Literature Review Paper

The Journal of Social Psychology, 2012, 152(6), 727–745

Copyright © Taylor & Francis Group, LLC

How Voluntariness of Apologies Affects Actual and Hypothetical Victims’

Perceptions of the Offender

ALAYNA JEHLE R&D Strategic Solutions

MONICA K. MILLER MARKUS KEMMELMEIER

University of Nevada

JONATHAN MASKALY University of South Florida

ABSTRACT. Apologies are important in social interactions. Study 1 investigated participants’ reactions after being insulted by a confederate and receiving no apology, a vol- untary apology, a coerced apology with consequences (i.e., explicitly coerced apology), or a coerced apology without consequences (i.e., implicitly coerced apology). Receiving any apology produced more positive perceptions of the offender and less serious recommended punishments than no apology. Study 2 replicated Study 1, except participants read about the insult and imagined being a victim (instead of being an actual victim as in Study 1). Actual victims distinguished between types of apologies while hypothetical victims did not. Results have implications for court-ordered apologies.

Keywords: apologies, coercion, conflict resolution, impression management, psychology and law

APOLOGIES ARE A WAY in which relationships between offenders and their victims can be repaired. To have this beneficial effect, it is typically assumed that offenders truly and sincerely feel sorry about what they have done (Lazare, 2004). However, in many contexts this assumption clearly does not hold. For instance, parents and teachers may order a child to apologize for a wrong committed against a peer. Can such an apology truly restore the relationship between the children? One might argue that, when forcing their child to apologize, parents

Address correspondence to Alayna Jehle, R&D Strategic Solutions, 117 W. 58th St., 2G, New York, NY 10019, USA; [email protected] (e-mail).

727

728 The Journal of Social Psychology

might be less interested in promoting peer-to-peer reconciliation than teaching their child their manners. However, when courts order an otherwise reluctant defendant to apologize, this often occurs with the explicit intent to improve, if not restore relationships between the victim and defendant (Kahan, 1996). Does such a coerced apology truly restore relations between the two parties and improve the victim’s perceptions of the defendant? The current research was designed to answer this question.

In addition, the current research seeks to determine whether actual vic- tims and hypothetical victims respond similarly to a transgression. Judges may assume that, if they force offenders to apologize, that this will be beneficial to and welcomed by a victim. This assumes that judges are able to predict how they would actually feel if they were a victim. It is entirely possible, however, that people imagining to be victimized might have quite different reactions from actual victims (e.g., De Cremer, Pillutla & Folmer, 2011). Thus, the current study will determine whether hypothetical and actual victims respond to transgressions similarly.

Apologies and Forgiveness

Victims are likely to accept an apology because acceptance is the socially nor- mative response according to the corrective interchange process (Goffman, 1955). Victims are expected to forgive offenders when they apologize as a means of restoring the relationship, and victims who reject an apology are viewed unfa- vorably for violating the social norm (Bennett & Dewberry, 1994). When the victim accepts the apology, the relationship and both parties’ “faces” are restored (i.e., they regain a positive self-image), which fulfills the role of the corrective interchange in social situations (Goffman, 1955).

When offenders apologize, perceivers view them more favorably on a variety of characteristics, including blameworthiness, likeability, and being a good per- son (Darby & Schlenker, 1982). The apology demonstrates that the transgression does not reflect the offender’s disposition, suggesting that the wrongdoing is not a behavioral constant (Bennett & Dewberry, 1994). For these reasons, victims are expected to forgive the offender, which would suggest that other perceivers (e.g., bystanders) also should forgive. Then, these forgiving responses will continue to improve the attitudes toward the offender.

Voluntary Versus Coerced Apologies

Voluntary apologies are motivated by internal causes, whereas coerced apolo- gies are brought about by external circumstances. Victims are more likely to for- give offenders who are truly sorry for their transgressions (Allan, Allan, Kaminer, & Stein, 2006); therefore, internally motivated apologies can be expected to be

Jehle et al. 729

more effective in promoting forgiveness and improving perceptions of the offender than externally motivated apologies. Hence, when relying on principles of inter- personal reasoning, such as Kelley’s (1971) discounting principle, the prediction would have to be that involuntary apologies are bound to be perceived as insincere, and that no good can come from them to their recipients. This is supported by a case study of a Victim Offender Mediation program that found apologies were not well received when the victim felt the offender’s apology was “lip service” and not genuine (Choi & Severson, 2009).

Risen and Gilovich (2007) examined the effects of coerced apologies on vic- tims and perceivers by comparing the effects of a freely given apology by an offender with a coerced apology that only occurred after a bystanding peer became upset and requested that the offender apologize. Interestingly, their victims did not differentiate between voluntary and coerced apologies when making assess- ments of the offender; that is, an apology resulted in an improved perception of the offender on the part of the victim regardless of whether this apology was issued voluntarily or not. Yet, bystanding perceivers did take the voluntariness of the apology into consideration by judging offenders giving coerced apologies more harshly. Specifically, perceivers rated their impressions of the offender more neg- atively, expressed less interest in working with her again, and punished her more by recommending less monetary payment when the offender provided a coerced rather than voluntary apology. This suggests that, although victims may want to reject an insincere apology, normative pressure demands that they should accept the apology (coerced or not), and therefore they do. Overall, the authors concluded that coerciveness of an apology appears to make a difference only for external perceivers and not for victims themselves.

