The Mandate of the Collective: Apology Representativeness Determines Perceived Sincerity and Forgiveness in Intergroup Contexts

Abstract

The sincerity of an apology is often critical for it to be viewed positively by victims. For collective apologies, we argue that sincerity takes on a particular meaning: It is a function of the apology’s perceived representativeness for the offender group’s will or sentiment. Consistent with this notion, when an apologetic (vs. nonapologetic) message was democratically chosen (Study 1) or explicitly endorsed by the majority of the offending outgroup (Study 2), it was considered more sincere and, through this, led to more forgiveness. Furthermore, while disagreement about an apology within the offender group reduced its perceived representativeness and sincerity, this was less so when the dissenters could be subtyped: when disagreement was correlated with an existing subgroup within the offending outgroup (Study 3) and in line with expectations for that subgroup (Study 4). This research shows that victim group members consider intragroup processes within the offending outgroup for attributions of sincerity.

Keywords

intergroup processes forgiveness collective apologies sincerity representativeness

Collective apologies have become a common occurrence in public life and politics, to an extent not seen just a few decades ago (Brooks, 1999; Okimoto, Wenzel, & Hornsey, 2015). Collective apologies are commonly invested with the hope that they facilitate moral and relational repair; that they help restore justice and foster positive relations between the groups (Smith, 2008). However, there has only been mixed evidence that collective apologies elicit forgiveness among victim group members (see Hornsey & Wohl, 2013, for a review). Psychological research has thus set out to better understand the conditions under which collective apologies may be effective in eliciting forgiveness and promoting reconciliation.

Rather than extrapolate from interpersonal apologies, it is important to identify the unique features of intergroup contexts that impact the apology–forgiveness link (Wohl, Hornsey, & Philpot, 2011). The present research undertakes such an investigation in relation to an issue that seems to be crucial for apologies in interpersonal and intergroup contexts alike: the attribution of sincerity to apologies. We propose that sincerity takes on a particular meaning in intergroup contexts, as collective apologies are a measure of the sentiments, true feelings, and benevolent intentions not so much of the individuals expressing their apology but rather the group they represent.

Apologies and Forgiveness

Evidence that collective apologies promote intergroup forgiveness is limited. Among the positive findings, Brown, Wohl, and Exline (2008, Study 1) found that experimentally manipulated information about a U.S. apology for a friendly fire killing of Canadian soldiers in Afghanistan increased forgiveness among Canadian participants. Leonard, Mackie, and Smith (2011) found a similar positive effect on university students’ forgiveness for an apology (vs. refusal to apologize) by university professors for their public criticism of student culture. In contrast, across four studies using varying contexts and research paradigms, Philpot and Hornsey (2008) found no evidence of collective apologies having a positive effect on intergroup forgiveness. Follow-up field research (Philpot & Hornsey, 2011) found that apologies were often misremembered, and that awareness of an apology was either weakly or nonsignificantly related to forgiveness.

Qualitative studies reinforce this mixed picture. In advance of the apology for the Stolen Generation, Indigenous Australians expressed optimism that prosocial outcomes would flow on from the apology, but forgiveness was not a topic that participants engaged with or discussed when thinking of the potential benefits of an apology (Mellor, Bretherton, & Firth, 2007; Philpot, Balvin, Mellor, & Bretherton, 2013). After a 6-year assessment of the South African Truth and Reconciliation Commission, Chapman (2007) noted that victims rarely raised the notion of forgiveness spontaneously, struggled to represent forgiveness in intergroup terms and, when prompted, were reluctant to grant it. Interviews with Chinese Canadians after the apology for the Head Tax imposed on Chinese immigrants in the 20th century suggested that high expectations around an apology are often tinged with disappointment and a sense of skepticism about the motives behind the apology (Blatz, Schumann, & Ross, 2009). Skepticism about motives and meaning of apologies, or conversely attributions of sincerity, would seem to be particularly problematic in politicized contexts in which collective apologies are usually situated.

Sincerity of Apologies

There are various mechanisms through which apologies might increase forgiveness. For example, apologies may return status, power, dignity, and respect to victims (e.g., Okimoto & Wenzel, 2008; Shnabel & Nadler, 2008). Apologies may also lead victims to reappraise the offenders’ moral character and their (re-)commitment to the values they violated (e.g., Steele & Blatz, 2014; Wenzel & Okimoto, 2010, 2014). However, apologies are unlikely to have any of these effects if they are not regarded as sincere. To the extent that an apology is regarded as insincere, the offender is not seen as truly believing in the validity or importance of the values they violated; the apology is meaningless, and a value consensus will not be repaired. Likewise, to the extent that an apology is regarded as insincere, it is seen as lacking the required humility and self-degradation on the part of the offender, and may instead be seen as arrogant and condescending; it will not rebalance status, power, or respect to victims. Apologies failing in these respects may be seen as manipulative attempts at mollifying the victim and avoiding further concessions, symbolic or practical. Correspondingly, insincere apologies are less likely to elicit forgiveness in both intergroup (e.g., Wohl, Hornsey, & Bennett, 2012) and interpersonal contexts (e.g., Schumann, 2012).

A critical question, then, relates to how victims infer whether an apology is sincere or not. Relevant research in interpersonal contexts has focused on three broad features of apologies: words, emotions, and deeds. First, while an apology can be worded minimally by merely expressing remorse and acknowledging responsibility, it can also be more elaborate and include other components, such as explicit self-degradation and recommitment to the violated value. In interpersonal contexts, more elaborate apologies tend to elicit more forgiveness (e.g., Darby & Schlenker, 1982). Furthermore, apologies tend to be perceived as more sincere when the apologizer expresses consistent and appropriate emotions. For example, the expression of guilt and shame has been found to increase, but pity to decrease, attributions of sincerity (Hareli & Eisikovits, 2006). Moreover, apologies that are costly to the offender are more likely to be perceived as sincere (Ohtsubo et al., 2012).

Similar arguments have been made for intergroup contexts. Steele and Blatz (2014) found no effects of apology elaborateness on forgiveness in collective contexts; however, there has been evidence that expression of the right emotions facilitates the acceptance of an apology. For example, an offender’s expression of shame reduces victims’ feelings of insult when being offered compensation (Giner-Sorolla, Castano, Espinosa, & Brown, 2008). Furthermore, when apologies are not followed with action (Blatz et al., 2009; Wohl et al., 2011), when change is seen to be unlikely (Wohl et al., 2015), or where with time it emerges that nothing has changed for victims (Wohl, Matheson, Branscombe, & Anisman, 2013), apologies can be seen as empty and/or manipulative.

Sincerity as Representativeness of an Offender Group

While words, emotions, and deeds are likely to matter in both interpersonal and intergroup contexts for apologies to be perceived as sincere, it is important to recognize distinctive features of intergroup contexts and how these features might change the way in which sincerity is perceived. Intergroup relations are characterized by an ingroup/outgroup categorization which, if salient, tends to imply a more competitive relationship and greater mistrust between the groups (Insko, Schopler, Hoyle, Dardis, & Graetz, 1990). Mistrust would determine whether an apology is seen to be genuine and sincere, or driven by ulterior motives (Hornsey & Wohl, 2013; Nadler & Liviatan, 2006).

