Running head: ANTICIPATED GROUP-BASED EMOTIONS 1 The Self-Regulatory Role of Anticipated Group-Based Shame and Guilt in Inhibiting Ingroup Favoritism Word count: 9,362 Running head: ANTICIPATED GROUP-BASED EMOTIONS 2 Abstract In three studies we examined whether the anticipation of group-based guilt and shame inhibits ingroup favoritism. In Studies 1 and 2 anticipated group-based shame negatively predicted ingroup favoritism; in neither study did anticipated group-based guilt uniquely predict ingroup favoritism. In Study 3 we orthogonally manipulated anticipated group-based shame and guilt. Here we found that the shame (but not the guilt) manipulation had a significant inhibitory effect on ingroup favoritism. Anticipated group-based shame (but not guilt) promotes egalitarian intergroup behavior. Keywords: Group-based guilt, group-based shame, ingroup favoritism, anticipated emotion Running head: ANTICIPATED GROUP-BASED EMOTIONS 3 The Self-Regulatory Role of Anticipated Group-Based Shame and Guilt in Inhibiting Ingroup Favoritism Given the countless incidents of group oppression, discrimination and genocide that have taken place in human history, it would be easy to conclude that group membership promotes selfish, hostile and immoral behavior. Research with minimal groups showing that merely being categorized into abstract, meaningless groups can evoke ingroup favoritism (Tajfel, Flament, Billig, & Bundy, 1971) is widely regarded as supporting this conclusion. In the present paper we contribute to the small but growing body of literature suggesting that groups are motivated to act morally. We argue that the anticipation of aversive group-based emotions can, under certain conditions, attenuate ingroup favoritism. There is a growing literature on the role of emotions in regulating intergroup behavior (e.g. Devine, Plant, & Buswell, 2000; Maitner, Mackie, & Smith, 2007). Researchers have found that experiencing interpersonal guilt for discriminatory behavior reduces such behaviors in the future (Amodio, Devine, & Harmon-Jones, 2007; Devine & Monteith, 1993; Monteith, 1993), and that group-based guilt serves a self-regulatory function (Maitner, Mackie, & Smith, 2006). Moreover, recent research suggests that anticipated emotions (such as guilt and shame) regulate moral behavior (Baumeister, Vohs, DeWall, & Zhang, 2007; Damasio, 1994; Haidt, 2001, 2003, 2007). These researchers suggest that when people anticipate feeling guilt or shame, they are likely to inhibit the behavior in question in order to avoid the aversive consequences of these emotions. We extend this hypothesis by applying it to group-based emotions. Previous research has demonstrated that immoral ingroup behavior elicits group-based guilt and shame (Doosje, Branscombe, Spears, & Manstead, 1998; Iyer, Schmader, & Lickel, 2007; Lickel, Schmader, Curtis, Scarnier, & Ames, 2005). We argue that the desire to avoid the aversive consequences of these group-based emotions motivates ingroup members to inhibit immoral intergroup behavior. When people predict (or anticipate) Running head: ANTICIPATED GROUP-BASED EMOTIONS 4 that an ingroup action would elicit aversive group-based emotions, they should be proactively motivated to inhibit the behavior in question in order to avoid the emotions and their aversive consequences. The aim of the present research was to assess the role of anticipated group- based guilt and shame in inhibiting ingroup favoritism. The Emotional Consequences of Illegitimate Ingroup Actions Guilt and shame about the negative actions or attributes of one’s group may be experienced through association with the ingroup even in the absence of personal responsibility for these actions or attributes (Brown & Cehajic, 2008; Doosje et al., 1998). These group-based emotions are evoked when members of an advantaged group believe that their privileged ingroup status was achieved through illegitimate actions for which their group is responsible (Branscombe, Doosje, & McGarty, 2003; Iyer et al., 2007; Leach, Snider, & Iyer, 2002). The ingroup members’ interpretation of an illegitimate action determines whether group-based guilt or shame is elicited. Although researchers debate whether shame stems from actions implying a global (e.g., Lewis, 1971; Niedenthal, Tangney, & Gavanski, 1994; Tangney & Dearing, 2002) or a specific lapse of one’s identity and reputation (e.g., Gausel & Leach, 2011; Gausel, Leach, Vignoles, & Brown, 2012), there is consensus that the focus is on one’s identity, rather than