With regard to the victims’ response to the coerced apology, Risen and Gilovich (2007) face some limitations. Specifically, in their research the ele- ment of coercion was fairly weak because there was no clear power differential between the person coercing and the person being coerced, as the person offer- ing the coercion was a confederate participating in the research study with the actual participants. Additionally, if the participant chose not to apologize there were no consequences, calling into question whether there was truly coercion or simply a prompt for the apology. From this vantage point, it may not be surpris- ing if victims discounted the element of coercion and instead focused on the fact that they received an apology at all. Therefore, the current research set out to test stronger types of coerced apologies to determine if they elicit the same responses as voluntary apologies.

Overview of Research

The current research examined the influence of apologies on their recipients’ responses to an offender. At the heart of our investigation is the question whether victims will or will not differentiate between voluntary and involuntary apologies.

730 The Journal of Social Psychology

When responding to an offender’s apology, will victims overlook whether the apology was freely given or not, as proposed by Risen and Gilovich (2007)? Or will they pay close attention to whether the offender is acting voluntarily, with voluntary and sincere apologies resulting in more favorable responses on the part of the victim, but not necessarily coerced ones, as suggested by Allan et al. (2006) and Choi and Severson (2009)?

To tests these competing hypotheses, we studied two kinds of involuntary apologies. First, we focused on coerced apologies that were similar to those stud- ied by Risen and Gilovich (2007), which we categorize as “implicit coercion.” Implicitly coerced apologies involve a person telling someone to apologize but there are no explicit consequences for not apologizing. However, in our research, the coercion came from a status-superior individual rather than a same-status peer as in Risen and Gilovich. Arguably, this scenario is more realistic in that it resem- bles situations in which a supervisor requests that an employee apologize to a colleague, or a parent or teacher requests that a child apologize to another child.

Second, and more importantly, our research investigates an even stronger level of coerced apologies, which we refer to as “explicit coercion.” That is, beyond being ordered to apologize by a status-superior individual, the superior also holds real power over the person and threatens negative consequences should the person refuse the request. As a result, we hold that our research is the first to investigate explicitly coerced apologies, that is, apologies that, if not given, will result in seriously adverse consequences for the person being asked to apol- ogize. Such a situation is highly realistic in that it resembles particular situations, such as a judge ordering a defendant to apologize under threat of punishment. Though systematic data are difficult to come by, anecdotally, recipients of such involuntary apologies often feel satisfied with them (Latif, 2001). Our study will determine whether an involuntary apology would in fact be better than no apology, and further whether it would be similar to a voluntary apology.

A third important aspect of our research concerns the possible discrepancy between the actual consequences of involuntary apologies, and subjective theories about such consequences. Social psychological research has repeatedly shown peoples’ beliefs about appropriate responses to situations do not reflect actual responses to those situations. When people are asked to predict their behavior and emotions, their predictions are often quite different than how they actually react when they later experience the situation (Diekmann, Tenbrunsel, & Galinsky, 2003). De Cremer and colleagues (2011) found that participants who imagined that they had received an apology valued that apology much more, and were later more trusting of the transgressor, than participants who actually received an apol- ogy. This suggests that hypothetical victims are much more forgiving and trusting than actual victims. This may indicate that judges (like the hypothetical victims in the De Cremer et al. study) may overestimate the value of an apology. De Cremer et al. did not investigate different types of apologies, but do suggest that actual and hypothetical victims perceive apologies differently. Thus, the current studies expand on this line of research as well.

Jehle et al. 731

To tackle these issues, Study 1 examined the effects of voluntary and invol- untary apologies on actual victims, whereas Study 2 examined the same effects on hypothetical victims, requiring an application of subjective understandings with- out any actual experience with the situation. Simulation theory states that when observers try to predict someone’s behavior, the observers tend to picture them- selves in the other person’s position and think what they would hypothetically do in the same situation (Gordon, 1986). As such, both people who coerce apologies and people who criticize coerced apologies essentially take on the role of hypo- thetical victim in trying to predict how the victim would react. These observers believe they understand the effects (or lack thereof) that the apology will have even though they only have a subjective understanding of the situation. Therefore, these studies were designed to determine if hypothetical victims could indeed predict the actual victim’s response to a coerced apology.

Study 1

To obtain realistic results of actual victims, participants in this study were vic- timized by being the target of a verbal insult of their intellectual ability. We tested two competing hypotheses: (a) Victims will not distinguish between coerced and non-coerced apologies (Risen & Gilovich, 2007); thus all of the apology condi- tions will produce similar beneficial effects; and (b) More voluntary apologies will result in more positive impressions because the victim is more convinced of the sincerity of the offender’s intentions (Allan et al., 2006); thus voluntary apolo- gies will be most beneficial, followed by implicitly coerced, and then explicitly coerced apologies, with no apology causing the harshest judgments.

In order to qualify as an explicitly coerced apology, we considered it impera- tive that the recipient of the apology realizes that not apologizing would result in unfavorable consequences for the offender. Meanwhile, we consider an apology that is demanded but without there being a clear threat, as was the case in Risen and Gilovich’s study, an implicitly coerced apology.

Method

Participants

Participants were 99 college students who received course credit for their participation. Seven participants indicated suspicion regarding the study’s purpose and were thus excluded. The final sample consisted of 92 participants, the majority of whom were female (73%).

Design

The design of this study was a between-groups design with four conditions: voluntary apology, implicitly coerced apology (with no consequences), explicitly coerced apology (with negative consequences), and no apology.

