《社会心理学》课程教学资源（文献资料）conformity——On the Social Influence of Emotions in Groups - Interpersonal Effects of Anger and Happiness on Conformity Versus Deviance

88ar286 On the Social Influence of Emotions in Groups:Interpersonal Effects of Anger and Happiness on Conformity Versus Deviance Mare W.Heerdink.Gerben Homan,and Agneta H.Fischer How do f 's d of hap e hy the es tcd which co dingly,in Study 2.the av ivegottmedms nd this effec of em sand the role of emotions in groups your react s if annoyedv tion may inform him or ber about how others feel about the emotions of dby the ns of other group members In the pre ch,we 002 20D Ofentimes,the emotions that mot ing ways:happir y shared opn HeteroPydalogy.UaigotAaeAa Emotions in Groups Our theorizing is informed by a social-functionl perspective on r.We otions expressed by individualsor (factions within)groups may e this article should be addressed to Marc affect the behavior of other individual Wees rential proc eby individuals use others m.w.heerdink@uva.nl

On the Social Influence of Emotions in Groups: Interpersonal Effects of Anger and Happiness on Conformity Versus Deviance Marc W. Heerdink, Gerben A. van Kleef, Astrid C. Homan, and Agneta H. Fischer University of Amsterdam How do emotional expressions of group members shape conformity versus deviance in groups? We hypothesized that angry and happy responses to a group member’s deviating opinion are interpreted as signals of imminent rejection versus acceptance. In 5 studies, the majority’s expressions of anger led the deviant individual to feel rejected, whereas expressions of happiness made the deviant feel accepted. Because conformity can be seen as strategic behavior aimed at gaining (re)acceptance, the effects of emotional expressions on conformity should be moderated by social-contextual factors that determine the motivation to be accepted by the group and by the extent to which conformity is a means to this end. Accordingly, in Study 2, the availability of alternative groups determined whether a deviant conformed to the current group or abandoned the group after an angry reaction. In Study 3, anger and happiness were only associated with conformity pressure in situations that were perceived as cooperative (rather than competitive). Employing an interactive group task in Study 4, we showed that individuals who received an angry reaction contributed less in a cooperative group task than did those who received a neutral or happy reaction. Finally, in Study 5, peripheral group members conformed more after an angry reaction than after a happy reaction, but prototypical group members did not. Moreover, conformity was still manifest 3 weeks after the experiment, and this effect was mediated by feelings of rejection. We discuss implications of these findings for theorizing about social functions of emotions and the role of emotions in groups. Keywords: conformity, deviance, social influence, emotions, interpersonal effects At a conference, you and several colleagues decide to go out for dinner together. After some discussion, the group decides on finding a pizzeria. After wandering around fruitlessly for a while, you propose to change plans and go to a nearby Asian restaurant instead. Unexpectedly, your colleagues react as if annoyed— even a little angry. Does this negative emotional reaction of your col￾leagues lead you to abandon your new plan and conform to the rest of the group? Or do you decide to leave the group and go to a place that you like? As this example illustrates, emotions are an integral part of group life. Many events that take place in or outside groups trigger emotions in group members by affecting individual or group-based concerns or goals (Smith, 1993)—such as finding a restaurant. Oftentimes, the emotions that are elicited in a group context do not remain private. Rather, they tend to be expressed, deliberately or not, through facial displays, verbal expressions, bodily postures, and tone of voice (Ekman, 1982). Thus, when a group member elicits an emotion in other group members, the expression of this emotion may inform him or her about how others feel about the situation (Keltner & Haidt, 1999). Given how much time most of us spend in groups, it stands to reason that we be influenced by the emotions of our fellow group members. Surprisingly, however, past research has largely neglected the question of how an indi￾vidual group member’s behavior is influenced by the emotional expressions of other group members. In the present research, we examined one important manifestation of