the behavior (Allpress, Barlow, Brown, & Louis, 2010; Ferguson, Burgman, White, & Eyre, 2007). Similarly, believing that an ingroup transgression threatens the ingroup’s moral identity should evoke group-based shame (Brown, Gonzalez, Zagefka, Manzi, & Cehajic, 2008; Lickel et al., 2005). In the case of guilt, people focus on the immoral action rather than the self, and feel bad because they committed a transgression (‘I did a bad thing;’ Niedenthal et al., 1994). Similarly, appraising the ingroup as responsible for a controllable immoral action should elicit group-based guilt (Iyer, Leach, & Crosby, 2003; Lickel, Schmader, & Barquissau, 2004). Running head: ANTICIPATED GROUP-BASED EMOTIONS 5 Group-based guilt and shame have aversive consequences for ingroup members through the threat that these emotions pose to the group’s identity (Branscombe, Ellemers, Spears, & Doosje, 1999). For these emotions to arise people must believe that their group was responsible for an immoral action. Accepting this appraisal associates the ingroup with a transgression (Doosje et al., 1998). Group-based shame poses an additional threat to social identity because people presumably believe that the transgression tarnishes their image (Johns, Schmader, & Lickel, 2005; Lickel et al., 2005; Lickel, Schmader, & Spanovic, 2007). Social identity theory (Tajfel & Turner, 1979, 1986) postulates that people are motivated to maintain a positive group identity, and ingroup members will therefore try to protect their group’s image by means of various ‘identity management strategies’ (Ellemers, Wilke, & van Knippenberg, 1993; van Knippenberg, 1989). These strategies serve to defend the ingroup’s identity by reassigning responsibility for the negative event or legitimizing the ingroup’s actions. For example, group identity might be protected by externalizing the blame (Doosje & Branscombe, 2003; Zebel, 2005), dehumanizing the victim (Castano & Giner-Sorolla, 2006; Imhoff & Banse, 2009; Zebel, Zimmermann, Viki, & Doosje, 2008), or denying the credibility of the source (Doosje, Branscombe, Spears, & Manstead, 2006). These identity management strategies are post hoc methods of avoiding aversive group-based emotions. Ingroup members employ these strategies after the incident has taken place. Identity management strategies may also be used to avoid these emotions in an anticipatory fashion, before an incident has taken place. For example, the use of propaganda to dehumanize an outgroup can alleviate the need to act morally towards this group (Bar-Tal, 1990; Staub, 1989), thereby preventing group-based guilt and shame for any future transgressions. People may also use pre-emptive pro-social strategies to avoid these aversive group-based emotions and the threat that they pose to social identity (Shepherd, Spears, & Manstead, 2013a, 2013b). One such strategy is not to commit the transgression in the first Running head: ANTICIPATED GROUP-BASED EMOTIONS 6 place. Indeed, recent research has found that the anticipation of aversive group-based emotions promotes collective action against a proposed ingroup transgression (Shepherd et al., 2013a) and moderates the amount of ingroup favoritism exhibited by member of high- status groups in stable hierarchies (Shepherd et al., 2013b). Therefore, when ingroup members find themselves in a situation in which they can achieve an advantage over an outgroup by illegitimate means (such as discrimination or ingroup favoritism), the anticipation of aversive group-based emotions should signal that undertaking the immoral action would result in aversive emotions that might threaten the ingroup’s identity, thereby motivating group members to inhibit transgressions. Although both group-based guilt and shame threaten social identity, shame is more damaging because it implies that the behavior is seen as a reflection of the immoral character of the ingroup rather than of a specific action. As a result, group-based shame is more closely related to the valence of the ingroup’s identity or reputation than guilt (Johns et al., 2005; Jones et al., 2009; Lickel et al., 2005, 2007). Anticipated group-based shame should therefore be a stronger signal of social identity threats than guilt. Because social identity theory suggests that the valence of the ingroup’s identity is a primary concern for ingroup members, anticipated group-based shame is more likely than guilt to be a negative predictor of ingroup favoritism. In line with this, previous research has found that anticipated