Differentiated Intergroup Emotions and Ingroup Identiﬁ Cation in Soccer Fans

Group Processes & Intergroup Relations 2007 Vol 10(1) 9–26

Seeing Red or Feeling Blue: Differentiated Intergroup Emotions and Ingroup Identiﬁ cation in Soccer Fans

Richard J. Crisp and Sarah Heuston University of Birmingham Matthew J. Farr Loughborough University Rhiannon N. Turner University of Birmingham

In this study we examined how social identity threat, contextualized as soccer fans’ reactions to their team’s success or failure, can lead to differentiated emotional expression as a function of ingroup identifi cation. We predicted that negative responses to threat (a team losing a match) would be qualitatively differentiated for lower and higher ingroup identifi ers in terms of both emotions and action tendencies. English male soccer fans were tested in three sessions (following matches resulting in two losses and one win). The fi ndings supported the hypotheses: following match losses lower identifi ers felt sad but not angry, whereas higher identifi ers felt angry but not sad. These qualitatively different negative emotional reactions to match loss mediated reported action tendencies. We discuss how these fi ndings support intergroup emotions theory and the predictive utility of social identifi cation in discerning differentiated emotional and behavioral reactions to intergroup threat. keywords intergroup emotions, social identity

For large numbers of people across the globe, risen to new heights. Spectator attendances attachment to their soccer club—whether it is at soccer matches in England are now at the smallest local amateur side or one of the their highest for many years (reaching nearly superpower teams with their galáctico players— 30 million in the 2003/4 season; by comparison, is something of considerable psychological in 1988/9, this ﬁ gure was around 18 million). signiﬁ cance. FIFA, soccer’s world governing body, appositely paints a vibrant picture of the Author’s note game as ‘physical movement that simultaneously Address correspondence to R. J. Crisp, School moves the emotions’ (Fédération Internationale of Psychology, University of Birmingham, de Football Association, n.d.). Truly the world’s Edgbaston, Birmingham B15 2TT, UK game, the global popularity of the sport has [email: [email protected]]

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1)

Unfortunately, however, intergroup confl ict and ingroup-threatening events on the pitch can social disorder remain an inescapable element affect emotions. Such emotions in turn affect of this pastime enjoyed by millions. Within the the likelihood of specifi c action tendencies last decade, research has suggested that vio- such as confrontation versus avoidance, with lence arises at around 10% of soccer matches confrontation being a potential precursor to in England (Marsh, Fox, Carnibella, McCann, social disorder. In this article, using intergroup & Marsh, 1996). Furthermore, social disorder emotions theory (Mackie, Devos, & Smith, 2000; has not been limited to being a solely domestic Smith, 1993, 1999), we conceptualize soccer- concern. The problems posed by ‘hooliganism’ related social identity processes in terms of and widespread social unrest at high-profi le differing emotional responses to an ingroup international tournaments (particularly within threat (for example, the team’s defeat), which the European Union), such as the 1998 FIFA may occur in terms of differing levels of World Cup in France, and UEFA Euro 2000 commitment (i.e. social identifi cation) to that in the Netherlands and Belgium, have proved ingroup. to be of a suffi cient magnitude to dominate the headlines, both in Britain and the wider Social identity and sporting contexts international community. Fortunately, recent work by social psychologists The social identity approach (by this we mean has demonstrated that the prognosis for the social identity theory, Tajfel & Turner, 1986, game, and the English national identity abroad, and self-categorization theory, Turner, Hogg, need not be bleak. Greater understanding of the Oakes, Reicher, & Wetherell, 1987) provides a processes which govern intergroup dynamics, logical and coherent account of the processes as understood through the traditional social which lead individuals to connect themselves identity theory framework (Tajfel & Turner, to groups, and what this connection—that is, 1979), has been making a tangible impact in social identifi cation—means for the individuals terms of reducing soccer-related social disorder concerned. In particular, because of a motivational (see Stott & Adang, 2004). The contribution of drive toward self- and group-enhancement (e.g. social psychologists to analyses of fan behavior Tajfel & Turner, 1986, see also Abrams & Hogg, and the policing of matches helped to ensure 1998; Hogg & Abrams, 1988), we can consider that the recent UEFA Euro 2004 tournament was ingroups by their very nature to be of emotional the fi rst in recent memory to take place free signifi cance to the self-concept. The sporting from signifi cant incidences of soccer-related world provides many sources for strong and disorder. enduring social identifi cation. In spectator sport, Social identity approaches, and in particular, the personal psychological lives of individuals the elaborated social identity model of crowd can be tied to the fates of their chosen teams. behavior (Reicher, 1996) are useful in explaining, Cialdini and colleagues’ (1976) classic study and helping to prevent, social disorder which of ‘basking in refl ected glory’ demonstrated arises as a consequence of intergroup processes the simple power of winning and losing upon (e.g. how opposing fans, and police, react to each collective self-esteem, a vivid illustration of the other as a function of their ingroup–outgroup grip that meaningful social memberships can category representations). There has been exert over self-perception. In this study more valuable progress made with respect to our under- students demonstrated their affi liation with standing of processes of self-categorization, their university after their university team won perceptions of (il)legitimacy of actions, and a football game than after losing a game. Their the function of norms (e.g. Stott & Drury, 2000; fi ndings also hinted at the strategy of ‘cutting Stott & Reicher, 1998a, 1998b; Stott, Hutchison, off refl ected failure’, later elucidated by Snyder, & Drury, 2001). Attention has not, however, Lassegard, and Ford (1986) and Wann and been focused on the ways in which particular Branscombe (1990); following group failure

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion affiliation is de-emphasized, and members self as indivisible from the social context), the distance themselves from the group. Both theory addresses those emotions people ex- strategies appear to be means of managing self- perience when they think about or confront esteem; either by capitalizing on membership members of an outgroup. of a successful group, or by preventing oneself from being tarnished by affiliation with an Intergroup emotions unsuccessful group. This apparent link between sports team affi li- Intergroup emotions theory (Mackie et al., 2000; ation and self-esteem was subsequently confi rmed Smith, 1993, 1999) builds on a combination of by Hirt, Zillmann, Erickson, and Kennedy two social psychological perspectives; appraisal (1992). They showed that fans who considered theories of emotion (Frijda, 1986), which their team allegiance to be an important part argue that emotions follow from events related of their identity considered their team’s success to the individual or the personal self, and and failure as constituting personal success and social identity theory (Tajfel & Turner, 1979), failure. In particular, a team’s poor performance which posits that when group membership is could markedly lower supporters’ self-evaluations salient it constitutes an integral part of the self, and mood. Recent studies have also suggested whereby individuals come to regard themselves a link between high levels of identifi cation and as relatively interchangeable with other group mood, as well as a link to aggressive tendencies. members. The theory argues that when group Wann, Dolan, McGeorge, and Allison (1994, membership is salient, people react emotionally Study 2) had college students attend their local when situations or events affect the ingroup. basketball teams’ home games and complete Smith (1993) argued that specifi c appraisals measures of identifi cation and affect. They made during experience with the outgroup found that following a loss, high identifi ers would lead to specifi c emotions and subsequent showed a greater increase in negative mood action tendencies. If an outgroup was seen as and a greater decrease in positive mood than pushy, for example, this might elicit anger, low identifi ers. Highly identifi ed fans have also leading to a tendency to aggress against the been shown to report higher levels of hostile group, or if an outgroup was seen as dirty, this and instrumental aggression (Wann, Carlson, & might elicit disgust and lead to avoidance of Schrader, 1999), and to be more likely to injure that group. some member of a rival team (Wann, Hunter, Mackie et al. (2000) investigated the effect Ryan, & Wright, 2001; Wann, Peterson, Cothran, of the strength of the ingroup relative to the & Dykes, 1999). outgroup on different emotions and action Beyond such generalized self- and group- tendencies. Specifi cally, they proposed that based affect, research in the social identity fi eld when the ingroup is in a strong position relative to has not prioritized the decomposition of such the outgroup, this should elicit anger, leading to affective responses. Whereas feelings toward offensive action tendencies toward the outgroup. an outgroup can be summarized as constituting In contrast, when the ingroup is in a weak position negative affect (put simply, ‘bad feeling’), we relative to the outgroup, this should elicit fear, can actually go further in exploring the specifi c resulting in avoidance of the outgroup. Across emotional components which constitute such three studies, participants categorized them- ‘bad feelings’. Such a focus on specifi c emotions, selves into one of two groups depending on their which provide a qualitatively and experientially position on an emotive issue (e.g. punishments richer understanding of intergroup behavior, for drug use, equal rights for homosexual forms the central pillar of intergroup emotions couples). The degree of support for the point theory (Mackie et al., 2000; Smith, 1993, 1999). of view of the ingroup relative to the outgroup Building upon the basic principles which was then measured (Study 1) or manipulated underlie the social identity approach (i.e. the (Studies 2 and 3). Mackie et al. (2000)