Importantly for the present research, the ingroup/outgroup categorization changes the relevant social entity that is held accountable for the wrongdoing; the implied outgroup categorization of perpetrators means that the group as a whole shares collective guilt (Wohl, Branscombe, & Klar, 2006). This shared guilt is, of course, exactly what underpins demands for a collective apology, and the sincerity of the apology should therefore also be an attribution to the group as a whole. Thus, perceived sincerity of a collective apology should be judged not so much by the inner state of the person issuing the apology but rather by the inner state of the group that the apologizer represents. It might therefore be only a secondary consideration whether an individual who delivers a collective apology truly believes in it. What might be more crucial is whether the apology reflects the sentiment of the offender group as a whole. As Tavuchis (1991) argued, an apology “. . . lacking the moral imprimatur of the group, amounts to no apology at all. It means nothing because it represents the unaccredited One and not the mandate of the Many” (p. 101).

As such, critical to the perceived sincerity of an intergroup apology is the extent to which the remorse communicated in the act is deemed to be representative of the broader offender group. One way to help ensure that an apology is seen as representative of the group is to base a collective apology on democratic decision making. If the decision to offer an apology is ostensibly agreed on by a majority of the offender group, its members will be more likely to see the apology as legitimate (Miller, Jackson, Mueller, & Schersching, 1987), and victims witnessing this process may do the same. Even where group members have no formal vote in the decision process, offender group sentiments could be revealed in public opinion surveys, which allow inferences about the credibility of conflict responses by gauging the public opinion of the adversary group. Whether based on formal decision process or voiced public opinion, victim group members may form perceptions of the apology being more or less representative of the offender group’s views.

Subtyping

Although we argue that perceptions of representativeness for the wider offender group are key to attributions of sincerity of a collective apology, it is also important to realize that in reality an offender group may not be an unstructured homogeneous unit. Rather, it may consist of different subgroups (social strata, classes, generations, political opinion groups, etc.) that may be expected to differ on a number of issues, including an apology for wrongdoing toward the victim group. The question is: Do all views within the offender group contribute equally to what is perceived to be the group’s representative position? For example, if a section of the offender group that voices unapologetic sentiments can be identified as a special subgroup (e.g., political extremists), is it possible that their views are discarded as different and not reflective of the broader offender group?

We argue that, depending on the availability of relevant information, victims may engage in implicit forms of subtyping of an offender group to make sense of their discrepant voices. Subtyping processes were originally discussed in research on stereotyping: It describes the process by which counterstereotypical exemplars may be ineffective in changing the overall stereotype about an outgroup because respondents may subtype these as being exceptional types of outgroup members (Richards & Hewstone, 2001). Declared as exceptions, counterstereotypical exemplars are not regarded as reflective of the wider group’s position, and the existing stereotype is maintained.

Equivalent processes may come into effect with attributions of remorse. We have argued that divisions within the offender group about the appropriateness of an apology can undermine the perceived sincerity of that apology. However, if victims perceive the division within the offender group as being predictably along subgroup lines, the dissenting voices that are inconsistent with the official apology could be “understood” and discounted as marginalized views. Thus, subtyping processes may provide a solution toward reducing the negative effects of intragroup dissent among the offender group about an apology.

Overview of the Studies

The present research conceptualizes sincerity of collective apology as a group-level characteristic (rather than a characteristic of the individual speaker). Studies 1 and 2 investigated whether an apology determined by democratic decision making (or not) will affect perceptions of sincerity and victims’ willingness to forgive. Studies 3 and 4 tested whether disagreement with an apology in the wider offender group might be discounted if participants could attribute the disagreement to a distinct subgroup.

Study 1

In this study, a transgression by one group against another was staged in the laboratory. An apology either was or was not offered, and this communication was allegedly either from a randomly chosen individual of the offender group (with low, unclear representativeness) or decided on by majority vote (high representativeness). In line with our arguments in the introduction, a representative apology should be seen as more sincere, and as a consequence elicit greater forgiveness.

Method

Participants and design

Participants were 112 undergraduate students (69 female; M_age = 24.82) who responded to advertisements on campus, and received AU$15 for participation. They were randomly assigned to a 2 (Representativeness: low, high) × 2 (Apology: no, yes) between-subjects design. An a priori power analysis using G*Power, based on an estimated medium effect size (f = 0.30) and statistical power of 0.80, determined a required sample size of 119, of which our sample fell slightly short.

Procedures

The study was allegedly about personality and problem solving. Participants received all instructions and completed all tasks on individual computers in separate cubicles. After answering demographic questions, they completed a personality test, the results of which they were led to believe would determine their allocation to two 3-person teams. Participants were led to believe that they would be interfacing with the other team online; in fact, there were no teams, and the responses of the “other” team were preprogrammed.

The “two” teams then competed against each other in two rounds of a cognitive task. The winning team of each round could decide to take a bonus payment either from the “bank” or from the other team (see Eaton, Struthers, & Santelli, 2006; Wenzel & Okimoto, 2015, for a similar procedure). The transgression occurred when the “other” team won both rounds and, without good reason, decided to take the money from the participant’s team. After the competition, the two teams were instructed to exchange messages with each other.

Experimental manipulations

The representativeness of the message was manipulated by assigning participants to either a democratic or random process. In the democratic process, each supposed member of the participant’s group drafted a message, and then all group members voted on which message to send (high representativeness). In the random process, one group member was allegedly randomly selected to draft and send a message on behalf of the group (low representativeness).

For the participant’s group, in the democratic process condition, the message drafted by the participant appeared on screen with the two messages allegedly drafted by the fellow team members (which were preprogrammed, benign messages such as “Well played guys, thanks for the game”). Group members voted to send the participant’s message to the other group. In the random process, the participant was supposedly randomly selected to send a message on behalf of their group. In either case, the message formulated by the participant was the one sent to the other team.

The participant’s group then received the message sent by the other group, ostensibly based on the same democratic or random process. The content of the message constituted the manipulation of apology. In the no-apology condition, the message was as follows: “Thanks for the game . . . and the money,” whereas in the apology condition the message read, “Sorry about taking all your money guys—it was a bit harsh of us. Hope there’s no hard feelings?” (grammatical errors were deliberate).¹

Dependent variables

The main dependent variable was forgiveness toward the outgroup, measured by 20 items of Philpot and Hornsey’s (2008) Intergroup Forgiveness scale (α = .94). We used 10 items that measure forgiving affect (“Toward the other team I feel” . . . happy, negative, angry, warm, etc.) and 10 items that measure forgiving cognitions (“I think the other team is” . . . worthless, immoral, of good quality, worthy of respect, etc.; 1 = not at all, 7 = very much).

For a measure of sincerity consistent with our theoretical discussion above, tapping truthfulness, humility, and meaningfulness, participants rated how they found the message on five items (1 = not at all, 7 = very much): sincere, arrogant, condescending, manipulative, and meaningless (the latter four were reverse coded; α = .80).

To check the representativeness manipulation, participants were asked in a forced-choice format whether the message their group received was “sent by a single member of the group or did the group collectively decide to send the message?”

Results

The majority of participants (89%) correctly recalled that the message was from an individual member in the low representativeness condition and from the group collectively in the high representativeness condition. However, 12 participants failed to recall this information correctly and were excluded from analyses, leaving a sample of 100 (60 female).² Means and standard deviations for all measures are summarized in Table 1.

Table 1.

Means (and Standard Deviations) for Study 1.

DVs	Representativeness
	Low		High
	No apology	Apology	No apology	Apology
Message sincerity	3.87 (1.23)	3.84 (1.38)	3.90 (1.62)	5.16 (1.57)
Forgiveness	4.83 (1.15)	4.46 (1.33)	4.21 (1.03)	5.17 (1.04)

Note. DV = dependent variable.