group-based shame (but not guilt) promotes collective action against an impending ingroup transgression (Shepherd et al., 2013a). Moreover, anticipated group-based shame (but not guilt) moderates ingroup favoritism in stable high-status groups but not unstable high-status or (stable or unstable) low-status groups (Shepherd et al., 2013b, Study 1). Although the above research is promising, it could be argued that the inhibitory role of anticipated group-based shame and guilt has only be demonstrated in groups with superior status over an outgroup. Indeed, Shepherd et al. (2013b) found that anticipated group-based Running head: ANTICIPATED GROUP-BASED EMOTIONS 7 shame only moderated ingroup favoritism in stable high status groups. Although status was not manipulated or measured in Shepherd et al. (2013a), the fact that the ingroup was able to perpetrate an aggressive transgression implies that the status and/or power of the ingroup was greater than that of the victim group. This raises the question whether a status differentiation is needed for anticipated group-based emotions to self-regulate the actions of ingroup members. The present study enhanced previous research (Shepherd et al., 2013a, 2013b) by assessing the role self-regulatory role of anticipated group-based shame and guilt on ingroup favoritism when there was not a salient status difference between the ingroup and the outgroup. The present study also tested whether other factors could account for the self- regulatory role of anticipated group-based shame, such as social dominance orientation (SDO) or ingroup identitfication. SDO reflects people’s tendency to support group hierarchies and to believe that it is legitimate for some groups to dominate others (Sidanius & Pratto, 1999). It has been found that SDO is positively related to the justification of immoral ingroup actions (Leidner, Castano, Zaiser, & Giner-Sorolla, 2010) and ingroup favoritism (Sidanius, Haley, Molina, & Pratto, 2007; but see Reynolds, Turner, Haslam, Ryan, Buzumic, & Subasic, 2007). Thus people high in SDO might justify ingroup favoritism, reducing their likelihood of anticipated group-based shame and increasing their likelihood of exhibiting ingroup bias. Similarly, under certain conditions, high identifiers are more likely than low identifiers to discriminate against an outgroup (Tajfel & Turner, 1979, 1986; for an overview, see Turner, 1999) and to legitimize an ingroup transgression in order to avoid group-based shame (Johns et al., 2005). Ingroup identification may therefore negatively predict anticipated group-based shame and positively predict ingroup favoritism. It is possible that the negative relationship between anticipated group-based shame and ingroup favoritism may therefore be due to the fact that both variables are related to SDO and/or ingroup identification. The Running head: ANTICIPATED GROUP-BASED EMOTIONS 8 negative relationship of SDO and/or ingroup identification and anticipated group-based shame in combination with the positive relationships of SDO and/or identification with ingroup favoritism might have given rise to the negative relationship between shame and ingroup favoritism. The present studies therefore enhanced previous research by assessing whether anticipated group-based shame predicted ingroup favoritism after controlling for SDO (Study 1) and ingroup identification (Studies 2 and 3). Study 1 In Study 1 we assessed the relationship of anticipated group-based guilt and shame to ingroup favoritism. This study used an inter-university intergroup context. Participants were asked to rate the extent to which they anticipated feeling group-based guilt and shame if their own university were to discriminate against a rival university. Participants then distributed research funding to anonymous applicants from their own and the rival university. According to university league tables the rival university was of a similar status to the ingroup university. Importantly, the league table position of the universities was not made salient to the participants at any time throughout the study. Moreover, in Study 1 we measured SDO in order to check the extent to which the predicted negative relationship between anticipated group-based shame and ingroup favoritism, if obtained, was due to this factor. Method Participants and Design A total of 68 undergraduate students (53 women and 15 men) participated in this study in exchange for course credit or £3.00 (approximately $4.80). The age range of respondents