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) found when the ingroup position was stronger and so on? Through considering an ingroup- than the outgroup position, more anger was threatening event, individuals can be thought elicited and there was a greater inclination to to experience an affective response which con- oppose or confront the outgroup. Crucially, stitutes an emotion directed toward ‘them’, the effect on approach action tendencies was the outgroup, on the basis of a dichotomized, mediated by anger. There was no evidence that intergroup frame of reference. Specifi cally, and being a member of a weak group led to avoidance of importance for the study we report here, action tendencies via fear. There was, however, research on intergroup emotions theory supports some limited evidence (Study 3) that being in the notion that in intergroup contexts specifi c the weak group led to feelings of contempt, emotions can be experienced that are linked which was associated with the desire to move directly to group membership, that qualita- away from the outgroup. tively different emotions will be experienced as Silver, Miller, Mackie, & Smith (2001; cited a function of context, and that these different in Devos, Silver, Mackie, & Smith, 2002) found emotions will lead to different behavioral greater evidence for the role of being in a weak (action) tendencies. In the research we report, group position on fear and avoidance action we tested a novel moderator of experienced tendencies. Female undergraduates were asked emotion that was derived directly from work on to imagine themselves walking down a local social identity theory—perceivers’ degree of street late at night and being unjustly accused ingroup identifi cation. of breaking a car’s side mirror. To induce appraisals that might lead to fear, the male This research owner of the car was described as threatening and aggressive. Silver et al. (2001) found that We argue that as well as existing moderators participants appraised themselves as weak in that lead to qualitatively different intergroup the situation, more so when they were the emotions and action tendencies (Mackie et al., sole ingroup member encountering multiple 2000; Silver et al., 2001) we can specify a further outgroup members. They also showed high moderator derived directly from work on social levels of fear and a desire to move away from identity theory. We possess multiple social iden- the outgroup. The weaker participants felt, the tities, which vary in the degree to which they are more fear they experienced and the more they salient (or not) at particular times (for a review, wanted to move away from the antagonist. The see Crisp & Hewstone, 2006). Put another way, effect of an appraisal of weakness on avoidance we can identify to a greater or lesser extent with action tendencies was mediated by participants’ a range of social categories at different times. reported levels of fear. Ingroup identifi cation can be defi ned as ‘the In sum, group-based emotions—for example, extent to which group members feel strong ties shared fear, sadness, anger, guilt—arise when a with their group’ (Spears, Doosje, & Ellemers, particular intergroup event is weighed up in 1999, p. 85). Consistent with the above, it is terms of its implications for the ingroup. One can well established that such ingroup identifi ca- see how this basic idea makes a lot of sense in all tion is variable (e.g. Branscombe, Wann, Noel, sorts of intergroup contexts. For instance, when & Coleman, 1993; Doosje, Ellemers, & Spears, the news emerged that three of the perpetrators 1995; for reviews see Brown & Williams, 1984; of the July 7, 2005 terrorist attacks on London Hinkle & Brown, 1990), and this variability in came from the English city of Leeds, residents of social identifi cation has strong predictive validity Leeds were forced to consider the implications when combined with value or distinctiveness of this for their own group memberships (e.g. threat. This is the case with respect to a BBC News Online, 2005). That is, how does this range of intergroup attitudes and behaviors event impact upon the notion of ‘us’ as citizens ranging from self-stereotyping (e.g. Spears, of Leeds, English people, Muslims, non-Muslims, Doosje & Ellemers, 1997), through stereotypic

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion differentiation (Doosje et al., 1995), perceived Although not in a sporting context, some group variability (Ellemers, Spears, & Doosje, studies have also indicated that identifi cation 1997) and intergroup bias (e.g. Branscombe & may not only affect emotions, but also associated Wann, 1994; Jetten, Spears, & Postmes, 2004; action tendencies. Recategorization from two Spears, Jetten, & Scheepers, 2002). subgroups into a merged superordinate group As we noted earlier, some work has already can constitute a threat to identity, especially established ingroup identifi cation as an important for highly identifying subgroup members predictor of generalized affective reactions in (see Crisp, 2006; Crisp, Stone, & Hall, 2006). sporting contexts. Wann et al. (1994) found that Yzerbyt, Dumont, Wigboldus, and Gordijn following a loss, high identifi ers showed a greater (2003) examined emotional reactions and action increase in negative mood and a greater decrease tendencies in response to recategorization as a in positive mood than low identifi ers. Although function of identifi cation. They asked French- these fi ndings only show that identifi cation had speaking Belgian participants enrolled at the an impact on broad negative or positive affect, Catholic University of Louvain (UCL) to read Wann et al.’s mood questionnaire included both a bogus newspaper story about how English items that referred to anger emotions (e.g. was being enforced as the sole language in angry) and depressed emotions (e.g. sad). the third, fourth and fi fth years of University. This raises the possibility that had these specifi c Participants either had their common identity negative emotions been differentiated then a (recategorization) with these students made dissociation may have been observed for lower salient (they were told they were being compared, and higher identifi ers. as students, with professors), or their subgroup Some further evidence that specifi c emotions membership was made salient (they were told can be elicited as a function of identifi cation that UCL students were being compared with comes from Sloan (1989). In this study 46 students at other universities). After reading regular fans of a basketball team (who could the newspaper article, participants reported the be considered committed, high identifi ers) extent to which they felt anger, sadness, fear and recorded their emotions before and after eight happiness, and reported their action tendencies; home basketball games. Sloan found that anger whether they intended to exhibit approach increased after a loss, but was almost nonexistent behaviors (e.g. intervene, get angry), do nothing, after a win. In contrast, happiness decreased after or avoid the situation (e.g. stop thinking about a loss and rose following a win. These fi ndings it). Yzerbyt and colleagues found that in the were also replicated among American football common ingroup condition high identifi ers fans. In contrast, fans watching an amateur showed signifi cantly more anger and a greater boxing tournament, where the competitors intention to engage in approach behaviors than had only recently volunteered, and therefore low identifi ers, who were marginally more likely where there had been little opportunity for fans than high identifi ers to have a tendency to avoid to form an attachment (i.e. participants could the situation. be regarded as lower identifi ers), showed no There is therefore indicative evidence that changes in anger or happiness across the course identifi cation has a role to play in predicting of the matches. Sloan’s study is important in that emotional reactions to ingroup threat and asso- it indicates that the degree of commitment to ciated action tendencies. However, although a sports team can have an impact on specifi c some of the above studies have focused in on emotions (anger and happiness). However, specifi c emotions (e.g. anger), none have shown because the two emotions used here are still a dissociation of specifi c emotions that cannot broadly positive versus negative, we cannot be be attributed to broadly affectively positive or sure that the results simply refl ect differences negative reactions. In other words, to be sure in generalized affective reaction, rather than that qualitatively different reactions are being being attributable to qualitatively different experienced in some context it is necessary to emotional reactions. show that two similarly valenced, but distinct