Sincerity

Main effects of apology, F(1, 96) = 5.35, p = .023, $η_{p}^{2}$ = .053, and representativeness, F(1, 96) = 4.50, p = .036, $η_{p}^{2}$ = .045, emerged on sincerity. Importantly, these effects were qualified by a significant interaction between representativeness and apology, F(1, 96) = 4.83, p = .030, $η_{p}^{2}$ = .048. An apology that was high in representativeness was rated as more sincere than an apology low in representativeness, F(1, 96) = 9.12, p = .003, $η_{p}^{2}$ = .087. In contrast, representativeness had no effect in the nonapology condition, F(1, 96) = 0.003, p = .957, $η_{p}^{2}$ < .001.

Forgiveness

Only a significant interaction emerged on forgiveness, F(1, 96) = 8.35, p = .005, $η_{p}^{2}$ = .080. In line with hypotheses, an apology that was high in representativeness led to greater levels of forgiveness than an apology low in representativeness, F(1, 96) = 8.49, p = .004, $η_{p}^{2}$ = .081. For a nonapologetic message, representativeness did not have a significant effect, F(1, 96) = 1.33, p = .252, $η_{p}^{2}$ = .014.

Moderated mediation

We tested for moderated mediation using the bootstrapping PROCESS approach by Hayes (2013; Model 7). A significant Index of Moderation showed that the indirect effect of apology on forgiveness via sincerity was significantly moderated by representativeness, b = .29, 95% confidence interval (95% CI) = [0.04, 0.62]. The conditional indirect effect of an apology on forgiveness mediated by sincerity was significant when representativeness was high, b = .30, 95% CI = [0.13, 0.50], but not when representativeness was low, b = .01, 95% CI = [−0.21, 0.19].

Discussion

Study 1 suggests that the sincerity of an offender group’s response is a key determinant of intergroup forgiveness. Moreover, in intergroup contexts the sincerity of an apology is dependent on the decision-making process within the offender group that led to their response. We assumed that an apology would be more representative of the group’s will when it was the outcome of a democratic vote than when offered by a randomly chosen individual member speaking on behalf of the group. As a consequence, an apology based on a democratic process would be considered more sincere, and therefore would be more likely to lead to forgiveness. The present data support this view.

In Study 1, we felt that it was important that participants experienced firsthand the decision-making process; they themselves had previously engaged in a democratic decision-making process when sending their own group’s message, so the decision process leading to the outgroup message was clear. However, it is technically possible that the absence or presence of a democratic process within the ingroup could have been responsible for the effects on sincerity and forgiveness. Also, in the random process condition, the degree of representativeness was unclear but not necessarily low. We therefore conducted a second study in which the democratic process was kept constant between conditions, but the representativeness of its outcome within the outgroup was varied.

Study 2

Method

Participants and design

Based on the observed effect size in Study 1 for the interaction effect (f = 0.225) and a desired power of 0.80, we determined a required sample of 160. The recruited sample comprised 161 undergraduate students (113 female; M_age = 22.98). Participants volunteered to participate in response to advertisements placed across the university campus, and received payment of AU$15. They were randomly assigned to a 2 (Representativeness: low, high) × 2 (Apology: no, yes) between-subjects design.

Procedure

The study followed the same general procedure as Study 1, except that participants in all conditions were told that group members were to draft messages to send to the opposing team, and that groups were to vote on which message to send. The outcome of the outgroup’s vote was varied to manipulate representativeness.

Experimental manipulations

Participants were told that if a group could not reach majority consensus on any of the messages drafted by its members, one message would be randomly selected and forwarded to the other team. The participants’ ingroup allegedly always reached a consensus. In the low representativeness condition, the transgressor outgroup did not appear to have a majority view on which message to send (implying that all three messages received one vote each), and so a message was chosen randomly. In the high representativeness condition, the outgroup message was allegedly endorsed by a majority. The nonapologetic and apologetic messages were the same as in Study 1.

Dependent variables

Measures for sincerity (α = .73) and intergroup forgiveness (α = .91) were the same as in Study 1. To test for comprehension of the representativeness procedure, a question at the end of the study checked whether participants correctly recalled whether the outgroup’s message was agreed on by the majority or randomly chosen. In addition, four items measured the perceived representativeness of the outgroup’s message for its members’ views (e.g., “How much do you think the members of the other team share the views expressed in this message?” 1 = not at all, 7 = very much, α = .80).

Results and Discussion

Participants who did not correctly identify whether the message received from the transgressor group was randomly selected or endorsed by the majority (n = 32) were excluded from analyses, leaving a sample of 129 (92 female, M_age = 22.52).³ Means and standard deviations are summarized in Table 2.

Table 2.

Means (and Standard Deviations) for Study 2.

DVs	Representativeness
	Low		High
	No apology	Apology	No apology	Apology
Message sincerity	4.24 (1.17)	4.37 (1.10)	3.98 (1.25)	4.98 (1.32)
Forgiveness	4.99 (1.03)	4.49 (0.79)	4.47 (1.04)	4.91 (0.72)

Note. DV = dependent variable.

Manipulation check of representativeness

As expected, 2 × 2 ANOVA showed that participants in the nonrepresentative condition perceived the message from the outgroup to be significantly less representative of the views of all outgroup members (M = 3.97) than those in the representative condition (M = 4.42), F(1, 125) = 5.83, p = .017, $η_{p}^{2}$ = .045. The effect of apology and the interaction were not significant, Fs < 1.28, ps > .260.

Sincerity

A main effect of apology on sincerity, F(1, 125) = 6.87, p = .010, $η_{p}^{2}$ = .052, was qualified by a significant interaction with representativeness, F(1, 125) = 4.04, p = .047, $η_{p}^{2}$ = .031. An apology high in representativeness was rated as more sincere than an apology low in representativeness, F(1, 125) = 3.91, p = .050, $η_{p}^{2}$ = .030, whereas for a nonapologetic message representativeness did not have a significant effect, F(1, 125) = 0.74, p = .391, $η_{p}^{2}$ = .006.

Forgiveness

As predicted, a significant interaction emerged between representativeness and apology on forgiveness, F(1, 125) = 8.63, p = .004, $η_{p}^{2}$ = .065. An apology high in representativeness led to marginally more intergroup forgiveness than an apology low in representativeness, F(1, 125) = 3.40, p = .068, $η_{p}^{2}$ = .026. In contrast, when receiving a nonapologetic message, forgiveness was lower in the high versus low representativeness condition, F(1, 125) = 5.36, p = .022, $η_{p}^{2}$ = .041.

Moderated mediation

We tested for moderated mediation using the PROCESS approach by Hayes (2013; Model 7). The Index of Moderation was significant, showing that the indirect effect of apology on forgiveness via sincerity was significantly moderated by representativeness, b = .19, 95% CI = [0.02, 0.41]. The conditional indirect effect of an apology on forgiveness mediated by sincerity was significant when representativeness was high, b = .22, 95% CI = [0.08, 0.37], but not when representativeness was low, b = .03, 95% CI = [−0.10, 0.14].

In sum, representativeness of an apology increased perceptions of sincerity and, mediated through this, increased victim forgiveness toward the transgressor group. This finding was consistent with Study 1.