such social influence of emotions (van Kleef, van Doorn, Heerdink, & Koning, 2011), namely, the effects of a majority’s emotional expressions on a deviant group member’s behavior. In doing so, we focus on two emotions that have the potential to affect a group member who deviates from a consensually shared opinion or behavior in oppos￾ing ways: happiness and anger. Emotions in Groups Our theorizing is informed by a social–functional perspective on emotion. According to this perspective, emotions serve social functions in dyads, in groups, and between groups alike (e.g., Fischer & Manstead, 2008; Fridlund, 1994; Frijda & Mesquita, 1994; Keltner & Haidt, 1999; Parkinson, 1996; van Kleef, 2009). Emotions expressed by individuals or (factions within) groups may affect the behavior of other individuals or groups via affective processes (e.g., emotional contagion or liking; van Kleef, 2009) or through inferential processes, whereby individuals use others’ emotional expressions to infer information about their motives and This article was published Online First June 17, 2013. Marc W. Heerdink, Gerben A. van Kleef, Astrid C. Homan, and Agneta H. Fischer, Department of Psychology, University of Amsterdam, Amster￾dam, the Netherlands. This research was facilitated by a grant from the Netherlands Organi￾sation for Scientific Research (NWO 452-09-010) awarded to the second author. We would like to thank Marwan Sinaceur for his helpful guidance in developing the emotion instructions used in Study 4; Stephanie Mertens, Rosa Mulder, and Pomme Woltman for their assistance in setting up and conducting parts of this research; and Verlin Hinsz for valuable sugges￾tions for improving this article. Correspondence concerning this article should be addressed to Marc W. Heerdink, Department of Social Psychology, University of Amster￾dam, Weesperplein 4, 1018 XA Amsterdam, the Netherlands. E-mail: m.w.heerdink@uva.nl This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology © 2013 American Psychological Association 2013, Vol. 105, No. 2, 262–284 0022-3514/13/$12.00 DOI: 10.1037/a0033362 262

ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 263 inentions(van Klcef.09:van Kleef.De Dreu.Manstead. Cialdini.&Kenrick.006 Hayes.007:Homsey&Jeten.04 search on the functio ins ha yas any behavior o mainly been con ned with hov affect spreads in g oups.for t from other group m behaviors or opin ed as an act o 994).M 002:Barsade Gibsor 1998:Kelly Barsade 001:Smith arily in line with those of th 99 Spoor ey,200 group eek ou Leach.2004)and how group affe Fons 2004).For state of th the n many situati ions deviance is important for attaining group dra(2005)showed that teams with a leader 1991).Work on hidde n profile 1995)and hbited bettero indeed suffer wher ler who expresse negative affect expen more effor 100 n the communicative aspects of emotion instead of 198 an Knippen ers De Dre anoth of a on. of the hange a group dec 、n for the such motivation s ws that de stimulat Jeh is inte f the argu 100 e.but it s and behavic of fellow dange ring the harmony and tru ist in eatens effectiv goal pursuit when s re we an ested in how deviant Up n mbers'behavior 1991).For instance within plain h rring ce is often s ure on the role of deviance in groun soal attainment alons with oun hy placing them outside what is a nembers may c cally exeluded from the group by Deviance and Group Goals Although group members have a general tendency to maintain g and disen integral part of group life (Griskevicius.Goldstein.Mortensen. and Levine.1989)by increasing their communication to the

intentions (van Kleef, 2009; van Kleef, De Dreu, & Manstead, 2010). Previous research on the functionality of emotions in groups has mainly been concerned with how affect spreads in groups, for instance via “primitive” emotional contagion (i.e., contagion via mimicry and afferent feedback; Hatfield, Cacioppo, & Rapson, 1994). Much research in this domain has focused on the interplay between individual-level and group-level affect (e.g., Barsade, 2002; Barsade & Gibson, 1998; Kelly & Barsade, 2001; Smith, 1993; Spoor & Kelly, 2004; Totterdell, 2000; Totterdell, Kellett, Teuchmann, & Briner, 1998; van Zomeren, Spears, Fischer, & Leach, 2004) and how group affect and affective variability within groups shape group outcomes (e.g., George, 1995; Tiedens, Sut￾ton, & Fong, 2004). For instance, Barsade (2002) found that a confederate’s affective state influenced the mood of the other group members and that contagion of positive affect increased cooperation and group