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) emotions can be experienced, with associated Harvey, 1999). Arguably these examples of (but again distinct) action tendencies. people identifying with groups as a way of dealing Here we extend theorizing in both the social with threat suggest that for such perceivers the identity and intergroup emotion domains by ingroup must be seen as a source of strength. predicting qualitatively different emotional re- Put another way, groups are seen as stronger, actions to intergroup threat and corresponding more potent, and better able to serve as a viable differentiation in terms of action tendencies. source of positive self-esteem to the extent that We make our predictions with respect to two one identifi es with the group. negative, but qualitatively different, relevant We may therefore expect, as with higher per- emotions in the context of sporting team suc- ceived group strength, that higher identifi ers cess or failure. Compared to team success, we will be more inclined (than lower identifi ers) expected the degree of ingroup identifi cation to feel angry following a threat, and to be more to predict specifi c experienced emotions, and inclined to report tendencies toward approach, for this to be particularly the case following a aggression-related behaviors (consistent with loss, an intergroup threat in this context. Mackie et al., 2000). As we have noted, higher We expected identifi cation to predict qualita- identifi ers are more likely to be ingroup protect- tively different emotions due to different ap- ive under threat, showing bolstering behaviors praisals of the general context at hand (i.e. a team like stereotyping, ingroup homogenization, loss). Above we discussed how the perception of self-stereotyping and bias. Approach-type be- the ingroup as being in either a strong or weak haviors, and by association the experience of position can lead to different appraisals of the angry emotions, are consistent with these general same situation, and correspondingly different trends. Indeed, Wann and Branscombe (1990) emotions and action tendencies (Mackie et al., found that highly identifi ed sports team fans 2000; Silver et al., 2001). Specifi cally, when the were less likely to Cut Off Refl ected Failure ingroup is in a strong position relative to the (CORF) than low identifi ers, arguing that they outgroup, this should elicit approach-related instead use other coping mechanisms when emotions, leading to offensive action tendencies they lose, such as expressing anger, hostility toward the outgroup. In contrast, when the and aggression. ingroup is in a weaker position relative to Anger and approach behaviors therefore the outgroup, this should elicit withdrawal-type seem consistent with what we already know about emotions, resulting in avoidance of the outgroup. how high identifi ers react to ingroup threat. For several reasons we believe it is reasonable But what about lower identifi ers? Typically we to equate perceiving either a strong or weak might expect lower identifi ers to simply exhibit position relative to the outgroup with ingroup less intense behaviors than higher identifi ers identification. According to social identity (their identity is less bound up with their group’s theory’s self-esteem hypothesis (Abrams & Hogg, successes or failures). They are less likely to react 1988), people can use group membership as with ingroup-bolstering or protective behaviors a resource—as a means of acquiring positive than higher identifi ers (e.g. ingroup favoritism) self-esteem. This tendency appears to vary and are more likely to try alternative self-esteem as a function of identification. Martinot, maintenance/restoration techniques, such as Redersdorff, Guimond, and Dif (2002) found dis-identifying and individuating themselves that identifi cation mediated the extent to which (Ellemers, van Knippenberg, & Wilke, 1990; group members self-categorized to protect Ellemers, Wilke, & van Knippenberg, 1993; themselves from threatening comparisons Martinot et al., 2002). Based on what we know (and retain positive self-esteem, Redersdorff about lower identifi ers in general we might & Martinot, 2003). Similarly, the negative therefore not predict any qualitative difference impact of perceiving oneself to be victim of in specifi c emotion experienced, but rather discrimination can be alleviated by higher simply a less intense reaction to that of higher ingroup identifi cation (Branscombe, Schmitt, & identifi ers.

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion

In the context of sports team allegiance, how- a weaker position predicts avoidance action ever, there is a critical difference to much of tendencies and related emotions (Mackie et al., the existing work on the moderating impact of 2000; Silver et al., 2001). Here we expect not ingroup identifi cation. That is that members in fear, but a different negative emotion, sadness, these contexts have chosen their allegiance to a to evoke avoidance. We would not expect fear to particular identity, and, as fans, should show a be particularly applicable in a sports-fan context general positively skewed level of identifi cation because it is an emotion that is typically elicited with the team. That is, sports fans should all be when thinking about future negative events: Here, fairly committed to the group. In this context, sports fans are appraising a negative event that the terms lower and higher identifi cation take on has happened in the past—a match loss. From a new meaning, but a meaning that is consistent research on counterfactual thinking we know with our prediction of a qualitative dissociation that thinking about negative past events can elicit of specifi c emotions. negative affect (Roese, 1997), and, in particular, While typically we might regard people who this negative affect is characterized by sadness are lower identifi ers with groups to simply feel (Mandel, 2003). The idea that sadness should less intensely about threats to that identity, lead to avoidance is consistent with Neuberg here we might expect a qualitatively different and Cottrell (2002) who argue that sadness experience for people who all feel a negative is associated with self-contemplation and self- emotion. Importantly, given the skew toward high imposed seclusion (for a similar argument see commitment, we expect all fans will experience Crisp, Ensari, Hewstone, & Miller, 2002). This a negative emotion following a loss, but the type dissociation in predicted emotional reactions of emotion experienced will depend upon how is also consistent with what we know of lower committed a fan they are. Consistent with the identifi ers in threatening group contexts. It is fi ndings from Sloan (1989) this represents a lower identifi ers who will be most likely to ‘cut key conceptual distinction between affi liations off refl ected failure’ (Snyder et al., 1986), via that are freely chosen and developed, like sports avoidance arising from associated emotions team allegiance—identities that matter to all like sadness. While lower identifying fans are members of the group—and types of groups likely to feel negative about a loss like higher that are not chosen by participants (such as identifi ers, their reaction will be more associated gender and race) which can engender very low with distancing themselves, rather than engaging levels of commitment (indeed for such groups in ingroup-protective behavior, and this will be dis-identifi cation can be observed, when people refl ected in their action tendencies. try to distance themselves from what they might In sum, we argue that while all participants will perceive to be a devalued identity; Steele & experience negative affect following a loss (all Aronson, 1995). As such, the positive skew being fans with a baseline skew toward higher toward high identifi cation is representative of a levels of identification), the nature of how precondition required to observe the qualitative this negative affect is experienced and repre- differentiation of emotions we predict here. sented in behavioral tendencies will vary as a Given that sporting contexts are typically function of the degree of identifi cation above those where all members have a baseline level this baseline. Consistent with the above theory of commitment, we may therefore expect the and empirical fi ndings, lower identifying fans degree of identifi cation reported by fans to will react to the loss with CORF-like tendencies predict different forms of experienced negative such as avoidance, and they will experience affect. Lower identifying fans should feel negative avoidance-related emotions such as sadness. but they should feel specifi cally more sad—a As identifi cation rises for these fans we expect less action-oriented emotion—than higher more anger-related and aggressive approach identifi ers, and so be more inclined to want to tendencies to be exhibited, consistent with avoid the outgroup. We can equate this with what we know of the ingroup bolstering previous work which has found that being in reactions to threat observed in previous