Study 3

In real-life situations, it would be unusual for the membership of a group to be directly involved in the decision making on a collective apology, as was the case in Studies 1 and 2. Indeed, we are unaware of any historical precedent where a referendum was held among the citizens of a country to decide on an apology toward a victim group. Instead of direct involvement in the decision making, it is often the case that the wider group membership engages in debate and voices their opinion about the issue. The tone of public debate (and data from opinion surveys) should provide an indication of the representativeness of an official statement for the wider group’s sentiment. This is the situation we utilized for Studies 3 and 4; we referred to a real political incident and provided participants with alleged survey results that indicated the level of public opinion for or against an official apology.

The incident we referred to concerned the abuse of Australian prisoners of war by the Japanese in World War II (WWII). In the course of the war, the Japanese captured about 22,000 Australians, of whom more than a third died. There have been apologies by Japanese officials on various occasions for abuses during the war, but only a small minority of the Australian public seem to be aware of such apologies (Philpot & Hornsey, 2011).

We presented participants with a fabricated online news article reporting on the official apology by Japan’s Foreign Minister. As part of the article, we provided information (verbal, with a supporting graph) about the level of support among the wider Japanese public for such an apology. This alleged public opinion survey showed either overwhelming support for an apology, or only a moderate majority supporting it (and a sizable minority objecting to it). Crossed with this factor, we provided additional information about the opinions of two subgroups: younger and older generation Japanese. In the condition with sizable dissent, younger generation Japanese strongly supported an apology, while older generation Japanese strongly objected to it. However, in the condition citing broad support, this subgroup distinction was only nominal, with both younger and older generation Japanese showing high levels of support. We reasoned that, given the historical nature of the wrongdoing, generational subgroups might be relevant and meaningful for an appraisal of the division with the offender group. If the public opinion survey implicitly “explained” the division of apology views within the offender group as being due to an identifiable, “subtypable” subsection of Japanese, the division would be discounted and the apology would be seen as being more typical of offender group views (and thus as more sincere).

Method

Participants and design

Based on an anticipated effect size of f = 0.225 and a desired power of 0.80, we determined a required sample of 160. We recruited 175 Australian citizens through the online survey platform Qualtrics. We excluded two participants who spent an excessive amount of time reading the news article presented to them (>10 min, suggesting that they might have cross-checked the information provided). This left 173 participants (43.4% female; M_age = 49). Participants were randomly assigned to the conditions of a 2 (Overall support: moderate, high) × 2 (Subgroup information: no, yes) between-groups design.

Procedure

After providing demographic details, participants were presented with a fake online news article titled “Japan Apologizes to Australian Prisoners of War.” The article reported about a formal apology to Australia issued by the Japanese Foreign Minister for war crimes committed against Australian soldiers and nurses: “Japan inflicted tremendous damage and suffering on people in many countries. We would like to apologize to Australia in particular, for our treatment of Prisoners of War.”

Experimental manipulations

To manipulate overall support for an apology in the offender group, the article either stated that there was “widespread support throughout Japan for this apology . . .” and an opinion survey “showed 88% of Japanese respondents supported an apology to Australia” (high support), or “There has been division in Japan regarding an apology to Australians . . .” and an opinion survey “showed 55% of Japanese respondents supported an apology to Australia” (moderate support). In the no-subgroup information conditions, this was all that was shown to participants. But in the subgroup information condition, a breakdown of survey results for younger and older generation Japanese was provided. In the high support condition, the article reported “similar levels of support among the young and old,” and a graph showed approximately 90% support among the young and 86% among the old. In the moderate support condition, the article reported “88% of younger people but only 12% of older people supporting the apology.”

Dependent variables

The typicality of supporters and nonsupporters was measured by two items (How TYPICAL of the Japanese population were those who SUPPORTED [DID NOT SUPPORT] the apology? 1 = not at all typical, 7 = very typical; r = −.51), and subtracting the nonsupporter typicality from supporter typicality. For sincerity, we adopted a four-item measure from Wohl et al. (2012) that measures participants’ trust in the genuineness of the emotions expressed in their apology (e.g., “I believe that the Japanese really are ashamed and humiliated by their role in these events”; 1 = not at all, 7 = very much; α = .92).⁴ For intergroup forgiveness, we again used Philpot and Hornsey (2008) scale, but this time included all 30 items (α = .96).

To check the effectiveness of the manipulations, one item assessed the extent of division over the apology in the offender group (“How much division is there among the Japanese people about the Japanese government’s apology to Australia?”), and another item assessed perceived differences in opinion between younger and older Japanese about the apology (“How much division is there among the older and younger generations of Japanese people about the Japanese government’s apology to Australia?” 1 = very little, 7 = very much).

Results

Means and standard deviations are summarized in Table 3.

Table 3.

Means (and Standard Deviations) for Study 3.

DVs	Overall support
	Moderate		High
	Subgroup info		Subgroup info
	No	Yes	No	Yes
Overall division	4.20 (1.46)	4.20 (1.39)	3.20 (1.59)	3.61 (1.63)
Division young–old	4.62 (1.30)	5.25 (1.73)	3.78 (1.67)	3.52 (1.77)
Relative typicality	−0.27 (1.97)	1.20 (2.45)	1.10 (2.86)	0.86 (3.00)
Sincerity	3.93 (1.53)	4.92 (1.48)	4.94 (1.35)	5.06 (1.58)
Forgiveness	4.14 (1.26)	4.20 (1.34)	4.30 (1.25)	4.21 (1.33)

Note. DV = dependent variable.

Manipulation checks

Consistent with manipulations, in the high support condition (M = 3.41) participants perceived less division in public support than in the moderate support condition (M = 4.20); F(1, 169) = 11.86, p < .001, $η_{p}^{2}$ = .067. The effect of subgroup information and the interaction were not significant, Fs < 0.82.

For the measure of division between younger and older generation Japanese, a main effect of support, F(1, 169) = 27.05, p < .001, $η_{p}^{2}$ = .138, was qualified by the expected interaction with subgroup information, F(1, 169) = 3.16, p = .077, $η_{p}^{2}$ = .018, albeit only marginally. In the high support conditions, levels of perceived division between young and old did not differ between subgroup information conditions, F(1, 169) = 0.50, p = .479, $η_{p}^{2}$ = .003. However, when support was moderate, participants indicated a marginally higher level of young–old division when the subgroup information was provided than when it was not, F(1, 169) = 3.32, p = .070, $η_{p}^{2}$ = .019. Overall, the manipulations were successful.

Typicality

A significant interaction between support and subgroup information emerged on typicality, F(1, 169) = 4.69, p = .032, $η_{p}^{2}$ = .027. In the absence of subgroup information, supporters of the apology were considered more typical of the Japanese population when overall support was high rather than low, F(1, 169) = 7.17, p = .007, $η_{p}^{2}$ = .041. However, when subgroup information was provided, the perceived typicality of supporters did not differ as a function of support, F(1, 169) = 0.17, p = .677, $η_{p}^{2}$ = .001. These results are in line with the prediction that dissenting views can be subtyped when disagreement with an apology is attributable to a particular subgroup, resulting in a belief that supportive views are more typical of the offender group.

