI [.69, 1.44]), but not when the observers were weakly involved (b = -.09,95% CI [-.54, .51])or when the victims were highly involved with the transgressor (b = .22,95% CI [-.22, .54]).In sum, the results imply that the critical distinction between observers and victims exists in the degree that they reflect the trait evaluation of the transgressor onto the forgiveness they express, as the observers become more prone to do so while the victims become less prone to do so as results of high involvement.

Discussion
The present study presented people with hypothetical conflict situations to consider whether their responses to an apology's sincerity are moderated by whether one is a victim or an observer in the situation and also by the magnitude of relationship one holds with the apologizer.Our data signified that three variables-standing, relational involvement, and apology sincerity-interact to determine people's forgiveness.We predicted that the victim/observer discrepancy in their reactions to an apology's sincerity will exaggerate under a highly involved relationship, and this prediction generally aligned with the overall pattern of obtained results.In particular, the data supported our hypothesis for the victim's forgiveness (H 1 ): only under a weak relationship with the transgressor did they discriminate the quality of an apology, while when assuming a high relational involvement, they were less prone to incorporate the quality of the offered apology into their decisions to forgive.This pattern of victims under high involvement contradicted that of the observers in the corresponding condition, since the observers differentiated their response according to the type of apology even when they were highly involved with the transgressor.
In the hypothesis for the observers (H 2 ), we predicted that relational involvement will increase the magnitude that observers will discriminate the type of apology.This hypothesis did not receive support; the observers in our experiment unconditionally discriminated the apology's quality regardless of involvement.Perhaps the observers make distinction of insincere and sincere apologies according to the social script they internalize, as indicated by Risen and Gilovich (2007), and such script may take effect without being affected by the relationship factor.Effect of Relationship on Victim/Observer Forgiveness Our study further sought to address a cognitive component potentially contributing to the victim/observer discrepancy.We revealed that the two parties differ in how trait evaluation of the transgressor mediates the effect of apology on forgiveness.This evidence suggests that victims and observers do not differ in how they interpret the transgressor's character from the given apology; rather, an asymmetry based on standing exists in the subsequent process of how the trait evaluations determine their decisions on forgiveness.For the observers, we found an increased association between trait evaluation and expressed forgiveness to the extent that observers were relationally involved with the transgressor.In contrast, among the victims under high relational involvement, trait evaluation no longer mediated the effect of the apology, having minimal associations with their expressed forgiveness.Why would there be such a difference in how the trait evaluation affect the forgiveness of two parties?Observers under high involvement expecting future interactions with the transgressor would likely perceive greater risks of victimization to oneself or a threat to the group's norms; such situation would motivate the observers to deliberately consider the acceptability of the target via a scrutiny of the target's character.Meanwhile, the victim's tendency to disregard the perceived character of the transgressor may be an indication that other motivational factors, such as concerns for relational resources (Burnette et al., 2012), gain salience under high relational concerns, and powerfully-or perhaps intuitively (e.g., Karremans & Aarts, 2007)-determine their judgments.Future studies could more specifically test out the underlying motivational sources for the victim/observer discrepancy.
Our findings extend the theorization on victim and observer's asymmetrical responses to an apology by identifying a situational circumstance which contributes to the phenomenon's occurrence: high relational involvement.By independently operationalizing standing and relationship, we were able to clarify that the tendency for victims to minimize their attention to the apology (Green et al., 2008;Risen & Gilovich, 2007) occurs not solely based on their victimized status, but as a cross-product of being a victim and also perceiving a strong need to settle the situation.The data particularly indicated that when victims decide on forgiving somebody relationally close, their decisions dissociate from how they evaluate the target's character.Our study provides one empirical demonstration on the tendency of victims to express instrumental or extrinsically-driven forgiveness which do not fully reflect how they feel at the heart (Ohbuchi & Takada, 2001;Risen & Gilovich, 2007;Worthington & Scherer, 2004).
To generalize the findings, one important direction for future studies would be to test them in different contexts.
For example, to suit the university student sample, our vignettes focused on predicaments occurring among same aged peers in classroom settings.However, the motivation to restore harmony might be a less important determinant of forgiveness among older individuals (Girard & Mullet, 1997), perhaps because adults who engage in formal relationships are concerned less about maintaining relationships and more about the actual consequences and rewards of taking certain measures.Thus victims, as wells as observers, may show different patterns of forgiveness or reactions to apologies under different settings, such as in workplace predicaments.Testing the results of this study under a wider range of contexts and relationship constraints would be a crucial step in drawing a precise and complete picture about how different parties decide on forgiveness.
Another limitation of our study is that we employed hypothetical transgressions.Using vignettes, we were able to strictly control for the information people received when deciding on forgiveness.However, the simulational nature of the methodology may have yielded responses which does not necessarily reflect how they would actually respond in a real-life situation.Through a combination of other modes of research, such as by asking people to recall past experiences of conflicts or staging transgressions in lab experiments, we would be able to gain further external validity of the findings.
Also to further test the generalizability of the findings, it would be important to identify and control for various individual-level attributes of study participants.For instance, factors such as dispositional forgivingness, empathy, and perspective-taking account for people's forgiveness (Schimel, Wohl, & Williams, 2006;Takaku, 2001;Thompson et al., 2005).Thus, controlling for the individual differences of such factors would contribute to acquiring a clearer understanding of the phenomenon we captured in this study.It would likewise be important to bear in mind that the study utilized a Japanese sample.According to cross-cultural studies, the Japanese people indicate relatively more positive attitudes toward an apology than the North American people (Ohbuchi, Atsumi, & Takaku, 2008;Takaku, 2000).Also worthy to note, the Japanese people may be more prone to attend to the context and make less dispositional inferences than people from individualistic cultures (Miyamoto & Kitayama, 2002).Such cognitive tendency may have played a role when, in our study, the victim's evaluation of the transgressor's traits did not conduce to forgiveness under high relational involvement.The current study, together with data obtained elsewhere (Hashimoto & Karasawa, 2010), indicates that the phenomenon of victim/observer asymmetry in forgiveness is observed among the Japanese population, parallel to the findings reported in North America (Risen & Gilovich, 2007).While this by itself contributes to the literature, we cannot conclude at this point what parts of our findings are culturally specific.
In many cases of disputes, the observers do not merely observe; they are potential interveners who readily take action when necessary.Thus, our findings on how victims and observers diverge in their decisions provide insights into disputes where third-party intervention significantly determine its resolution.These include cases where friends intervene on peer-related conflicts, teachers on school-bullying, or supervisors in a workplace setting seeking to resolve injustice among subordinates.In these situations and many other day-to-day conflicts, both the victim and the third-party intervener are socially proximal to the transgressor-a condition analogous to the setting of high relational involvement in our experiment.Our data suggests that such situations likely harbor discords between how the victim and the intervener (i.e., a former observer) view an apologizer, potentially disrupting a resolution.Our study provides a framework toward understanding such issues by specifying when apologies does or does not work for parties involved in a conflict.