732 The Journal of Social Psychology

Procedure

Participants reported individually to an experiment supposedly about distrac- tion and performance. Upon arrival, a same sex pair of experimenter and research assistant introduced themselves. Participants were told the study examined how background noise affects performance on an easy anagram task. Participants also learned they would evaluate the assistant at the conclusion of the study and that the assistant’s course grade would reflect how well the assistant ran the study.

After signing the informed consent form, the experimenter exited and the research assistant informed the participant that s/he had been randomly assigned to the control condition, which meant there would be no distraction. The research assistant gave the participant a list of ten anagrams and said “Complete them as quickly as you can, but do not worry, they are very easy.” In actuality, only three of the anagrams could be solved. After one minute, the research assistant asked, “You’re not done yet?” When the participant responded, the research assistant responded, “Even a six year old would be finished by now, they are so easy. Just keep working.” Thirty seconds later, the experimenter entered the room and the research assistant said, “We may as well end this now. It is obvious that s/he is never going to finish. No one else has taken this long. I don’t know what is wrong with him/her.” Next, the experimenter said to the participant, “Okay, we will just use the ones you have done.” Then, the experimenter handed the ana- gram list to the research assistant and said: “Take this to my office while we do your evaluation.” The research assistant exited and the apology manipulation followed:

In the voluntary apology condition and the no apology condition, the exper- imenter said to the participant, “Oh, I forgot I locked my office. I will be right back.”

In the implicitly coerced apology condition, the experimenter said to the participant, “I am very annoyed at how rude s/he was to you. That was very unpro- fessional to rush you and talk down to you like that. I am going to tell him/her that s/he must apologize to you.”

In the explicitly coerced apology condition, the experimenter’s statement was identical to the implicitly coerced apology condition, except that that the last sen- tence included an explicit threat of negative consequences: ”I am going to tell him that s/he must apologize to you or else s/he is going to receive a failing grade for his/her course.”

The experimenter then exited for a brief time and instructed the research assis- tant whether to apologize or not before both re-entered. Except for the no apology condition, the research assistant was blind to the apology condition and had no knowledge whether the apology condition was voluntary, implicitly coerced, or explicitly coerced.

The experimenter handed the research assistant a credit slip to fill out for the participant while the participant completed the evaluation. In all experimental

Jehle et al. 733

conditions, except the no apology condition, the research assistant said: “I am sorry I was so rude to you” before leaving the room. In the no apology condition, the research assistant simply took the credit slip and left.

The experimenter was trained to not give any facial or body language reaction to the insult or the apology, and instead to simply give his/her prompted lines in a pleasant tone.

Questionnaires

After the apology manipulation, participants filled out a “Research Assistant Evaluation” form, which they were told would be used to determine the research assistant’s grade for the course. The experimenter asked participants to be completely honest as the research assistant would never see their responses. Participants evaluated the research assistant on five perception measurements (courteous, likeable, responsible, respectful, overall perception) using a 9-point Likert scale. The five variables were highly correlated (α = .93) and therefore were combined. Participants also recommended a grade on a 100-point scale that the research assistant should receive for his/her class.

Participants also filled out a “Problems Conducting Experiment” form. The experimenter explained that this was a standard form required whenever there was a problem with a research study, which in this case was the research assistant’s unprofessional conduct. The experimenter added that participants’ responses on this form would be weighted heavily in the research assistant’s final grade. Participants assessed how much they believed that the research assistant should be removed from his/her position. As a manipulation check, participants answered whether or not they received an apology and then rated the apology’s voluntariness and sincerity.

Next, participants were debriefed and given a credit slip to obtain credit for their classes. During the debriefing, the research assistant reconciled with the par- ticipant and gave a sincere apology for being rude as part of his/her role in the experiment. The experimenter also informed the participant that most of the ana- grams in the study’s task were unsolvable. Most participants believed the cover story and the few who indicated suspicion were removed from the analysis.

Results1

Manipulation Check

Eight participants (8%), who were roughly equally distributed across exper- imental conditions, χ 2(df = 3) = 6.09, p = .11, incorrectly answered whether the research assistant apologized. However, analysis with and without these participants led to identical conclusions. Thus, due to the possibility that the manipulation check failures were actually an evaluation of the offender rather

734 The Journal of Social Psychology

than forgetting/mishearing an apology, the analysis including these participants are reported.

Perceptions of the Apology

To assure the validity of the experimental manipulation, it was important that participants viewed the voluntary apology as less coercive than the two coerced apology conditions. A one-way ANOVA across the apology conditions revealed a significant effect on voluntariness ratings, F(2, 60) = 32.17, p < .001, ηp2 = .52. Participants rated the offender’s apology in the voluntary apology condition (M = 7.55, SD = 1.64) as significantly more voluntary than the implicitly coerced apology (M = 2.86, SD = 2.39), p < .001, and explicitly coerced apology conditions (M = 2.82, SD = 2.36), p < .001, with the latter two conditions not being different from each other, p = 1.00. This confirmed that the apology manipulation was effective in conveying the distinction between voluntary and different involuntary apologies.

There was also an effect on the perceived sincerity of the apology, F(2, 60) = 4.49, p = .028, ηp2 = .13. Participants rated the offender’s apology in the volun- tary apology condition (M = 5.75, SD = 2.67) as significantly more sincere than the explicitly coerced apology (M = 3.91, SD = 1.66), p = .017 and somewhat more sincere than the implicitly coerced apology condition (M = 4.24, SD = 1.89), p = .063, while the two coerced apology conditions did not vary, p = .87.