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) studies. In the study reported below, we tested reactions to other groups. We were interested these predictions. in measuring two specifi c action tendencies; the tendency to move toward the outgroup Method and the tendency to move away from it, which we hypothesized would be associated with Participants and design anger and sadness respectively. The tendency Sixty male soccer supporters participated in to move toward the outgroup was measured this study; their ages ranged from 16 to 52 years. by two items asking participants to rate the All participants were fans of Northfi eld Town extent to which the result of the match made Football Club, a small amateur club in the city them want to: ‘Confront the opposition’s fans’, of Birmingham, England. Founded in 1966, and ‘Argue with the opposition’s fans’, which they play in the Midland Combination League. were signifi cantly correlated with one another Fans were sampled on three occasions (Ns = 16, (r(60) = .86, p < .0001). These items were 23 and 21), following one win (score: 1–0) and therefore averaged to form a single index of two losses (0–3; 0–2) for their team. As such, approach tendency. The tendency to avoid the win or loss constituted a dichotomous between- outgroup was measured by two items, asking subjects variable, and identifi cation a continuous participants to rate the extent to which the predictor for anger, sadness and happiness result of the match made them want to, ‘Avoid emotions, and approach and avoidance action the opposition’s fans’, and ‘Have nothing to do tendencies. with them’, which were also highly correlated (r(60) = .68, p < .0001). These two items were Dependent measures therefore averaged to form a single index of Intergroup emotion Our measure of intergroup avoidance. Participants responded to all four emotion (adapted from Mackie et al., 2000), was items on a 7-point Likert scale, ranging from 1 intended to measure participants’ emotional (not at all) to 7 (very much so). reactions (anger, sadness and happiness) to the results of each match. Participants were Procedure asked to state the result of the match before The questionnaire was designed following con- rating their emotions about the result using a sultation with a number of soccer fans about the 7-point Likert scale, ranging from 1 (not at all) nature of supporter behavior at soccer matches. to 7 (very much so). The emotions participants The study was carried out at the soccer team’s were asked to rate were: angry, hostile, annoyed, ground when home games were being played. irritated, furious (measuring anger; α = .967); sad, Sixty participants were tested over a three depressed, tense, distressed, upset (measuring sadness; match period, being told that the purpose of α = .947); and happy, pleased, cheerful, calm, elated the study was to ‘gain information about how (measuring happiness; α = .883). people think about their team for research into spectator behavior’. At the end of each soccer Action tendencies To measure the extent to which match, participants were approached by the different emotions would facilitate different experimenter as they left the soccer ground behaviors, we also adopted the measure of and asked if they would complete a voluntary action tendencies used by Mackie et al. (2000). questionnaire. An introductory page, covering These measures are useful because while issues of informed consent and confi dentiality, social psychologists typically focus on negative and giving a brief explanation of the research evaluations of a group and its members, such area was first presented. Participants were measures are limited because they do not explain then asked to complete items relating to ingroup the wide variety of reactions to the outgroup. identifi cation. These items asked participants Action tendencies, group members’ desire to to rate the extent to which they agreed with behave in a particular way toward the outgroup, the following four statements (adapted from may help us to understand a broader range of Jetten, Branscombe, Spears & McKimmie, 2003;

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion

Luhtanen & Crocker, 1992): ‘I identify strongly effects on the anger plus sadness index equate with other Northfi eld Town FC supporters’, to between-subjects effects observed in a typical ‘Being a Northfi eld Town FC supporter is an ANOVA. For our purposes of differentiating important part of who I am’, ‘I feel strong ties specifi c types of emotion, the former index is with other Northfi eld Town FC supporters’, most relevant here. The procedure involving ‘I feel a sense of solidarity with Northfi eld Town the anger minus sadness index creates what is FC supporters’ (1-not at all, 9-very much). These essentially a within-subjects factor of emotion items formed a reliable index (α = .976) so were (anger vs. sadness) and effects on this index averaged accordingly to form a single meas- indicate diverging trends to report anger and ure of identifi cation. Following this, participants sadness emotions respectively. Interactions can completed the intergroup emotions items, noting then be decomposed like any other higher-order fi rst their team’s fi nal score and the opposing ANOVA interaction by carrying out the separate team’s fi nal score (to reinforce the win-loss threat analyses for the two levels of the emotion factor context). Finally, the participants completed the (sadness and anger). action tendency items before being thanked This analysis revealed that, overall, as iden- and debriefed. tifi cation increased so too did general negative affect (anger plus sadness) (β = .262; t = 2.90, Results p = .005). In addition, negative affect (anger plus sadness) was greater following a loss than a win, We used moderated regression (Aiken & West, irrespective of degree of ingroup identifi cation 1991) to investigate the impact of a threat to the (β = –.713; t = –7.89, p < .0005). There was no ingroup (loss vs. win) on intergroup emotions result × identifi cation interaction on general and action tendencies as a function of ingroup negative affect (anger plus sadness) (β = .059; identifi cation. We computed an interaction t = .185, p = .854). These fi ndings confi rm that, variable by contrast coding our threat conditions overall, higher and lower identifi ers felt equally as 0 and 1 (loss vs. win) and multiplying it by the negative following a loss compared to a win. centered continuous identifi cation scores for These between-subjects effects (using the each participant. We then entered this interaction anger plus sadness index) were qualified, variable into a multiple regression on a second however, by several interactions that show a step following the insertion of the categorization dissociation of the different types of emotion and identifi cation factors independently at step (using the anger minus sadness index). There 1. Notably, mean identifi cation did not vary over was a result × emotion interaction (β = –.244; sampling sessions (F(2, 57) = .280, p = .757) t = –2.47, p = .017), and an identification × (Ms = 4.76, 5.17, 4.73) and across all sessions the emotion interaction (β = .648; t = 6.55, p < .0005). overall identifi cation mean was 4.86 (SD = 1.94) The latter indicated that degree of ingroup which was signifi cantly higher than the mid- identifi cation was positively related to anger point of the scale (4) (t(59) = 3.43, p = .001). (β = .506; t = 4.46, p < .0005), while degree of This confi rmed the expected positive skew of identifi cation was negatively related to sadness the identifi cation scores in this context. (β = –.321; t = –2.58, p = .012). Broadly, this supports the notion that ingroup identifi cation Intergroup emotions will differentiate intergroup emotions: at higher We first computed two indices: anger plus levels of identifi cation participants reported sadness, and anger minus sadness. This created more anger, and at lower levels of identifi cation the variables needed to test whether there was participants reported more sadness. Both of these an interaction between result, identifi cation effects were qualifi ed by the predicted result × and specifi c emotion. The anger minus sadness identifi cation × emotion interaction (β = –.985; index equates to entering anger and sadness as t = –3.06, p = .003). This interaction suggested different levels of an emotion within-subjects that participants reported qualitatively different factor in a typical analysis of variance (ANOVA); emotions depending on their degree of ingroup