performance. Similarly, Sy, Côté, and Saavedra (2005) showed that teams with a leader who expressed positive affect developed a more positive “group affective tone” (George, 1995) and exhibited better coordination, whereas teams with a leader who expressed negative affect expended more effort on the task. Focusing on the communicative aspects of emotion instead of how affect spreads within groups, another study showed that expressions of anger (as opposed to happiness) on the part of a leader can increase team performance when team members are motivated to consider the implications of the leader’s emotions. When such motivation was high, team members inferred from the leader’s anger that their performance was unsatisfactory, which led them to increase their efforts. The leader’s happiness, on the other hand, was interpreted as a sign that performance was satisfactory, and this inference led to a decrease in effort (van Kleef, Homan, Beersma, van Knippenberg, van Knippenberg, & Damen, 2009). This brief overview of prior research on emotions in groups is far from comprehensive, but it suffices to demonstrate that emo￾tional expressions of group members have the potential to influ￾ence the emotions, inferences, and behaviors of fellow group members. It also reveals that previous research has not considered the perspective of the individual within the group, and how other group members’ emotional expressions that are contingent on one group member’s behavior influence this individual. More specif￾ically, we are interested in how deviant group members’ behavior is influenced by the majority’s emotional expressions in response to their behavior. Thus, we take a communicative approach to the interpersonal effects of emotions within groups to explain how happiness and anger, when expressed by a majority within a group, may influence a deviant individual’s tendency to persist in devi￾ance or to yield to this majority by conforming. Before theorizing about the process underlying these effects, we first review litera￾ture on the role of deviance in group goal attainment, along with evidence regarding naturally occurring emotional responses to deviance. Deviance and Group Goals Although group members have a general tendency to maintain their similarity to others by conforming to the opinion and behav￾ior of other group members (e.g., Asch, 1956), deviance is an integral part of group life (Griskevicius, Goldstein, Mortensen, Cialdini, & Kenrick, 2006; Hayes, 2007; Hornsey & Jetten, 2004; Ridgeway, 1978). We define deviance broadly as any behavior or expression of an opinion or idea that is intentionally or uninten￾tionally different from other group members’ behaviors or opin￾ions. Thus, for instance, in the opening example, the suggestion to go to a nearby Asian restaurant could be construed as an act of deviance. Unintentional deviance is likely to surface frequently within groups, as an individual’s preferences, ideas, intentions, beliefs, and behavior are not necessarily in line with those of the group. In addition, group members may intentionally seek out different roles or diversifying positions to maintain a sense of uniqueness while still belonging to the group (Brewer, 1991; Homan, Greer, Jehn, & Koning, 2010; Hornsey & Jetten, 2004; Mullen & Hu, 1989). In many situations, deviance is important for attaining group goals. These include situations that require creativity and divergent thinking to find the optimal solution to a problem (see, e.g., Kruglanski & Webster, 1991). Work on hidden profiles, diversity, groupthink, and the common knowledge effect show that group performance may indeed suffer when group members suppress deviance by being too critical of new information and converging too much (e.g., Gigone & Hastie, 1993; Janis, 1982; Stasser & Titus, 1985, 2003; van Knippenberg, De Dreu, & Homan, 2004). To avoid this situation, deviance is required. For instance, an expert individual may seek (minority) influence through deviance to change a group decision for the better (Moscovici, Mucchi￾Faina, & Maass, 1994). Similarly, work on constructive conflict shows that deviance can indeed stimulate group performance (Amason, 1996; Jehn, 1995; Schweiger, Sandberg, & Ragan, 1986). Thus, because deviance can be crucial for obtaining good group outcomes, groups that are aware of the value of deviance may encourage it (for similar arguments, see de Wit, Greer, & Jehn, 2012; Kruglanski & Webster, 1991; Tjosvold, Wedley, & Field, 1986) and respond to it with happiness or enthusiasm. On the other hand, deviance may