was 18-37 years, with a mean age of 20.18. The predictor variables were anticipated group-based guilt and shame1. The outcome variable was ingroup favoritism. This outcome variable was measured using the so called ‘Tajfel matrices’ (see Tajfel et al., 1971). Ingroup favoritism was measured using the pull score of FAV (maximum differentiation and Running head: ANTICIPATED GROUP-BASED EMOTIONS 9 maximizing ingroup profit) against parity (P). The pull score was measured and calculated using the procedure outlined by Bourhis and colleagues (Bourhis, Sachdev, & Gagnon, 1994). Pull scores can range from -12 to 12. Greater positive FAV on P values indicate a stronger preference for ingroup favoritism over parity and greater negative values demonstrate stronger outgroup favoritism. Values of zero demonstrate that the participant distributed the resources equally between the ingroup and the outgroup. Materials and Procedure Anticipated group-based emotions. After giving consent, participants were informed that this research was investigating the attitudes of students at different universities and that a similar study was being conducted at the rival university. This information was included to strengthen the participants’ belief that at the end of the study the resources would be allocated between the ingroup and the outgroup. Participants were then asked to rate the extent to which they anticipated experiencing group-based guilt and shame if their own university were to discriminate against the rival university. Anticipated group-based guilt and shame were assessed using two scales adapted from Schmader and Lickel (2006). The guilt items were: ‘guilty,’ ‘regret,’ ‘sorry,’ and ‘remorse’ (α = .84). The shame items were ‘ashamed,’ ‘humiliated,’ ‘disgraced,’ and ‘embarrassed’ (α = .87). Four positive (or at least non- negative) anticipated group-based emotion items were also rated: ‘unconcerned,’ ‘confident,’ ‘apathetic,’ and ‘indifferent.’ Participants were asked ‘If [ingroup] University students were to discriminate against students from the [outgroup] University in some way, to what extent would you feel [emotion word]?’ All items were rated on a 7-point scale, ranging from not at all (0) to extremely intensely (6). Ingroup Favoritism. Participants were then informed that the people responsible for conducting this research wanted to know how people thought research funding should be distributed between academics at the ingroup and outgroup university. Participants were told Running head: ANTICIPATED GROUP-BASED EMOTIONS 10 that the grants were distributed in the form of credits. The more credits an applicant had, the more money they would receive. Participants were then asked to distribute credits between anonymous academics at the ingroup and outgroup university. SDO. Participants also completed an SDO measure. This was adapted from the SDO6 (Sidanius & Pratto, 1999) and contained 8-items (e.g., ‘In getting what you want, it is sometimes necessary to use force against other groups’; α = .83). Participants rated the extent to which they agreed with each of these items on a 7-point scale (1= ‘strongly disagree,’ 7 = ‘strongly agree’). When the participant had completed this scale they were thanked and debriefed. Results Anticipated Group-Based Guilt and Shame Confirmatory factor analysis was conducted to test whether anticipated group-based guilt and shame were separate constructs. This analysis was conducted using AMOS 19 (Arbuckle, 2010). The model tests were based on the covariance matrix and maximum likelihood was used as the method of estimation. The two-factor hypothesized model provided an adequate fit to the data, χ²(19) = 27.65, p = .090. This was confirmed by the other fit indices: goodness-of-fit index (GFI) = .91, comparative fit index (CFI) = .97, normed fit index (NFI) = .91, and root mean square error of approximation (RMSEA) = .082. The alternative single-factor solution did not fit the data well: χ²(20) = 52.58, p < .001, GFI = .83, CFI = .89, NFI = .84, and RMSEA = .156. Importantly, the two-factor solution provided a significantly better fit to the data than the single factor solution, χ²(1) = 24.93, p < .001. We therefore concluded that although anticipated group-based shame and guilt were strongly correlated (see Table 1), they were empirically distinct constructs. The mean levels of anticipated group-based shame (M = 2.79, SD = 1.32) and guilt (M = 2.82, SD = 1.28) were significantly greater than zero, t(67) = 17.44, p < .001 and t(67)