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) identifi cation following losses versus wins. To result × identifi cation interaction (β = .813; decompose this interaction, we examined the t = 2.11, p = .039) (see Figure 2). Simple slope effects of result and identifi cation on anger and analysis within result conditions revealed sadness separately. that identification had a marginal positive relationship with sadness when the soccer Anger Moderated regression analysis revealed team had won the match (β = .440; t = 1.83, that as identifi cation increased so too did anger p = .088), but a signifi cant negative relationship (β = .566; t = 7.06, p < .0005). In addition, with sadness when the soccer team had lost the anger was greater following a loss than a win, match (β = –.401; t = –2.84, p = .007). The less irrespective of degree of ingroup identifi ca- the participants identifi ed with their team, the tion (β = –.620; t = –7.72, p < .0005). These more sadness they experienced following a effects were qualifi ed, however, by the predicted loss—a qualitatively different relationship than result × identifi cation interaction (β = –.559; that observed between identifi cation and anger, t = –2.05, p = .045) (see Figure 1). Simple which was a positive relationship.1 slope analysis within result condition revealed that identifi cation had a marginal effect on anger when the soccer team had won the match (β = .417; t = 1.72, p = .11), but was signifi cantly positively related to anger when the soccer team had lost the match (β = .720; t = 6.73, p < .0005). The higher the participants identifi ed with their team, the more anger they experienced following a loss.

Figure 2. Reported sadness as a function of threat and ingroup identiﬁ cation.

Action tendencies We first computed two indices: avoidance plus approach, and avoidance minus approach. This created the variables needed to test whether there was an interaction between result, identifi cation and action tendencies. This Figure 1. Reported anger as a function of threat and analysis revealed only two theoretically signifi - ingroup identifi cation. cant effects. First, there was an identifi cation × action tendency interaction (β = .581; t = 5.35, Sadness Moderated regression analysis revealed p < .0005). This interaction indicated that that as identifi cation increased sadness decreased while degree of ingroup identifi cation was posi- (β = –.281; t = –2.47, p = .02). In addition, tively related to approach tendencies (β = .579; sadness was greater following a loss than a win, t = 5.40, p < .0005), it was negatively related irrespective of degree of ingroup identifi cation to avoidance tendencies (β = –.289; t = –2.30, (β = –.413; t = –3.64, p = .001). These effects p = .025). This supports the basic hypothesis were qualified, however, by the predicted that degree of identifi cation will differentiate

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion action tendencies: higher identifi ers tended following a loss than a win, irrespective of toward approach behaviors, lower identifi ers degree of ingroup identification (β = .221; tended toward avoidance behaviors. t = 1.78, p = .080). These effects were, however, These effects were qualifi ed, however, by a qualifi ed by a result × identifi cation interaction result × identifi cation × action tendency inter- (β = .854; t = 2.03, p = .048) (see Figure 4). Simple action that approached signifi cance (β = –.635; slope analysis within result condition revealed t = –1.70, p = .094). This interaction suggested that identification had a marginal positive that participants reported qualitatively different relationship with avoidance when the team had action tendencies depending on their degree won the match, β = .429; t = 1.78, p = .098. When of ingroup identification following losses the soccer team had lost the match, however, versus wins. To decompose this interaction, we there was a signifi cant negative relationship examined the effects of result and identifi cation between identifi cation and avoidance, β = –.367; on the two action tendencies separately. t = –2.55, p = .014. Lower identifi ers were more likely to avoid the outgroup than high identifi ers Approach action tendency As identifi cation following a loss. increased so too did tendencies to approach the outgroup (β = .609; t = 6.08, p < .0005). In addition, tendencies to approach were greater following a loss than a win, irrespective of degree of ingroup identifi cation (β = –.313; t = 3.12, p = .003). These effects were not qualifi ed by a result × identifi cation interaction (β = –.129; t = –.366, p = .716), indicating that as identifi cation increased, so too did approach tendencies, both when the team had lost a match, but also when the team had won a match (see Figure 3).

Figure 4. Reported avoidance tendencies as a function of threat and ingroup identiﬁ cation.

Mediational analysis We computed mediational analyses to assess whether the relationship between ingroup identifi cation and the tendency to approach or avoid the outgroup could be explained by anger and sadness respectively following a match loss. Figure 3. Reported approach tendencies as a Regarding the effect of identifi cation on the function of threat and ingroup identifi cation. tendency to approach the outgroup, in step 1, the outcome measure (approach tendency) was Avoidance action tendency As identifi cation regressed onto the predictor (identifi cation). decreased tendencies to avoid the outgroup This analysis revealed a signifi cant relation- increased (β = –.268; t = –2.16, p = .035). In ship between the predictor and the outcome addition, there was a marginal tendency (β = .643; p < .0005). In step 2, the analysis to report more avoidance of the outgroup revealed a significant relationship between

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) identifi cation and the potential mediators, anger Discussion (β = .720; p < .0005) and sadness (β = –.401; p = .007). In step 3, approach behavior was In this research we set out to integrate work on regressed onto the mediators while controlling the distinctive reactions of lower and higher for identifi cation. This analysis revealed that identifi ers in social judgment with intergroup while there was a signifi cant relationship between emotions theory. In the context of soccer fans’ anger and approach tendency (β = .909; p < .0005), reactions to wins or losses, we measured negative, there was not between sadness and approach but qualitatively distinct, intergroup emotions tendency (β = –.057; p = .453). Finally, when the and action tendencies. Broadly, we found strong mediators were controlled in the fi rst equation, support for the overall dissociation of emotions the relationship between the predictor and the and action tendencies as a function of ingroup outcome became nonsignificant (β = –.034; identification. Anger was experienced to a p = .742), and a Sobel test was signifi cant (z = 5.43; greater extent for higher identifi ers than lower p < .0005) see Figure 5 top). In sum, although identifi ers, and sadness was experienced to a identifi cation predicted both anger and sadness, greater extent for lower identifi ers than higher it was only anger that mediated the effect of identifi ers. These tendencies were, however, identifi cation on the tendency to approach qualifi ed by threat (consistent with previous work the outgroup. on social identity theory), here contextualized We next computed a mediational model of as a team win or loss. When participants’ group the effect of identifi cation on the tendency identity was threatened (a loss), qualitatively to avoid the outgroup. In step 1, the outcome different emotional responses were observed for measure (avoidance) was regressed onto the higher and lower identifi ers: higher identifi ers predictor (identifi cation). This analysis revealed were more angry than lower identifi ers, whereas a signifi cant relationship between identifi cation lower identifi ers were more sad than higher and avoidance (β = –.367; p = .014). In step 2, identifi ers. Identifi cation with the ingroup also the analysis revealed a signifi cant relationship predicted differences in the tendency to ap- between identifi cation and potential mediators, proach or avoid the outgroup following a loss: sadness (β = –.401; p = .007) and anger (β = .720; lower identifi ers were more likely to avoid the p < .0005. In step 3, avoidance was regressed outgroup than higher identifi ers after a loss; onto the mediators while controlling for iden- higher identifi ers were more likely to approach tifi cation. This analysis revealed a signifi cant the outgroup than lower identifi ers after a loss relationship between sadness and avoidance (although this latter tendency was also observed (β = .897; p < .0005), but not anger and avoid- when the ingroup team had won). Crucially, the ance (β = .028; p = .78). Finally, when the effect of identifi cation on action tendencies after mediators were controlled in the fi rst equation, a loss was fully mediated only by the specifi c the relationship between identifi cation and associated emotion: approach tendencies avoidance became nonsignifi cant (β = –.027; were mediated by anger, not sadness; avoidant p = .797), and a Sobel test was signifi cant (z = tendencies were mediated by sadness, but not 2.75; p = .006) (see Figure 5 bottom). In sum, anger. More generally we supported the basic although identifi cation positively predicted both positive relationship between identifi cation and anger and sadness, only sadness mediated the anger (and consequent approach tendencies) effect of identifi cation on avoidance following and the basic negative relationship between a match loss. These distinct mediational paths identifi cation and sadness (and consequent identifi ed for approach and avoidance action avoidance tendencies). These fi ndings support tendencies further support the dissociation of intergroup emotions theory, which proposes emotions and action tendencies as a function that specifi c emotions are associated with specifi c of identifi cation.2 action tendencies, but they also show that these