Perceptions of the Offender

An initial omnibus ANOVA revealed that the combined perceptions varied significantly across the four conditions, F(3, 87) = 6.73, p < .001, ηp2 = .19 (see Table 1, top panel). However, to test more focused hypotheses, we used a series of planned contrast analyses. Relying on the approach by Rosenthal, Rosnow and Rubin (2000), we were able to test competing hypotheses using non-orthogonal contrasts (see also Rosnow, Rosenthal & Rubin, 2000). Rosenthal et al.’s proce- dures allow the analyst to examine how well a predicted pattern fits the data by specifying a sequence of contrast weights (λ) that reflects the pattern of mean differences expected under a particular hypothesis. These authors’ approach also permits one to test if one hypothesis provided a better fit than another. In addi- tion to F-statistics, which reflects to what extent the match between prediction and observation might be the product of chance alone, we also report so-called alerting correlations (ralerting) which reflect how closely the observed patterns of means corresponds to the predicted patterns (Rosenthal et al., 2000). Other cor- relational parameters recommended by Rosenthal et al. (2000), e.g., rcontrast or reffect size, are not reported here as they led to conclusions that are redundant with the conclusions we drew based on the alerting correlations.

Jehle et al. 735

The first analysis tested the hypothesis that voluntary and implicitly coerced apologies create a more favorable impression than an explicitly coerced apology and no apology, contrast weights (+1, +1, –1, –1; see Table 1 for order of condi- tions). Results confirmed this “two vs. two (conditions)” hypothesis, F(1, 87) = 15.54, p < .001, ralerting = .89. Next, we generated contrasts that examined Risen and Gilovich’s claim that, because victims do not distinguish between coerced and non-coerced apologies, all apologies should produce similar effects, contrast weights (+1, +1, +1, –3). This hypothesis was found also to be compatible with the data, F(1, 87) = 15.22, p < .001, ralerting = .84, with a comparative test of these two contrasts showing only a minimal difference in fit, F�contrasts(1, 87) = .06, �ralerting = .05 (see Table 2 for a summary of the contrast analyses).

Lastly, we tested the hypothesis that victims’ perceptions of the offender are indeed a function of the voluntariness of the apology such that more voluntary apologies result in more positive impressions because the victim is more convinced of the sincerity of the offender’s intentions (Allan et al., 2006). Recall that this “degree of voluntariness” hypothesis is in stark contrast to Risen and Gilovich (2007) who argued that voluntariness should be of little consequence to victims’ perceptions of an offender. We tested this alternative hypothesis using another set of contrast weights, which assumed that voluntariness of the apology steadily

TABLE 1. Perceptions and Punishment of the Offender as a Function of Apology Type, Study 1 and 2

Apology type

Voluntary Implicit coercion

Explicit coercion No apology Omnibus

M (SD) M (SD) M (SD) M (SD) F

Study 1 (enacted) Comb.

percept. 4.62a (1.98) 4.17a (1.96) 3.42ab (1.57) 2.51b (1.38) 6.73

∗∗∗

Grade 82.95a (12.87) 76.22ab (11.10) 71.36bc (9.58) 63.16c (16.57) 9.37 ∗∗∗

Removal 3.45a (2.93) 3.73a (2.07) 4.61ab (1.90) 6.36b (2.34) 7.44 ∗∗∗

Study 2 (hypothetical) Comb.

percept. 2.44 (0.91) 2.08 (1.14) 2.15 (0.97) 1.98 (0.80) 1.19

Grade 69.74 (9.01) 65.56 (11.64) 65.86 (10.30) 65.47 (11.57) 1.02 Removal 5.67 (1.84) 6.85 (2.15) 6.45 (1.94) 6.40 (2.13) 1.60

∗∗∗p < .001 Notes: Means that do not share the same superscript differ at the .05 level based on Tukey or, in case of heterogeneous variances, on Dunnett’s T3 test. “Grade” was based on a 100-point scale and refers to the recommended grade that participants gave for the research assistant. The combined perception ratings and “removal” (belief that the research assistant should be removed from the experiment) were each based on a 9-point Likert scale.

736 The Journal of Social Psychology

TABLE 2. Summary of Contrast Analysis, Study 1 and 2

Contrasts

Hypothesis name “Two vs. two” “Risen & Gilovich”

“Degree of voluntariness”

+1 +1 −1 −1 +1 +1 +1 −3 +3 +1 −1 −3 Contrast weights λ F ralerting F ralerting F ralerting

Study 1 (enacted) Comb. percept. 15.54∗∗∗ .89 15.22∗∗∗ .84 19.12∗∗∗ .99 Grade 20.31∗∗∗ .85 20.21∗∗∗ .89 27.45∗∗∗ .995 Removal 14.87∗∗∗ .83 19.74∗∗∗ .93 19.02∗∗∗ .95

Study 2 (hypothetical) Comb. percept. 1.25 .57 1.44 .62 2.71 .86 Grade 0.96 .56 0.48 .38 1.94 .78 Removal 0.21 .19 0.03 .08 1.15 .47

∗∗∗p < .001

declined from the voluntary to the no apology condition (+3, +1, –1, –3). This contrast resulted in the best fit for the data, F(1, 87) = 19.12, p < .001, ralerting = .99 (see also Table 2). A comparative test between this contrast and the first and second contrast revealed that the contrast based on the degree of voluntariness hypothesis provided a noticeably better fit, F�contrasts(1, 87) = 0.86, p = .36, �ralerting = .21 and F�contrasts(1, 87) = 0.88, p = .35, �ralerting = .22, though not at conventional levels of statistical significance (cf. Rosenthal et al., 2000).