Sincerity

Main effects of support, F(1, 169) = 6.44, p = .012, $η_{p}^{2}$ = .037, and subgroup information, F(1, 169) = 5.93, p = .016, $η_{p}^{2}$ = .034, were qualified by a (marginally) significant interaction, F(1, 169) = 3.66, p = .058, $η_{p}^{2}$ = .021. When no-subgroup information was provided, the apology was considered more sincere when public support was high than moderate, F(1, 169) = 9.71, p = .002, $η_{p}^{2}$ = .054. However, when subgroup information was presented that allowed the attribution of the moderate support to a particular dissenting subgroup, support had no effect, F(1, 169) = 0.20, p = .656, $η_{p}^{2}$ = .001. These results closely follow the pattern identified for typicality attributions above. Moderated mediation analysis with bootstrapping methods (Hayes, 2013; Model 7) confirmed that high support increased (dissent decreased) perceived sincerity indirectly via perceived typicality of apology supporters in the absence of subgroup information, b = .25, 95% CI = [0.05, 0.46], but not when subgroup information was provided, b = −.06, 95% CI = [−0.29, 14].

Forgiveness

Contrary to predictions, levels of forgiveness did not differ significantly between conditions, all Fs < 0.17. Despite the lack of a total effect of the manipulations on forgiveness, perceived sincerity was nonetheless a strong predictor of forgiveness, b = .48, p < .001, and the indirect effect of the interaction between support and subgroup information via sincerity on forgiveness was significant, b = −.21, 95% CI = [−0.46, −0.01]. High support increased (dissent decreased) forgiveness indirectly via perceived sincerity in the absence of subgroup information, b = .24, 95% CI = [0.11, 0.44], but not when the subgroup information was provided, b = .03, 95% CI = [−0.13, 19].

Discussion

The findings show once again that dissent within the offender group about the appropriateness of an apology can reduce its perceived sincerity, which is in turn a strong determinant of victim group members’ forgiving sentiments. Furthermore, Study 3 shows that when the dissent is due to an identifiable subgroup, its negative impact is reduced or neutralized. The subtyping of dissenters allows the attribution of greater typicality to supporters of the apology, which implies that the apology is more representative of the wider offender group, and thus more sincere in a group-level sense.

The subgrouping of younger and older generation Japanese in the present study was an intuitive choice, as it seemed to fit the context of a historical wrongdoing. Indeed, it can be argued that not just any subgroup information will allow relevant subtyping processes to occur; instead, the subgrouping has to fit (comparatively and normatively; see Oakes, 1987) the given context. It may not be sufficient that the subgrouping coincides with greater and lesser support for an apology (comparative fit). Rather, the subgroups might be expected to disagree on the issue; the dissenting views can be expected and thus explained (and potentially explained away) because of their subgroup membership (normative fit). For war crimes and abuses of prisoners of war, it might not be unexpected that older generation Japanese would be more reluctant to accept guilt and less willing to move forward toward reconciliation than younger generation Japanese (whether or not that is actually true). If so, the dissenting views can be subtyped as the views of the old and as less representative of the wider offender group. The next study investigated this qualification.

Study 4

As in Study 3, Study 4 referred to Japanese abuses against Australian prisoners of war in WWII and provided participants with a news article about an official apology by Japan’s Foreign Minister. However, we kept the level of overall support at a moderate level and provided subgroup information in all conditions. In one condition, we reported that the dissent was largely due to older generation Japanese, as in Study 3; we expected this to be a “fitting” subgrouping as explained above. This was compared with conditions where the dissent was largely due to younger generation Japanese holding apology-opposing views (counternormative subgrouping), or where the Japanese from the North or the South largely held apology-opposing views (irrelevant subgrouping). We predicted that the apology would be seen as more representative, and as a consequence more sincere, when fitting subgroup information was presented rather than counternormative or irrelevant information.

We also predicted this process to lead to greater levels of forgiveness in the fitting condition than in the other conditions. However, the previous study only provided evidence for an indirect effect of the subgroup information, with no overall effect on forgiveness. Thus, in the present study we switched to a forgiveness measure that more directly measures individuals’ feelings of forgiveness toward another group.

Moreover, it is possible that the fitting subgroup information has an additional, counteracting effect in the present context. In Study 3, the fitting subgrouping involved older generation Japanese not supporting an apology. However, victims may regard the older generation Japanese as more closely implicated in the wrongdoing, and it should be them in particular (rather than the younger ones) who have the obligation and “standing” to apologize. “Standing”—originally a legal term—is “a procedural requirement ensuring that only legitimate disputants adjudicate claims” (Smith, 2008, p. 52). In Smith’s understanding of a “categorical apology,” this means that only the party causally responsible for wrongdoing can truly apologize. In the fitting subgroup condition, the younger generation Japanese carry the apology; in this case, victim group members may perceive a lack of standing, which may undermine the perceived sincerity attributed to the apology and/or their forgiveness response. In the present study, we included a measure of lack of standing to investigate this potential counteracting process, both as a sequential mediation via sincerity and directly on forgiveness.

Finally, in Study 3 we had used a sincerity measure that was different from the scale used in Studies 1 and 2. In the present study, we included both measures to demonstrate their equivalence.

Method

Participants, design, and procedure

Based again on an effect size of f = 0.225 and a power of 0.80, we determined a desired sample size of 195. The sample comprised 200 Australian citizens above 18 years old recruited through the online survey platform Qualtrics; 16 participants failed a procedural reading check and were excluded from the analysis. We also excluded four participants who spent an excessive amount of time reading the news article presented to them (>10 min). This left 180 participants (48% female; M_age = 54). Participants were randomly assigned to three experimental conditions of a between-subjects design: fitting, nonfitting, and irrelevant subgrouping. The procedure was the same as in Study 3.

Experimental manipulations

To manipulate different subgroup support in the offender group for an apology, the article stated, “There has been division in Japan regarding an apology to Australians . . .,” and an opinion survey “showed 55% of Japanese respondents supported an apology to Australia.” In the fitting subgroup information condition, it was stated that the dissent was mainly due to older Japanese people (“88% of younger people but only 12% of older people [supported] the apology”), in the nonfitting subgroup condition the dissent was mainly due to younger Japanese, and in the irrelevant subgroup condition it was mainly due to people from South Japan or (counterbalanced) North Japan. A bar graph was used to illustrate the unequal levels of support.

Dependent variables

Perceived representativeness of the apology was measured by a single item (“I sense that the official apology of the Japanese government represents the views of the Japanese people”). For sincerity, we used the same four-item measure as in Study 3 (α = .89); as an alternative, we also included the five-item scale we used in Studies 1 and 2 (α = .87). However, both scales were highly correlated (r = .70), and the results for both were virtually the same. For economy of presentation, we therefore report here the findings for a combined nine-item measure of sincerity (α = .91). For intergroup forgiveness, we used a four-item measure adopted from Wohl et al. (2012): for example, “I forgive Japan for the harm done to Australian Prisoners of War” (1 = strongly disagree, 7 = strongly agree; α = .92).

Perceived lack of standing was measured by a single item, in nontechnical terms that would be accessible to participants: “The official apology is meaningless because those who should apologize are least willing to support it” (1 = not at all, 7 = very much). Furthermore, we included a number of manipulation checks toward the end of the study. First, participants indicated how much division there was among the older and younger generations of Japanese people about the Japanese government’s apology to Australia (1 = very little, 7 = very much). Second, two measures checked for the assumed normative fit of the subgroupings. Participants were asked to drag a slider to the left or right to the extent that they expected younger or older Japanese people to be more supportive of an apology (−3 = younger more supportive, 3 = older more supportive), and another slider for whether they expected northern or southern Japanese to be more supportive (−3 = northern more supportive, 3 = southern more supportive).