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion

Figure 5. Top: mediational model of the role of reported anger in explaining the identifi cation–approach action tendency relationship following a match loss. Bottom: mediational model of the role of sadness in explaining the identifi cation–avoidance action tendency relationship following a match loss. mediational paths are moderated by ingroup Smith, 1993, 1999). When group membership identifi cation. Below we discuss the implications as a soccer fan was salient as a result of watching of these fi ndings for intergroup emotions theory, a soccer match with another team, participants social identity theory and efforts to apply social reacted emotionally to losing the match, a psychological theory to real intergroup contexts situation that affected the ingroup. For lower such as those involving sports fan behavior. identifi ers, moreover, losing a soccer match led to the action tendency of outgroup avoidance, Theoretical implications an effect mediated by the emotional reaction The current fi ndings provide strong support for of sadness; for higher identifi ers losing a match intergroup emotions theory (Mackie et al., 2000; led to outgroup approach behaviors, an effect

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1) mediated by anger. While previous research has emotions would only arise when social identity shown that contempt (Mackie et al., 2000) and was salient. The current fi ndings strengthen fear (Silver et al., 2001) led to an action tendency the link between intergroup emotions theory of avoidance, we believe the current study is the and social identity theory by demonstrating that fi rst to show a sadness–avoidance link, and in ingroup identifi cation moderates the effect of a particular that sadness mediates the relationship threat to the ingroup on intergroup emotions. between identifi cation and an action tendency Specifically, they showed that threat to the to avoid the outgroup. Although sadness was not ingroup led to anger for high identifi ers and specifi ed in the original theory as a mediator of sadness for low identifi ers. Previous research has the effect of an appraisal about the intergroup focused on the effects of the intergroup context context on avoidance of the outgroup, we in producing different intergroup emotions, associate sadness with withdrawal from social showing, for example, that when the ingroup is in situations (see Crisp et al., 2002). Summarizing a strong position, there is anger and aggression research on the associations between specifi c toward the outgroup, but when the ingroup is intergroup emotions and action tendencies, in a weak position, there is fear and avoidance of Neuberg and Cottrell (2002) concluded that the outgroup (Mackie et al., 2000). In contrast, sadness was associated with self-contemplation we have found that characteristics of the group and self-imposed seclusion. It therefore makes members themselves infl uence which intergroup sense that sadness emerged as a mediator of emotions arise. Enhancing our understanding the relationship between identifi cation and of the factors that moderate when specifi c inter- avoidance. group emotions and associated action tendencies It is important to note that our fi ndings here will arise represents an important development apply to a context where all members are to in intergroup emotions theory. some extent highly committed. Soccer fans (and Finally, we know that variability in ingroup more generally sports fans), by defi nition, may identifi cation is a reliable predictor of a wide have a baseline level of identifi cation above which range of intergroup attitudes (ranging from they can vary, and this may represent a context self-stereotyping to intergroup bias, e.g. Jetten distinct from that in which identifi cation has et al., 2004). This research has added a specifi c been studied previously. On the one hand this dissociation of emotions and action tendencies to is a potentially important boundary condition this list. Higher identifi cation, as well as leading defi ning when the dissociation of qualitatively to more self-stereotyping, homogeneity and different emotions might be observed, but it intergroup bias, also appears specifi cally to lead also opens up a potentially new and exciting to more anger and approach behaviors. Lower avenue for future research in which defi nitions identifi cation, as well as leading to a greater of ingroup identifi cation can become more tendency to individuate oneself and dissociate fine-grained, potentially revealing a more from the ingroup, appears to specifi cally lead qualitatively detailed picture of the implications to more sadness and avoidance tendencies, of commitment to groups. Broadly speaking, especially under threat. It is this qualitative dis- these fi ndings imply that identifi cation can sociation that is particularly interesting here, not only predict differences in the intensity of supporting the general notion that lower and responses but can also sometimes reveal more higher identifi ers think and act differently in fi ne-grained, specifi c affective or behavioral intergroup contexts. tendencies in particular intergroup contexts, such as those relating to sports team allegiance, Sport and social identity where affi liation has been adopted as a potential The fi ndings of the present study have import- means of providing self-conceptual benefi ts. ant implications for future research within the Intergroup emotions theory is grounded in fi eld of soccer-related types of social disorder social identity theory; Smith (1993) proposed which seeks to uncover systemic intergroup that intergroup—rather than individual level— processes using experimental social psychology

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion approaches, but which strives to keep at its soccer is, increasingly, less credibly attached to center the socially situated affective experiences widespread ‘hooliganism’. As a consequences of the participants. The integration of self- of this, intergroup process explanations of such categorization and social identity principles with incidents now have greater explanatory power. intergroup emotions theory is promising for Stott et al.’s (2001) ethnographic study of the future developments in this applied fi eld, because 1998 World Cup seems to confi rm this point; it incorporates the particular acute affective states they argue that their fi ndings show that, ‘. . . it is felt by, and described by, those who routinely simply not adequate to see all violence . . . purely attend soccer matches. Indeed, we believe that and exclusively in terms of the presence and this is the fi rst fi eld test of the processes outlined predisposition of hooligans’ (p. 379). While by intergroup emotions theory. Marsh et al.’s (1978) social anthropological Marsh, Rosser, and Harré’s (1978) expositive observations are rooted in the decade they were analysis of soccer-related disorder in the UK made, their argument regarding the highly in the late 1970s was timely because of the structured and rule-based nature of social difference between the widely received view of behavior at soccer matches still rings true. Such the phenomenon in question and what they a structured and organized intergroup envir- observed through social scientifi c methods. onment is a dream setting in which to apply They argued that the public perception of and test intergroup emotions theory; there is disorder at soccer matches (shaped through no question that group membership is highly political and media rhetoric) was that it was salient and situations are being constantly chaotic and irregular, the sum total of a large evaluated in terms of their repercussions for number of uneducated aggressive hooligans the ingroup. behaving erratically and nonnormatively. To some extent, all British soccer supporters—but Conclusion particularly those designated ‘hooligans’—were In this research we argue that emotions and action subject to a degree of infrahumanization at that tendencies can be qualitatively differentiated time, and this state of affairs was exacerbated depending on the extent to which people see by the attitude of the right-wing press and gov- the relevant ingroup as being important for their ernment in the 1980s. ‘Hooligans’ were being self-concept. The data we report here support categorized as distinct creatures from ‘normal this hypothesis. Following match losses lower people’. Intraindividual, rather than intergroup, identifiers felt sad but not angry, whereas processes, were to blame. This in itself might higher identifi ers felt angry but not sad and have made it possible to dismiss social-level these negative, but qualitatively different, processes from being applicable on the many emotional reactions mediated reported action occasions on which social disorder was taking tendencies. The fi ndings support the basic links place. Our study suggests that soccer’s quite between specifi c emotion and action tendency particular emotional thesaurus is evident at outlined by intergroup emotions theory, as well a personal level and is linked to what can be as providing a point of synthesis with existing termed loosely as disorder-facilitating (not work on the importance of social identifi cation causing) ‘approach’ tendencies, but stems from in describing and defi ning intergroup relations. a social-level process. This work may therefore aid the development The perception of soccer supporters in society of an elaborated account of the motivational is very different in 2005 following a change and emotive processes that defi ne thought and in the game’s spectator demographics (more behavior in social contexts. women and children following perceived safety improvements in stadia, ‘middle-class’ cultural acceptability since the early 1990s, season tickets Notes becoming prohibitively expensive for low income 1. We also carried out the moderated regression families). As such, social disorder attached to analysis on the items that formed the happiness