Punishment of the Offender

We used a similar approach when analyzing proposed punitive measures. Omnibus analyses revealed that the recommended grade differed by apology type, F(3, 86) = 9.37, p < .001, ηp2 = .25 (see Table 1 top panel). The contrast between the voluntary and implicit apology on the one hand and the explicitly coerced apol- ogy and no apology condition on the other hand was significant (“two vs. two” hypothesis), F(1, 86) = 20.31, p < .001, ralerting = .85, as was the contrast imple- ment of Risen and Gilovich’s (2007) assumption, F(1, 86) = 20.21, p < .001, ralerting = .89, with these two contrasts offering virtual identical fit with the data, F�contrasts(1, 86) = 0.04, �ralerting = .04. However, the contrast testing degree of voluntariness as predictor of the consequences of an apology provided the better fit again, F(1, 86) = 27.45, p < .001, ralerting = .995. As before, this particular con- trast was markedly better than the “two vs. two” contrast, F�contrasts(1, 86) = 2.64,

Jehle et al. 737

p = .11, �ralerting = .31, or the second, Risen and Gilovich inspired contrast, F�contrasts(1, 86) = 1.80, p = .18, �ralerting = .26, even when the difference did not surpass conventional levels of statistical significance.

Similar findings were also obtained for participants’ ratings that the offender should be removed from his position, omnibus F(3, 86) = 7.44, p < .001, ηp2 = .21. The contrast pitting two pairs of conditions against each other was again viable, F(1, 86) = 14.87, p < .001, ralerting = .83. However, both the contrast anal- ysis based on Risen and Gilovich (2007), F(1, 86) = 19.74, p < .001, ralerting = .93, and the analysis contrasting the degree of voluntariness as a predictor of the conse- quence of apology were not only compatible with the data, F(1, 86) = 19.02, p < .001, ralerting = .95, but both provided a slightly better model fit than the “two vs. two” contrast, F�contrasts(1, 86) = 0.22, p = .64, �ralerting = .10, and F�contrasts(1, 86) = 1.26, p = .26, �ralerting = .24, respectively. At the same time the last two sets of contrasts (Risen & Gilovich, degree of voluntariness) did not differ between each other in how well they accounted for the observed data, F�contrasts(1, 86) = 0.02, �ralerting = .03.

Discussion

Consistent with previous research (e.g., Darby & Schlenker, 1982), Study 1 found that victims viewed offenders more favorably and recommended more lenient punishments when they apologized compared to when they did not apol- ogize. All three of the models consistently supported this notion that apologies can be beneficial. However, the contrast model that best fit our data showed that victims were quite sensitive to the degree of voluntariness of the apology in that the voluntary apology produced the most beneficial effects, followed by the implicitly coerced apology, then the explicitly coerced apology. The no apology condition produced the most unfavorable perceptions and the harshest punish- ments. Therefore, it appears that actual victims adjust their perceptions of the offender based on the apology’s level of voluntariness, which is consistent with previous research (e.g., Allan et al., 2006) that found victims were more forgiv- ing when they were more convinced of the sincerity of the offender’s intentions. However, this does not support Risen and Gilovich’s (2007) claim that actual victims respond the same to any apology regardless of coercion.

Study 2

The goal of the second study was to examine whether people can anticipate the consequences of coerced apologies. This question is of particular relevance when it comes to the decision to force an offender to apologize to a victim, such as when a judge orders defendants to apologize to the victims of their crimes. It cannot be assumed that decision makers themselves know what it feels like to be on the receiving end of a coerced apology; thus, they must rely on their naïve

738 The Journal of Social Psychology

theories of such situations. Based on research in social psychology (e.g., Miller, Turnbull & McFarland, 1990), it is plausible to assume that people arrive at pre- dictions concerning the effects of apologies by simulating in their own minds how they would feel in such a situation (Gordon, 1986). Using this premise of simu- lation theory, observers essentially take on the role of hypothetical victims when they try to understand the emotions and behaviors of someone who has just been harmed. Indeed, much of social-psychological research assumes that individuals are able to accurately predict their own (and others’) responses without having to experience a situation first-hand. This ability to simulate one’s experiences is generally a powerful tool when navigating the social world even though research clearly demonstrates the limits of insight that hypothetical scenarios can produce (e.g., De Cremer et al., 2011).

If people do not have any first-hand experience with their own responses to coerced and voluntary apologies, it might be plausible for them to rely on pre- vious observation when they witnessed others receiving such apologies. Recall that Risen and Gilovich (2007) investigated different vantage points on apologies. In particular, these authors found that observers differentiated between whether an apology was given voluntarily or not and that they were much less forgiving with regard to coerced apologies than the recipients of the same apologies. This finding suggests that, to the extent that a person ever observed a coerced apology being offered, relying on this observation would bring about the correct intuition, even in cases when the recipient of an apology would not make this distinction him or herself. At the same time, one would have to assume the opportunity for such an observation to arise only rarely; hence, it is an open question if individuals would be able predict how they or anybody else would respond to receiving a coerced apology.