constitute a threat to the group’s goals, because it violates the shared reality in the group, thereby endangering the harmony and trust in the group (Mannetti, Levine, Pierro, & Kruglanski, 2010; Marques, Abrams, Paez, & Martinez-Taboada, 1998; Sani, 2005). Furthermore, deviance threatens effective goal pursuit when coordinated action is re￾quired or when one course of action should be decided upon (Kruglanski & Webster, 1991). For instance, in the restaurant example, one person’s deviating dinner preference may undermine or slow down group decision making. Indeed, in naturally occurring contexts, deviance is often se￾verely socially sanctioned by passively (e.g., ignoring, ostracizing) or actively (e.g., rejecting, bullying, see Molden, Lucas, Gardner, Dean, & Knowles, 2009) excluding the deviant individual. People holding a deviant opinion may be metaphorically ostracized from their group by placing them outside what is an “acceptable” opinion for members of their group (Eidelman, Silvia, & Biernat, 2006; Marques et al., 1998; see also Williams, 2007). Group members may even be physically excluded from the group by voting them out (Schachter, 1951). Deviance in groups usually does not lead to immediate social exclusion, however (Levine, 1989). Instead of ignoring and disen￾gaging from a deviant group member, groups may attempt to resocialize the deviant member (Levin & Moreland, 1994; More￾land & Levine, 1989) by increasing their communication to the This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 263

264 HEERDINK.VAN KLEEF.HOMAN.AND FISCHER deviant group member (Schachter.1951).This increase in com ard a dev roup membe reate the perception that one's belonging in the group is unde one 950 g hostility and aggn sion.For i Conformity in Response to the Majority's Emotional nd Wachtler (197.p.10)study: Expressions "The case was hypotheti the which n the table next to the confederate's face (the one who argued a d or reie cted.the maioritys em onal reaction may alloy the viant or mo ate the individua of anger. and ividual feels in tun rt.depending on the context and the situation the group i ee his or will the argue that onging. 0 o the may incite further deviance.and anger may motivate the individual Conformity can be defined as the act of adjustin one' (Nail.MacDo Anger,Happiness,and Inclusionary Status 2000) Although confo We propose that devian individuals interpret the majority's Gerard.1955).ther ulting c vior is si form to their the group nor ically. arg e that happiness and nger,if expre nity is that a deviant's challenge to the group's p about the deviant individa's o the appiness is elicited by events that an individual p Moreland le vine.1989).Conformity can therefore be 1991) context.e at gaining or i a195 nde.But 1986 had to of thre o)an that .1979 her theor pira nships when shared (Frednckso 1998.2001 his n g dev i( .Walter nity by miss 0sL1998) 2008) hap may mp an relativel be observe by this group.Fo elt re cople to change their behavi (cf.:Clr nents of their groups tha lid participants who felt less rejecte to the when their roup ocial distance between the deviant and the rest of the Adarves-Yomo.2006)Similarly. Dewall (2010 urt rmore,the evidence discussed suggests that D people who they e highly g in heir ude to be c with th e opinio s of the groups (Baumeister&Leary.1995: evine,2008:Sman Thus

deviant group member (Schachter, 1951). This increase in com￾munication may be motivated by anger, as studies show that group members feel anger toward a deviant group member (e.g., Ju￾vonen, 1992; Phillips, 2003; see also Festinger, 1950). Anecdotal reports indeed illustrate how the majority’s reactions to deviance may escalate into strong hostility and aggression. For instance, Nemeth (2010) described the reactions of naive subjects in the group to a confederate advocating a deviant position in her Nemeth and Wachtler (1974, p. 10) study: “The case was hypothetical— but the anger was so evident that subjects were pounding their fists on the table next to the confederate’s face (the one who argued a minority position on compensation).” Thus, attempts to force a deviant individual to conform may be accompanied by expressions of anger. In short, depending on the context and the situation the group is facing, deviance may be welcomed or not. To effectively pursue group goals, groups therefore need to be able to regulate deviance. We argue that the majority’s expressions of anger and happiness in response to deviance can be functional in this respect, as happiness may incite further