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1)

index and found the same result × identifi cation studies. Journal of Personality and Social Psychology, interaction (β = .566; t = 2.32, p =.024), with 34, 366–375. simple slope analysis revealing that identifi cation Crisp, R. J. (2006). Commitment and categorization did not signifi cantly affect happiness when the in common ingroup contexts. In R. J. Crisp & participants’ soccer team had won the match M. Hewstone (Eds.), Multiple social categorization: (β = .351; t = 1.41, p < .18), but was negatively Processes, models and applications (pp. 90–111). related to happiness when the soccer team had Hove, E. Sussex, UK: Psychology Press (Taylor & lost the match (β = –.418; t = –2.98, p < .005). Francis). Happiness is not, however, relevant to the Crisp, R. J., Ensari, N., Hewstone, M., & Miller, theoretical aims of the current research, so we N. (2002). A dual-route model of crossed do not discuss these fi ndings in detail. categorization effects. In W. Stroebe & 2. We also tested these mediational models M. Hewstone (Eds.), European Review of Social irrespective of intergroup threat (win or loss) Psychology (Vol. 13, pp. 35–74). Philadelphia: and observed a replication of all paths found to Psychology Press. be signifi cant above. Crisp, R. J., & Hewstone, M. (2006). Multiple social categorization: Context, process, and social consequences. In R. J. Crisp & References M. Hewstone (Eds.), Multiple social categorization: Abrams, D., & Hogg, M. A. (1988). Comments on Processes, models and applications (pp. 3–22). the motivational status of selfesteem in social Hove, E. Sussex, UK: Psychology Press (Taylor & identity and intergroup discrimination. European Francis). Journal of Social Psychology, 18, 317–334. Crisp, R. J., Stone, C. H., & Hall, N. R. (2006). Abrams, D., & Hogg, M. A. (1998). Prospects for Recategorization and subgroup identifi cation: research in group processes and intergroup Predicting and preventing threats from common relations. Group Processes Intergroup Relations, 1, ingroups. Personality and Social Psychology Bulletin, 7–20. 32, 230–243. Aitken, L. S., & West, S. G. (1991). Multiple Devos, T., Silver, L. A., Mackie, D. M., & Smith, E. R. Regression: Testing and interpreting interactions. (2002). Experiencing intergroup emotions. In London: Sage. D. M. Mackie & E. R. Smith (Eds.), From prejudice BBC News Online (2005). City’s tribute to bombing to intergroup emotions: Differentiated reactions victims. Retrieved November 29, 2005 from to social groups (pp. 113–134). Philadelphia: http://news.bbc.co.uk/1/hi/england/west_ Psychology Press. yorkshire/4682739.stm. Doosje, B., Ellemers, N., & Spears, R. (1995). Branscombe, N. R., Schmitt, M. T., & Harvey, R. D. Perceived intragroup variability as a function (1999). Perceiving pervasive discrimination of group status and identifi cation. Journal of among African Americans: Implications for Experimental Social Psychology, 31, 410–436. group identifi cation and well-being. Journal of Ellemers, N., Spears, R., & Doosje, B. (1997). Personality and Social Psychology, 77, 135–149. Sticking together or falling apart: In-group Branscombe, N., & Wann, D. (1994). Collective identifi cation as a psychological determinant of self-esteem consequences of outgroup group commitment versus individual mobility. derogation when a valued social identity is on Journal of Personality and Social Psychology, 72, trial. European Journal of Social Psychology, 24, 617–626. 641–657. Ellemers, N., van Knippenberg, A., & Wilke, Branscombe, N. R., Wann, D. L., Noel, J. G., & H. A. (1990). The infl uence of permeability of Coleman, J. (1993) In-group or out-group group boundaries and stability of group status extremity: Importance of the threatened social on strategies of individual mobility and social identity. Personality and Social Psychology Bulletin, change. British Journal of Social Psychology, 29, 19, 381–388. 233–246. Brown, R., & Williams, J. (1984). Group Ellemers, N., Wilke, H., & van Knippenberg, A. identifi cation: The same thing to all people? (1993). Effects of the legitimacy of low Human Relations, 7, 547–564. group or individual status on individual and Cialdini, R. B., Borden, R. J., Thorne, A., Walker, collective status-enhancement strategies. M. R., Freeman, S. & Sloan, L. R. (1976). Journal of Personality and Social Psychology, Basking in refl ected glory: Three football fi eld 64, 766–778.

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Crisp et al. identity and emotion