In Study 2, we examined if hypothetical victims would be able to anticipate how actual victims would feel after receiving a voluntary apology, an implicitly coerced apology, an explicitly coerced apology, or no apology. Specifically, par- ticipants of Study 2 were cued into precisely the situation that participants of Study 1 had experienced first-hand. With the actual victims of Study 1 and the hypothetical victims of Study 2 being from the same population, there was no rea- son to assume that there would be any differences in their actual reactions to the apologies. Thus, a direct comparison across experiments allowed us to determine to what extent our hypothetical victim would be able to accurately anticipate the response that they would likely show if they experienced the apologies themselves.

Method

Participants

Participants were 117 college students who received course credit for their participation (54 males).

Jehle et al. 739

Design, Procedure, and Study Materials

Participants participated online using SurveyMonkey. The design paralleled that of Study 1, except participants read instructions asking them to imagine them- selves in a psychological experiment in which they were insulted by a research assistant. Specifically, participants in the four conditions of Study 2 read a detailed summary of the scenarios that participants in Study 1 experienced first-hand. Subsequently, participants were asked to complete the same questionnaires pre- viously completed by Study 1 participants. Lastly, participants were thanked, debriefed, and granted course credit through the website.

Results

Manipulation Check

Thirteen participants (11%) incorrectly answered whether the research assis- tant apologized. The individuals were roughly equally distributed across condi- tions, χ 2 (df = 3) = 2.65, p = .45, but as in Study 1, their inclusion did not alter the results and thus they are reported.

Perceptions of the Apology

As in Study 1, participants rated the apology as much more voluntary when it was not coerced (M = 5.36, SD = 1.82), compared to the explicitly and implicitly coerced conditions, (M = 1.96, SD = 1.69 and M = 1.87, SD = 1.70, respec- tively), F(2, 72) = 32.68, p < .001, ηp2 = .48. Similarly, another ANOVA, F(2, 70) = 3.12, p = .05, ηp2 = .08, showed that a voluntary apology was perceived as more sincere (M = 3.16, SD = 1.55) than an explicitly coerced apology (M = 2.04, SD = 1.33), p < .05, with the implicitly coerced apology falling in between without being significantly different from either condition (M = 2.33, SD = 1.95).2 Thus, participants (like those in Study 1) were clearly aware of the nature and context in which the apologies were given.

Perceptions of the Offender

Unlike in Study 1, the combined perception of the offender in Study 2 (α = .81) did not vary as a function of apology condition, omnibus F(3, 113) = 1.19, p = .32, ηp2 = .03 (see Table 1, lower panel). When we repeated the series of contrast analyses of Study 1, there was no evidence that the combined voluntary apology and implicitly coerced apology conditions differed from the explicitly coerced apology and no apology conditions (“two vs. two hypothe- sis”), F(1, 113) = 1.25, p = .27, ralerting = .57 (see Table 2). Likewise, there

740 The Journal of Social Psychology

was no support for Risen and Gilovich’s notion that the three apology condi- tions should differ from the no apology condition, F(1, 113) = 1.44, p = .23, ralerting = .62. Moreover, Study 2 varied clearly from Study 1 in that the con- trast based on the decreasing degree of voluntariness of the apology did not provide a good model fit for perceptions of the offender, contrast F(1, 113) = 2.71, p = .102, ralerting = .86 (see Table 2 for a summary of all contrast analy- ses). Overall, these findings demonstrate that the hypothetical victims of Study 2 were unable to accurately anticipate how they would likely feel. In particular, their responses did not indicate that hypothetical apologies had the same bene- ficial effects on perceptions of the offender that were previously documented in Study 1 (and elsewhere in the literature, e.g., Darby & Schlenker, 1982). While Study 1 found that even coerced apologies might have a salubrious influence on the victim’s perceptions of the offender, this clearly was not the case for Study 2.

Punishment of the Offender

In Study 2 there were no significant differences in the grades that hypothet- ical victims assigned to the various offending research assistants, F(3, 107) = 1.02, p = .39, ηp2 = .03 (see Table 1 lower panel).3 Similarly, the contrast analy- sis comparing the voluntary and the implicitly coerced apology conditions with the explicitly coerced apology and the no apology conditions (“two vs. two”) were not significant, F(1, 107) = 0.96, p = .33, ralerting = .56, nor was the con- trast inspired by Risen and Gilovich, F(1, 107) = 0.48, p = .49, ralerting = .38. The contrast assessing if the degree of voluntariness of the apology predicted the response of hypothetical victims was also not reliable, F(1, 107) = 1.94, p = .17, ralerting = .78.

As with grade, there was no variation in whether hypothetical victims thought that the offender should be removed from running the study, F(1, 112) = 1.60, p = .19, ηp

2 = .04. Neither the first “two vs. two” contrast, F(1, 112) = 0.21, p = .65, ralerting = .19, the one based on Risen and Gilovich (2007), F(1, 112) = 0.03, p = .86, ralerting = .08, nor the one testing if the consequences of apologies are shaped by their degree of voluntariness were reliable, and F(1, 112) = 1.15, p = .29, ralerting = .47. Note that these findings occurred even though the victims of Study 1 and Study 2 were in complete agreement as to the sincerity and voluntariness of the apology.

Discussion

Study 2 revealed that hypothetical victims are not at all sensitive to the voluntariness of the apology. The hypothetical victims’ judgments of the offender revealed no differences based on apology type; they viewed the offender

Jehle et al. 741

negatively and judged him harshly regardless of whether his apology was volun- tary or coerced.