Results

Manipulation checks

A one-factorial ANOVA showed that experimental conditions differed significantly in the extent of division perceived between younger and older Japanese people, F(2, 177) = 6.54, p = .002, $η_{p}^{2}$ = .069. As intended, participants perceived greater divisions between young and old in the fitting and nonfitting subgroup conditions compared with the irrelevant subgroup condition (see Table 4). Second, participants expected younger Japanese to be more supportive of an apology than older Japanese, as shown by a t test against the midpoint of the scale, t(179) = −6.88, p < .001. However, ANOVA showed this to be significantly less the case in the nonfitting condition compared with the fitting and irrelevant conditions, F(2, 177) = 6.66, p = .002, $η_{p}^{2}$ = .070. On the contrary, participants did not expect northern and southern Japanese to differ in their support for an apology overall, t(179) = .26, p < .795, and there were no significant differences in this perception between conditions, F(2, 177) = 0.06, p = .946, $η_{p}^{2}$ = .001. Consistent with the intent of the manipulations, participants expected younger Japanese to be more supportive of an apology; thus, the fitting subgroup condition was indeed more normatively fitting.

Table 4.

Means (and Standard Deviations) for Control Variables in Study 4.

DVs	Subgrouping
DVs	Fitting	Nonfitting	Irrelevant
Young–old division	5.37 (1.47)	5.15 (1.42)	4.46 (1.44)
Support for apology by the young (vs. old)^a	−1.29 (1.62)	−0.25 (1.69)	−1.02 (1.53)
Support for apology in the North (vs. South)^b	0.03 (0.76)	0.04 (0.63)	−0.01 (1.32)
Lack of standing	4.58 (1.92)	3.40 (1.85)	3.72 (1.72)

Note. DV = dependent variable.

−3 = younger generation more supportive, 3 = older generation more supportive.

−3 = northern Japanese more supportive, 3 = southern Japanese more supportive.

Lack of standing

ANOVA showed a significant effect on perceived lack of standing, F(2, 177) = 6.56, p = .002, $η_{p}^{2}$ = .069. Our suspicions were supported: In the fitting subgrouping condition, where the apology was supported by younger Japanese rather than the older generation, participants perceived a significantly greater lack of standing than in the nonfitting and irrelevant conditions (see Table 4). Further analysis showed that this lack of standing was significantly related to intergroup forgiveness, B = −.46, p < .001, and we therefore controlled statistically for this variable in the following analyses.

Sincerity

An ANCOVA for perceived sincerity yielded a significant effect of the covariate, F(1, 176) = 130.78, p < .001, B = −.43; the less perceived standing, the less sincere the apology. There was also a significant effect of experimental conditions, F(2, 176) = 5.25, p = .006, $η_{p}^{2}$ = .056. In the fitting subgroup condition, the apology was perceived to be significantly more sincere than in the nonfitting and irrelevant subgroup conditions; the latter two conditions did not differ significantly from another (see Figure 1). Note, however, that the effect of the fitting subgrouping on sincerity would not have been significant without controlling for lack of standing. Illustrated in a regression framework with nonfitting and irrelevant conditions combined, the fitting subgroup had no significant total effect on sincerity, B = .02, 95% CI = [−0.16, 0.21]; only when accounting for an indirect effect via lack of standing, B = −.22, 95% CI = [−0.35, −0.08], was the direct effect of fitting subgrouping on sincerity significant, B = .24, 95% CI = [0.09, 0.39].

Figure 1.

Estimated means (with standard error bars) for apology sincerity and intergroup forgiveness, controlling for perceived lack of standing (Study 4).

Forgiveness

An equivalent ANCOVA for forgiveness yielded a significant effect of the covariate, F(1, 176) = 74.34, p < .001, B = −.46; the less perceived standing, the less forgiving were the participants. Furthermore, there was a significant effect of experimental condition, F(2, 176) = 3.18, p = .044, $η_{p}^{2}$ = .035; participants in the fitting subgroup condition were significantly more forgiving than in the nonfitting condition and irrelevant conditions (the latter two did not differ significantly from one another; see Figure 1). Again, as illustrated through regression techniques, the total effect on forgiveness of fitting subgrouping (vis-à-vis nonfitting and irrelevant subgrouping combined) was not significant, B = .02, 95% CI = [−0.22, 0.26]; only when accounting for an indirect effect via lack of standing, B = −.23, 95% CI = [−0.39, −0.10], was the effect of fitting subgrouping on forgiveness significant, B = .25, 95% CI = [0.04, 0.46].

Furthermore, mediation analysis using PROCESS (Model 6; Hayes, 2013) showed that this effect on forgiveness was significantly mediated by perceived sincerity; indirect effect B = .15, 95% CI = [0.06, 0.27]. In addition, there was a (counteracting) significant mediational sequence from fitting subgroup information to lack of standing, to reduced perceived sincerity and, in turn, reduced forgiveness, B = −.13, 95% CI = [−0.26, −0.06]. There was also a mediational sequence from fitting subgroup information to lack of standing directly to reduced forgiveness, B = −.10, 95% CI = [−0.21, −0.03]. Figure 2 shows the complete mediation relationships.

Figure 2.

Indirect effects of fitting subgroup information on forgiveness (against nonfitting and irrelevant subgroup conditions combined; Study 4).

Discussion

The present results support the argument that dissent in the offender group about the appropriateness of an apology can lead victims to attribute less sincerity to the apology and respond with less forgiveness. In conjunction with the findings from Study 3, however, they confirm an exception to the rule: Dissent will not necessarily decrease sincerity and forgiveness if it can be attributed to an offender subgroup whose disagreement with the apology is normatively expected. In this case, the dissenters may be subtyped as a particular subgroup of offenders, not typical of the offender group as a whole. Hence, the apology tends to be seen as more representative and sincere, and victims respond with greater forgiveness.

Although this would suggest a solution to the problem of offender group division diminishing the effectiveness of apologies, there is a significant caveat. A fitting subgrouping that would allow such subtyping to occur is one where dissenters to an apology are more or less expected to disagree with an apology. Such disagreement is likely to be more expected for a subgroup that is more implicated in the wrongdoing and for whom acceptance of guilt would have more serious identity implications. However, it is this subgroup that victims are particularly keen to see admit guilt and show remorse. Subtyping such dissenters would therefore not make the apology more sincere but rather less so, because those who victims think would have particular cause to be apologetic and who have standing to apologize are not part of the offender decision to apologize. In the present study, this process counteracted the beneficial effect of subgroup information, with no significant net effect on sincerity and forgiveness.

General Discussion

The present research provided evidence that victim group members find collective apologies less sincere—and they are less forgiving as a consequence—to the extent that these are not reflective of the wider offender group’s sentiment. A collective apology was perceived to be less sincere when it was the decision of an individual member rather than the group as a whole (Study 1), a group decision not carried with majority support (Study 2), or inconsistent with public opinion within the offender group (Studies 3 and 4). The findings support our argument that, in relation to collective apologies, sincerity has a distinctive meaning: An apology’s sincerity—its truthfulness, meaningfulness, and humility—is a function of it representing apologetic sentiments held within that group.

The Challenges of Collective Apologies

With such a group-level understanding of sincerity, one is better positioned to appreciate the challenges that collective apologies face compared with their individual-level counterpart. As discussed earlier, the evidence for the effectiveness of collective apologies is rather mixed and points to limits in their capacity to elicit forgiving responses in victims (Hornsey & Wohl, 2013; Wohl et al., 2011). The concept of sincerity advanced by the present research offers an understanding as to why this might be the case, and thus why collective apologies often fail to elicit forgiveness in the empirical literature. Specifically, the sincerity of collective apologies is a group-level attribution.