Fédération Internationale de Football Association. ingroup identifi cation. Personality and Social (n.d.). Retrieved November 16, 2005, from Psychology Bulletin, 28, 1586–1600. http://www.fi fa.com/en/organisation/president Neuberg, S. L., & Cottrell, C. A. (2002). Intergroup section/0,1650,2,00.html. emotions: A biocultural approach. In D. M. Frijda, N. H. (1986). The emotions. Cambridge, UK: Mackie & E. R. Smith (Eds.), From prejudice to Cambridge University Press. intergroup emotions: Differentiated reactions to social Hinkle, S., & Brown, R. (1990). Intergroup groups (pp. 265–283). Philadelphia: Psychology comparisons and social identity: Some links and Press. lacunae. In D. Abrams & M. Hogg (Eds.), Social Redersdorff, S., & Martinot, D. (2003). Impact of identity theory: Constructive and critical advances. upward and downward social comparisons on Hemel Hempstead, UK: Harvester Wheatsheaf. self-esteem: Importance of activated identity. Hirt, E. R., Zillmann, D., Erickson, G. A., & Annee Psychologique, 103, 411–443. Kennedy, C. (1992). Costs and benefi ts of Reicher, S. (1996). ‘The Crowd’ century: allegiance: Change in fans’ self-ascribed Reconciling practical success with theoretical competencies after team victory versus defeat. failure. British Journal of Social Psychology, 35, Journal of Personality and Social Psychology, 63, 535–553. 724–738. Roese, N. J. (1997). Counterfactual thinking. Hogg, M. A., & Abrams, D. (1988). Social Psychological Bulletin, 121, 133–148. identifi cations: A social psychology of intergroup Silver, L. A., Miller, D. A., Mackie, D. M., & relations and group processes. London: Routledge. Smith, E. R. (2001). The nature of intergroup Hogg, M. A., & Abrams, D. (1990). Social emotions. Unpublished manuscript, University of motivation, self-esteem, and social identity. In California, Santa Barbara. D. Abrams & M. A. Hogg (Eds.), Social identity Sloan, L. R. (1989). The motives of sports fans. theory: Constructive and critical advances (pp. In J. D. Goldstein (Ed.), Sports, games, and play: 28–47). Hemel Hempstead, UK: Harvester Social and psychosocial viewpoints (pp. 175–240). Wheatsheaf. Hillsdale, NJ: Erlbaum. Jetten, J., Branscombe, N. R., Spears, R., & Smith, E. R. (1993). Social identity and social McKimmie, B. M. (2003). Predicting the paths emotions: Toward new conceptualizations of of peripherals: The interaction of identifi cation prejudice. In D. L. Hamilton & D. M. Mackie and future possibilities. Personality and Social (Eds.), Affect, cognition, and stereotyping: Interactive Psychology Bulletin, 29, 130–140. processes in group perception (pp. 297–315). San Jetten, J., Spears, R., & Postmes, T. (2004). Diego, CA: Academic Press. Intergroup distinctiveness and differentiation: Smith, E. R. (1999). Affective and cognitive A meta-analytic integration. Journal of Personality implications of a group becoming a part of the and Social Psychology, 86, 862–879. self: New models of prejudice and of the self- Luhtanen, R., & Crocker, J. (1992). A collective concept. In D. Abrams & M. A. Hogg (Eds.), self-esteem scale: Self –evaluation of one’s social Social identity and social cognition (pp. 183–196). identity. Personality and Social Psychology Bulletin, Malden, MA: Blackwell. 18, 302–318. Snyder, C. R., Lassegard, M., & Ford, C. E. (1986). Mackie, D. M., Devos, T., & Smith., E. R. (2000). Distancing after group success and failure: Intergroup emotions: Explaining offensive Basking in refl ected glory and cutting off action tendencies in an intergroup context. refl ected failure. Journal of Personality and Social Journal of Personality and Social Psychology, 79, Psychology, 51, 382–388. 602–616. Spears, R., Doosje, B., & Ellemers, N. (1997). Self Mandel, D. R. (2003). Counterfactuals, emotions, stereotyping in the face of threats to group and context. Cognition and emotion, 17, 139–159. status and distinctiveness: The role of group Marsh, P., Fox, K., Carnibella, G., McCann, J., & identifi cation. Personality and Social Psychology Marsh, J., (1996). Football violence in Europe. Bulletin, 23, 538–553. Amsterdam: The Amsterdam Group. Spears, R., Doosje, B., & Ellemers, N. (1999). Marsh, P., Rosser, E., & Harré, R. (1978). The rules Commitment and context in the social of disorder. London: Routledge and Kegan Paul. perception. In N. Ellemers, R. Spears, & Martinot, D., Redersdorff, S., Guimond, S., & Dif, B. Doosje (Eds.), Social identity: context, S. (2002). Ingroup versus outgroup comparisons commitment, content. (pp. 59–83). Oxford, UK: and self-esteem: The role of group status and Blackwell.

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015 Group Processes & Intergroup Relations 10(1)

Spears, R., Jetten, J., & Scheepers, D. (2002). Wann, D. L., Dolan, T. J., McGeorge, K. K., & Distinctiveness and the defi nition of collective Allison, J. A. (1994). Relationships between self: A tripartite model. In A. Tesser, D. A. spectator identifi cation and spectators’ Stapel, & J. V. Wood (Eds.), Self and motivation: perceptions of infl uence, spectators’ emotions, Emerging psychological perspectives (pp. 147–171). and competition outcome. Sport & Exercise Washington, DC: American Psychological Psychology, 16, 347–364. Association. Wann, D. L., Hunter, J. L., Ryan, J. A., & Wright, Steele, C. M., & Aronson, J. (1995). Stereotype L. A. (2001). The relationship between team threat and the intellectual test-performance identifi cation and willingness of sports fans to of African-Americans. Journal of Personality and consider illegally assisting their team. Journal of Social Psychology, 69, 797–811. Social Behavior and Personality, 29, 531–536. Stott, C., Hutchison, P., & Drury, J. (2001). Wann, D. L., Peterson, R. R., Cothran, C., & Dykes, ‘Hooligans’ abroad? Inter-group dynamics, social M. (1999). Sport fan aggression and anonymity: identity and participation in collective ‘disorder’ The importance of team identifi cation. Journal of at the 1998 World Cup Finals. British Journal of Social Behavior and Personality, 27, 597–602. Social Psychology, 40, 359–384. Yzerbyt, V., Dumont, M., Wigboldus, D., & Stott, C. J., & Adang, O. M. J. (2004, June) Gordijn, E. (2003). I feel for us: The impact of ‘Disorderly’ conduct: Social psychology and the categorization and identifi cation on emotions control of football ‘hooliganism’ at ‘Euro2004’. and action tendencies. British Journal of Social The Psychologist, 318–319. Psychology, 42, 533–549. Stott, C. J., & Drury, J. (2000). Crowds, context and identity: Dynamic categorization processes in the Paper received 30 November 2005; revised version accepted ‘poll tax riot’. Human Relations, 53, 247–273. 9 May 2006. Stott, C. J., & Reicher, S. D. (1998a). How confl ict escalates: The inter-group dynamics of collective Biographical notes football crowd ‘violence’. Sociology, 32, 353–377. Stott, C. J., & Reicher, S. D. (1998b). Crowd action richard j. crisp is a reader in social psychology as inter-group process: Introducing the police at the University of Birmingham. He did his BA perspective. European Journal of Social Psychology, in experimental psychology at the University of 28, 509–529. Oxford and his PhD at Cardiff University. His research interests focus on social categorization, Tajfel, H., & Turner, J. C. (1979). An integrative group processes, and intergroup relations. theory of intergroup confl ict. In W. G. Austin & S. Worchel (Eds.), The social psychology of sarah heuston was formerly a student at the intergroup relations (pp. 33–48). Monterey, CA: University of Birmingham from where she was Brooks/Cole. awarded her BSc in Psychology. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. matthew j. farr is a lecturer in social psychology Worchel & W. G. Austin (Eds.), The psychology of at Loughborough University. He did his BA in intergroup relations (pp. 7–24). Chicago: Nelson- social and political sciences at the University Hall. of Cambridge and his PhD at the University of Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, Birmingham. His research interests have focused S. D., & Wetherell, M. S. (Eds.) (1987). on prejudice and discrimination, especially Rediscovering the social group: A self-categorization social identity approaches to confl ict and theory. Oxford, UK: Blackwell. intergroup relations. Wann, D. L., & Branscombe, N. R. (1990). Die-hard and fair-weather fans: Effects of identifi cation rhiannon n. turner is a postdoctoral research on BIRGing and CORFing tendencies. Journal of fellow at the University of Birmingham. She Sport and Social Issues, 14, 103–117. did her BSc at Cardiff University, her MSc at Wann, D. L., Carlson, J. D., & Schrader, M. P. the University of Kent, and her D.Phil. at the (1999). The impact of team identifi cation on University of Oxford. Her research interests the hostile and instrumental verbal aggression focus on intergroup relations, specifi cally on of sports spectators. Journal of Social Behavior and dimensions, mediators and consequences of Personality, 14, 279–286. intergroup contact.

Downloaded from gpi.sagepub.com at UCSF LIBRARY & CKM on March 9, 2015