General Discussion

The purpose of these two studies was to examine whether the coercive nature of apologies affect the responses of both an actual victim and a hypothetical vic- tim. Both studies supported the concept that apologies in general can produce beneficial effects for the offender as compared to no apology. However, Study 1 revealed that actual victims differentiated between apology types when judging the offender, while Study 2 demonstrated that hypothetical victims did not make such distinctions. Study 1 showed that actual victims were sensitive to the coer- cive nature of apologies in that the more voluntary the apology, the more favorable judgments they had of the offender and the more lenient they were with punish- ments. On the other hand, Study 2 participants’ responses did not vary based on the voluntary level of the apology, which suggests that hypothetical victims view any apology as somewhat beneficial without distinguishing between apology types.

Implications for Psychology Research

The findings of the current research contradict Risen and Gilovich’s (2007) finding that observers are sensitive to coerced apologies while actual victims are not. This difference may be a consequence of the specific methodology chosen. Risen and Gilovich used two confederates that were perceived to be of equal status, which means the apology could be seen as less than coercive. Coercion implies that there is a power differential between the person that is demanding the apology and the person that is apologizing. This power differential was at play in the current research, coming in the form of an experimenter responsible for giving a research assistant a grade for the class. Within the explicitly coerced apology condition, this power differential was made salient by the apology prompt of the researcher indicating negative consequences for non-compliance with the apology. In the Risen and Gilovich study, there was no implied power differential between the confederates and no negative consequences for non-compliance, which did not fully exploit the potential for forced apologies. The current research used a salient power differential which was better able to test victims’ reactions to truly coerced apologies.

More generally, these studies support previous research which indicates that people are poor predictors of their future behavior and emotions (Diekmann et al., 2003). For instance, hypothetical victims valued an apology more than the actual victims (De Cremer et al., 2011). Like De Cremer et al., we found that hypothetical victims reacted differently than actual victims on some measures, specifically that hypothetical victims differentiated between types of apologies less than actual victims.

742 The Journal of Social Psychology

Implications for the Justice System

Some justice systems around the world are utilizing restorative justice approaches to allow or force an offender to apologize to the victim (BBC, 2009). Although critics argue that coerced apologies are ineffective (Latif, 2001), our findings indicate that, from the victim’s perspective, a coerced apology may pro- duce beneficial effects compared to no apology if the apology can be made to appear minimally coerced. This is especially important because of some sentenc- ing practices utilized by courts in the United States. Currently, many courts are forcing offenders to issue formal apologies to their victims and certain states have statutorily mandated that court-ordered apologies be issued for certain crimes (e.g., hate crimes and vandalism; N.H. Rev. Stat. § 169-B:45; W.S.A. 938.34). Judges often reduce the defendants’ sentence if they comply (Latif, 2001), which may make these court-ordered apologies similar to the explicitly coerced apol- ogy in the current research. These judges appear to have the same view as our hypothetical victims in Study 2; the coercive nature of the apology does not mat- ter (even if the apology is given under coercion to avoid negative consequences) because any apology is better than no apology. As the current studies suggest, judges may be less sensitive to the type of apology compared to actual victims, much like our hypothetical victims of Study 2. Judges’ potential inability to dif- ferentiate between types of apologies can cause them to order apologies without any regard to how coerced the apology appears. Therefore, judges and offenders should do their best to make the apology appear less coerced and more sincere in order for victims to truly appreciate the apology.

Research has shown that apologies are particularly and consistently effective in criminal courts under a variety of circumstances. For example, defendants who are remorseful at trial are less likely to be sentenced to death and instead given a life sentence (Eisenberg, Garvey, & Wells, 1998). The results from the current research add further support for apologies in the legal system. Study 1 revealed that, although coerced apologies produced more negative judgments compared to a voluntary apology, implicitly coerced apologies produced more favorable responses than no apology at all. Even though voluntary apologies diminish the feelings of ill will toward an offender much more, some levels of coerced apolo- gies appear to be more beneficial than no apology. Therefore, victims will likely appreciate an apology if the apology can appear to be motivated by the offender’s desire to demonstrate true regret for his actions rather than merely to reduce his sentence.

Limitations and Future Research

These studies have two primary limitations. The first limitation stems from using students as the participants in the study because students sometimes respond differently than nonstudents in various situations (Gordon, Slade, & Schmitt, 1986). However, these studies tended to attribute the differences to students’ lack

Jehle et al. 743

of familiarity with the experimental task, whereas the current research involved tasks that both students and non-students are typically familiar with (i.e., judg- ing someone who is rude to them). Additionally, previous research has shown that college students and non-students do not differ in terms of their perceptions of offenders in the legal context (e.g., Bornstein, 1999), which is not too much different from making judgments about offenders in our research. Therefore, the implications likely still hold despite the sample. An additional limitation of the present research is that it only accounts for apologies in one specific situation; a relatively mild insult. It is possible that a person suffering a transgression in a different situation (e.g. medical malpractice; criminal activity) might not react the same way.

Future research should be designed to address some of these limitations. Specifically, future research should test whether or not coerced apologies have the same effect within the legal realm. This research would be particularly useful for judges deciding whether to force an offender to apologize. Research should also be conducted to determine whether similar results would be observed for more serious transgressions and transgressions in other settings. Despite the lim- itations, the results offer some initial support for transgressions in interpersonal relations.