The current findings illustrate that victim group members tend to attribute greater sincerity to an apology when it is based on a majority decision, or when it is consistent with a survey of sentiments within the wider offender group. However, such features are often not given in reality; often, apologies are decisions made by politicians and diplomats on behalf of a group, and often there is no information available as to what the wider offender group thinks or feels about it. Even if such information were available, it would have to compete with what could be rather entrenched prejudicial views about the offender outgroup, and with a categorical differentiation that fosters perceived negative interdependence, competition, and mistrust (Insko et al., 1990). Representativeness is ultimately an attribution by victims. As such, it is likely to be affected by psychological processes of stereotyping and perceptions of prototypicality, which are known to be influenced by categorization processes, the comparative and normative fit of information, salience, accessibility, and the perceivers’ motivations and identity concerns (Fiske, 1998; Oakes, Haslam, & Turner, 1998). Even an unapologetic minority within the offender group could thus be seen by victims as representing the group’s true face. The objective numbers of a majority vote may not matter if those unapologetic voices are more salient and dominate the debate, are regarded as leaders, or are considered to be more prototypical. In the context of a history of conflict and wrongdoing, the perceived prototype may indeed be contrasted away from conciliation, tending to make the unapologetic position a more prototypical one (Oakes et al., 1998). Attributions of representativeness are vastly more complex than simply being determined by a count of votes or voices.

Subtyping and Perceived Intragroup Dynamics

The qualification that victims give more weight to some voices than others can also be used to render an apology more effective. One such approach was investigated in Studies 3 and 4: When dissenters to an apology are identified as mainly belonging to a particular subgroup (in particular, one that might be more or less expected to oppose an apology), victims may fence them off from the main offender group that decided to offer an apology (Richards & Hewstone, 2001). Our findings suggest that through this process of subtyping dissenters, victims can regard the apology as relatively representative of the offender group and more sincere as a result.

While our studies provided supportive evidence for this process, they also pointed to some caveats. Specifically, the subtyping process led only indirectly to greater levels of forgiveness; overall, the total effect of subgroup information, which should theoretically elicit subtyping, did not increase forgiveness. The results suggested a counteracting effect: Knowing that the subgroup who were most implicated in the historical wrongdoing oppose the apology creates the opportunity for subtyping, but it also implies that the offender subgroup endorsing the apology have reduced standing. They may be seen as apologizing for others’ actions, for those more directly associated with the wrongdoing, but as lacking legitimacy to do so. Lack of standing robs the apology of some of its meaning, which canceled out the positive effect of subtyping in Study 4.

Furthermore, one could argue that there is an alternative or complementary interpretation of effects of subgrouping that distinguishes between younger and older generations of a group. Younger generation Japanese may be regarded not only as more representative but also as more influential as far as the future actions and direction of Japan are concerned. Hence, an apology supported by younger generation Japanese may be given greater weight because it is more predictive of Japan’s future conduct. However, in our view these two explanations in terms of representativeness versus influence are closely intertwined. First, those who are considered more representative (or prototypical) of a group tend to be more influential (Turner, 1991), and this is indeed part of the presumed process. To the extent that the “unapologetic” older generation Japanese are subtyped as atypical, the younger generation is more prototypical and influential, and the apology is seen as a more genuine reflection of Japanese sentiments and action tendencies. Second, what is considered prototypical or representative of a group may not only be based on the perceived static pattern of current positions held within a group but also the dynamic trajectory of where the group is heading; that is, it signifies not only what the group is but also what the group is to become (see Reicher, 1996). Representativeness may thus reflect current or impending influences within the group.

Interestingly, in situations where the salient subgrouping is between older and younger generations, the two counteracting effects of lack of standing versus representativeness could also be framed as two different perspectives on apologies, namely backward-looking and forward-looking. A backward-looking perspective focuses on how apologies are most effective in addressing and making amends for the wrong of the past; this would likely be an apology carried by those more closely implicated in the wrongdoing (who have standing). A forward-looking perspective would focus on the promise of an apology for the future, improved behavior, and reconciliation; such promise would be more likely if the apology was carried by those representing, or with influence on, the future.

Nevertheless, the findings show that victim group members do not necessarily see an offender group as an undifferentiated whole, and that they may be attuned to differences within the offender group. Not only do they understand that there may be different factions within an offender group that may predictably diverge in their conciliatory stance toward the victim group, but they also appreciate the form of decision making that has led an offender group to apologize. Indeed, this is the more fundamental lesson of the present research: Group members form impression of other groups (and their actions) based on their perceived or suspected intragroup composition and dynamics.

While the present research makes it clear that sincerity of collective apologies needs to be understood as a group-level phenomenon, it does not automatically follow that individual attributes of the speaker conveying the apology (e.g., the speaker’s demeanor, emotionality, or inner conviction) are unimportant for collective apologies. The processes are more complex than just counting votes or voices as if they were all equally reflective of the group as a whole; the views and actions of an outgroup leader are likely to be given greater weight than those of other outgroup individuals in forming impressions of the group’s true sentiment. Future research will need to investigate to what extent individual-level factors matter for victims’ reception of collective apologies and, if they do, how they combine with group-level factors to determine attributions of sincerity.

More generally, it is important to recognize that domestic politics and international relations are often entangled (Putnam, 1988). Collective apologies are communications to the offender group as much as to the victim group. Politicians may pursue (or resist) collective apologies for domestic political point scoring to placate certain sections of the electorate, or to gain license for the implementation of less popular political agendas. A multilevel approach to collective apologies is necessary that considers the interactions between intra- and intergroup dynamics.

Conclusion

In an age where they are increasingly expected yet decreasingly appreciated (Okimoto et al., 2015), collective apologies face an uphill battle to convince audiences of their sincerity. Well-chosen words, heartfelt emotions, and consequent deeds may only go so far if they are not seen to represent what the wider offender group thinks, feels, or does. Of course, leaving the emergence of apologetic sentiments to the whims of public opinion may do little to advance a process of conciliation. True leaders must endeavor to shape the group’s views and engage democratic participation in policy formation, to make sure that an apology represents the group’s will and sentiment.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a grant from the Australian Research Council, DP130101598.

Supplemental Material

The online supplemental material is available at .

Notes

References

Blatz

C. W.

Schumann

Ross

(2009). Government apologies for historical injustices. Political Psychology, 30, 219-241.

Brooks

R. L.

(1999). When sorry isn’t enough: The controversy over apologies and reparations for human injustice. New York: New York University Press.

Brown

R. P.

Wohl

M. J. A.

Exline

J. J.

(2008). Taking up offenses: Secondhand forgiveness and group identification. Personality and Social Psychology Bulletin, 34, 1406-1419.

Chapman

A. R.

(2007). Truth commissions and intergroup forgiveness: The case of the South African Truth and Reconciliation Commission. Peace and Conflict: Journal of Peace Psychology, 13, 51-69.

Darby

B. W.

Schlenker

B. R.

(1982). Children’s reactions to apologies. Journal of Personality and Social Psychology, 43, 742-753.

Eaton

Struthers

Santelli

A. G.

(2006). The mediating role of perceptual validation in the repentance-forgiveness process. Personality and Social Psychology Bulletin, 32, 1389-1401.