Notes
i) The current experiment did not employ a manipulation check item for the apology sincerity manipulation.To check how people interpret the apologies, we analyzed a data (N = 98) from a separate study (Irifune, 2013) using the same vignettes.Three items tapped on respondents' impressions toward the transgressor's actions: 7-point bipolar scales anchored on the ends with "insincere/sincere," "unreasonable/reasonable," and "inconsiderate/considerate."We averaged the scores as an index of perceived sincerity (alphas over .88 in both vignettes).This particular study incorporated manipulations of relational involvement and apology sincerity, but not observer/victim standing; all vignettes were written from the victim's standpoint.
There was one another experimental manipulation-induction of one's sense of power-so we included this variable in the analysis and conducted a mixed-design ANOVA predicting perceived sincerity with vignette type, involvement, apology, and power.We found a significant main effect of apology manipulation (F(1, 90) = 163.83,p < .001,η 2 p = .65)and an interaction of vignette and apology (F(1, 90) = 47.92,p < .001,η 2 p = .35).No other effects were significant (Fs < 1.95, ps > .16,η 2 p < .03).People viewed the transgressor's actions as more sincere, reasonable, and considerate when the individual apologizes sincerely as opposed to insincerely in both the injury (M insincere = 2.08, M sincere = 5.80) and the workgroup (M insincere = 2.18, M sincere = 4.11) cases (ps < .001,η 2 p > .34).As for the interaction of vignette and apology type, the type of vignette did not lead the perceived sincerity to differ in the insincere conditions (p > .57,η 2 p < .01),while people reported a stronger sense of sincerity toward one who sincerely apologized in the injury vignette than in the workgroup vignette (p < .001,η 2 p > .48).The results give support to how the manipulation does induce perceptions of apology's sincerity.ii) To additionally test for the factorial validity, we conducted maximum-likelihood, promax-rotated factor analyses of the 19 items.For each vignette, the analysis discriminated the three dimensions of revenge, avoidance, and benevolence under the critical eigenvalue of 1.The items loaded on the presupposed factor with loadings of more than .30except for one item in the workgroup vignette ("I have goodwill for X") which loaded most highly to the factor of revenge (-.20) instead of benevolence (.03).iii) Following a discussion that positive and negative motivations toward the transgressor may involve two distinct psychological processes (e.g., Worthington & Wade, 1999), we also conducted the same ANOVAs separately for the 7-item composites of benevolence and the 12-item composites of revenge and avoidance.In overall, the results resembled that of the single index of forgiveness, that is, we obtained the focal three-way interactions of standing, involvement, and apology (without further interaction with vignette type) for both the benevolence score (F(1, 108) = 6.10, p = .02,η 2 p = .05)and-albeit marginally significant-the composite of revenge and avoidance (F(1, 108) = 3.36, p = .07,η 2 p = .03).

Figure 1 .
Figure 1.Forgiving motivation as a function of observer/victim standing, relational involvement, and apology sincerity.Note.Error bars indicate standard errors.

Table 1
Condition Means and Standard Deviations for the Dependent Measures