Conclusion

Our two studies were designed to determine, first, if apologies that vary in the extent to which they are coerced or voluntary have a differential impact on actual victims of a transgression. Second, we were interested whether hypotheti- cal victims would be able to accurately anticipate the responses of actual victims. The answer to this first question is a clear Yes. Actual victims proved to be quite sensitive to whether the apologies were coerced or voluntary, with more volun- tary apologies resulting in more favorable responses to the offender. However, the answer to our second question is a qualified No. Responses of hypothetical victims were unable to accurately predict the responses of actual victims, even when they tended to anticipate that any apology, regardless if coerced or not, has favorable implications for actual victims’ perception of an offender. These results expand upon some of the previous research in the field; among other things because the present research makes salient the power differential that tends to be inherent in coerced apologies. These results could be particularly interesting to the legal sys- tem: apologies—even if coerced by judges—seem to be perceived positively by actual victims. Such findings indicate the value of apologies in the legal system, and in interpersonal interactions in general.

NOTES

1. Some variables contain missing data, and therefore, the degrees of freedom in the denominator vary slightly among different analyses.

744 The Journal of Social Psychology

2. Two statistical outliers were temporarily removed from this analysis. 3. Six extreme values were temporarily removed from this analysis.

AUTHOR NOTES

Alayna Jehle is affiliated with R&D Strategic Solutions, New York, NY, USA. Monica K. Miller is affiliated with the Department of Criminal Justice/Social Psychology, University of Nevada. Markus Kemmelmeier is affiliated with the Department of Criminal Justice/Social Psychology, University of Nevada. Jonathan Maskaly is affiliated with the Department of Criminal Justice, University of South Florida.

REFERENCES

Allan, A. Allan, M. M., Kaminer, D., & Stein, D. J. (2006). Exploration of the asso- ciation between apology and forgiveness amongst victims of human rights violations. Behavioral Sciences and the Law, 24, 87–102. doi:10.1002/bsl.689

BBC. (2009, October 29). Victim meetings “cut” youth crime. Retrieved from http://news. bbc.co.uk/2/hi/uk_news/8328529.stm.

Bennett, M., & Dewberry, C. (1994). “I’ve said I’m sorry, haven’t I?” A study of the iden- tity implications and current restraints that apologies create for their recipients. Current Psychology, 13, 10–21. doi:10.1007/BF2686855.

Bornstein, B. H. (1999). The ecological validity of jury simulations: Is the jury still out? Law and Human Behavior, 23, 75–91. doi:10.1023/A:1022326807441.

Choi, J. J., & Severson, M. (2009). “What! What kind of apology is this?”: The nature of apology in victim offender mediation. Children and Youth Services Review, 31, 813–820. doi: 10.1016/j.childyouth.2009.03.003

Darby, B. W., & Schlenker, B. R. (1982). Children’s reactions to apologies. Journal of Personality and Social Psychology, 43, 742–753. doi:10.1037/0022-3514.43.4.742.

De Cremer, D., Pillutla, M. M., Folmer, C. R. (2011). How important is an apology to you? Forecasting errors in evaluating the value of apologies. Psychological Science, 22, 45–48. doi: 10.1177/0956797610391101

Diekmann, K. A., Tenbrunsel, A. E., & Galinsky, A. D. (2003). From self-prediction to self-defeat: Behavioral forecasting, self-fulfilling prophecies, and the effect of com- petitive expectations. Journal of Personality and Social Psychology, 85, 672–683. doi:10.1037/0022-3514.85.4.672

Eisenberg, T., Garvey, S. P., & Wells, M. T. (1998). But was he sorry? The role of remorse in capital sentencing. Cornell Law Review, 83, 1599–1637.

Goffman, E. (1955). On face-work. Psychiatry: Journal for the Study of Interpersonal Processes, 18, 213–231.

Gordon, M. E., Slade, L. A., & Schmitt, N. (1986). The “science of the sophomore” revis- ited: From conjecture to empiricism. Academy of Management Review, 11, 191–207.

Gordon, R. M. (1986). Folk psychology as simulation. Mind and Language, 1, 158–171. Kahan, D. M. (1996). What do alternative sanctions mean? University of Chicago Law

Review, 63, 591–653. Kelley, H. H. (1971). Attribution in social interaction. In E. E. Jones, D. E. Kanouse, H. H.

Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior (pp. 1–26). Morristown, NJ: General Learning Press.

Latif, E. (2001). Apologetic justice: Evaluating apologies tailored toward legal solutions. Boston University Law Review, 81, 289–320.

Lazare, A. (2004). On apology. New York, NY: Oxford University Press.

Jehle et al. 745

Miller, D. T., Turnbull, W., & McFarland, C. (1990). Counterfactual thinking and social perception: Thinking about what might have been. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 305–331). San Diego, CA: Academic Press.

Risen, J. L., & Gilovich, T. (2007). Target and observer differences in the acceptance of questionable apologies. Journal of Personality and Social Psychology, 92, 418–433. doi:10.1037/0022-3514.92.3.418.

Rosenthal, R., Rosnow, R. L., & Rubin, D. B. (2000). Contrasts and effect sizes in behavioral research: A correlational approach. New York, NY: Cambridge University Press.

Rosnow, R. L., Rosenthal, R., & Rubin, D. B. (2000). Contrasts and correlations in effect- size estimation. Psychological Science, 11, 446–453. doi: 10.1111/1467-9280.00287

Received September 7, 2011 Accepted May 21, 2012

Copyright of Journal of Social Psychology is the property of Taylor & Francis Ltd and its content may not be

copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.