Fiske

S. T.

(1998). Stereotyping, prejudice, and discrimination. In Gilbert

D. T.

Fiske

S. T.

Lindzey

(Eds.), The handbook of social psychology (4th ed., Vol. 2, pp. 357-411). New York, NY: McGraw-Hill.

Giner-Sorolla

Castano

Espinosa

Brown

(2008). Shame expressions reduce the recipient’s insult from outgroup reparations. Journal of Experimental Social Psychology, 44, 519-526.

Hareli

Eisikovits

(2006). The role of communicating social emotions accompanying apologies in forgiveness. Motivation and Emotion, 30, 189-197.

10.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

11.

Hornsey

M. J.

Wohl

M. J. A.

(2013). We are sorry: Intergroup apologies and their tenuous link with intergroup forgiveness. European Review of Social Psychology, 24, 1-31.

12.

Insko

C. A.

Schopler

Hoyle

R. H.

Dardis

G. J.

Graetz

K. A.

(1990). Individual-group discontinuity as a function of fear and greed. Journal of Personality and Social Psychology, 58, 68-79.

13.

Leonard

D. J.

Mackie

D. M.

Smith

E. R.

(2011). Emotional responses to intergroup apology mediate intergroup forgiveness and retribution. Journal of Experimental Social Psychology, 47, 1198-1206.

14.

Mellor

Bretherton

Firth

(2007). Aboriginal and non-aboriginal Australia: The dilemma of apologies, forgiveness, and reconciliation. Peace and Conflict: Journal of Peace Psychology, 13, 11-36.

15.

Miller

C. E.

Jackson

Mueller

Schersching

(1987). Some social psychological effects of group decision rules. Journal of Personality and Social Psychology, 52, 325-332.

16.

Nadler

Liviatan

(2006). Intergroup reconciliation: Effects of adversary’s expressions of empathy, responsibility, and recipients’ trust. Personality and Social Psychology Bulletin, 32, 459-470.

17.

Oakes

P. J.

(1987). The salience of social categories. In Turner

J. C.

Hogg

M. A.

Oakes

P. J.

Reicher

S. D.

Wetherell

M. S.

(Eds.), Rediscovering the social group: A self-categorization theory (pp. 117-141). Oxford, UK: Basil Blackwell.

18.

Oakes

P. J.

Haslam

S. A.

Turner

J. C.

(1998). The role of prototypicality in group influence and cohesion: Contextual variation in the graded structure of social categories. In Worchel

Morales

J. F.

Paez

Deschamps

J.-C.

(Eds.), Social identity: International perspectives (pp. 75-92). London, England: SAGE.

19.

Ohtsubo

Watanabe

Kim

Kulas

J. T.

Muluk

Nazar

. . . Zhang

(2012). Are costly apologies universally perceived as being sincere? A test of the costly apology-perceived sincerity relationship in seven countries. Journal of Evolutionary Psychology, 10, 187-204.

20.

Okimoto

T. G.

Wenzel

(2008). The symbolic meaning of transgressions: Towards a unifying framework of justice restoration. In Hegtvedt

Clay-Warner

(Eds.), Advances in group processes (pp. 291-326). Bingley, UK: Emerald.

21.

Okimoto

T. G.

Wenzel

Hornsey

M. J.

(2015). Apologies demanded yet devalued: Normative dilution in the age of apology. Journal of Experimental Social Psychology, 60, 133-136.

22.

Philpot

C. R.

Balvin

Mellor

Bretherton

(2013). Making meaning from collective apologies: Australia’s apology to its indigenous peoples. Peace and Conflict: Journal of Peace Psychology, 19, 34-50.

23.

Philpot

C. R.

Hornsey

M. J.

(2008). What happens when groups say sorry: The effect of intergroup apologies on their recipients. Personality and Social Psychology Bulletin, 34, 474-487.

24.

Philpot

C. R.

Hornsey

M. J.

(2011). Memory for intergroup apologies and its relationship with forgiveness. European Journal of Social Psychology, 41, 96-106.

25.

Putnam

R. D.

(1988). Diplomacy and domestic politics: The logic of two-level games. International Organization, 42, 427-460.

26.

Reicher

(1996). Social identity and social change: Rethinking the context of social psychology. In Robinson

W. P.

(Ed.), Social groups and identities: Developing the legacy of Henri Tajfel (pp. 317-336). Oxford, UK: Betterworth-Heinemann.

27.

Richards

Hewstone

(2001). Subtyping and subgrouping: Processes for the prevention and promotion of stereotype change. Personality and Social Psychology Review, 5, 52-73.

28.

Schumann

(2012). Does love mean never having to say you’re sorry? Associations between relationship satisfaction, perceived apology sincerity, and forgiveness. Journal of Social and Personal Relationships, 29, 997-1010.

29.

Shnabel

Nadler

(2008). A needs-based model of reconciliation: Satisfying the differential emotional needs of victim and perpetrator as a key to promoting reconciliation. Journal of Personality and Social Psychology, 94, 116-132.

30.

Smith

(2008). I was wrong: The meanings of apologies. New York, NY: Cambridge University Press.

31.

Steele

R. R.

Blatz

C. W.

(2014). Faith in the just behavior of others: Intergroup apologies and apology elaboration. Journal of Social and Political Psychology, 2, 268-288.

32.

Tavuchis

(1991). Mea Culpa: A sociology of apology and reconciliation. Stanford, CA: Stanford University Press.

33.

Turner

J. C.

(1991). Social influence. Buckingham, UK: Open University Press.

34.

Wenzel

Okimoto

T. G.

(2010). How acts of forgiveness restore a sense of justice: Addressing status/power and value concerns raised by transgressions. European Journal of Social Psychology, 40, 401-417.

35.

Wenzel

Okimoto

T. G.

(2014). On the relationship between justice and forgiveness: Are all forms of justice made equal? British Journal of Social Psychology, 53, 463-483.

36.

Wenzel

Okimoto

T. G.

(2015). “We forgive”: A group’s act of forgiveness and its restorative effects on members’ feelings of justice and sentiments towards the offender group. Group Processes & Intergroup Relations, 18, 655-675.

37.

Wohl

M. J. A.

Branscombe

N. R.

Klar

(2006). Collective guilt: Emotional reactions when one’s group has done wrong. European Review of Social Psychology, 17, 1-37.

38.

Wohl

M. J.

Cohen-Chen

Halperin

Caouette

Hayes

Hornsey

M. J.

(2015). Belief in the malleability of groups strengthens the tenuous link between a collective apology and intergroup forgiveness. Personality and Social Psychology Bulletin, 41, 714-725.

39.

Wohl

M. J. A.

Hornsey

M. J.

Bennett

S. H.

(2012). Why group apologies succeed and fail: Intergroup forgiveness and the role of primary and secondary emotions. Journal of Personality and Social Psychology, 102, 306-322.

40.

Wohl

M. J. A.

Hornsey

M. J.

Philpot

C. R.

(2011). A critical review of official public apologies: Aims, pitfalls, and a staircase model of effectiveness. Social Issues and Policy Review, 5, 70-100.

41.

Wohl

M. J. A.

Matheson

Branscombe

N. R.

Anisman

(2013). Victim and perpetrator groups’ responses to the Canadian government’s apology for the head tax on Chinese immigrants and the moderating influence of collective guilt. Political Psychology, 34, 713-729.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.15 MB