deviance, and anger may motivate the individual to conform. Anger, Happiness, and Inclusionary Status We propose that deviant individuals interpret the majority’s emotional reaction to their behavior to estimate their position in the group, which may motivate them to change their behavior. More specifically, we argue that happiness and anger, if expressed toward a deviant individual in a group, may be interpreted as information about the deviant individual’s inclusionary status. In other words, these emotional expressions influence the degree to which a deviant feels accepted or rejected by the group. Happiness is elicited by events that an individual perceives as goal congruent (Lazarus, 1991). In a dyadic context, expressions of happiness are interpreted as a signal that the environment is safe (Klinnert, Emde, Butterfield, & Campos, 1986; Sorce, Emde, Campos, & Klinnert, 1985) and expressing happiness (i.e., smil￾ing) is considered a strategy for affiliation (Clark, Pataki, & Carver, 1996; Fridlund, 1991, 1994; Kraut & Johnston, 1979). Indeed, positive emotions such as happiness serve affiliative func￾tions (van Kleef, De Dreu & Manstead, 2010), as they help build social relationships when shared (Fredrickson, 1998, 2001). Sim￾ilarly, in the group context, positive affect is linked to the devel￾opment of trust and harmonious intragroup relations (e.g., Walter & Bruch, 2008). Thus, happiness may implicate that one’s belong￾ing in the group is secure. We therefore expected to find that a deviant individual would feel relatively accepted if the majority responds with happiness to his or her deviance. Anger, on the other hand, is often expressed in an attempt to get other people to change their behavior (cf. Averill, 1982; Clark et al., 1996; Fischer & Manstead, 2008; Fischer & Roseman, 2007), which implies that one’s opinion or behavior is currently unac￾ceptable to the expressers (Fischer & Manstead, 2008). In a group setting, expressions of anger may therefore draw attention to the social distance between the deviant and the rest of the group. Furthermore, the evidence discussed suggests that expressions of anger precede or accompany social exclusion in groups. Given that humans are highly sensitive to the safety of their belonging in groups (Baumeister & Leary, 1995; Kerr & Levine, 2008; Smart Richman & Leary, 2009; Williams, 2007; Williams, Cheung, & Choi, 2000), the majority’s expressions of anger may therefore create the perception that one’s belonging in the group is under threat. Hence, we predicted that a majority’s expression of anger would cause the deviant individual to feel rejected by the group. Conformity in Response to the Majority’s Emotional Expressions By affecting the extent to which a deviant individual feels accepted or rejected, the majority’s emotional reaction may allow the deviant individual to remain deviant or motivate the individual to conform. When the majority expresses happiness in response to deviance, and the deviant individual feels accepted in turn, the deviant is not likely to change his or her behavior and can therefore be expected to persist in deviance. In the case of an angry reaction, however, the deviant will feel rejected, and the deviant will there￾fore be motivated to restore the sense of belonging. One way to do restore this sense is by conforming to the majority’s position. Conformity can be defined as the act of adjusting one’s overt behavior in such a way that it becomes more in line with the apparent group norm (for a similar definition, see Nail, MacDon￾ald, & Levy, 2000). Although conformity may be attributed to various motives (e.g., Cialdini & Goldstein, 2004; Deutsch & Gerard, 1955), the resulting overt behavior is similar: Conformity involves movement toward the group norm. From the group’s point of view, the most important consequence of behavioral conformity is that a deviant’s challenge to the group’s position is removed. Thus, by conforming, a person can show a commitment to the group’s identity (i.e., identity performance; Klein, Spears, & Reicher, 2007) and group goals, which may increase acceptance from the group (cf. Hollander, 1960; Levine & Moreland, 1994; Moreland & Levine, 1989). Conformity can therefore be seen as strategic behavior aimed at gaining acceptance in a group. This idea is illustrated by prior research. For instance, Asch (1956), in an experiment in which participants had to choose which of three lines was the shortest, showed that even if people are really certain of their own judgment, they conform to the clearly erroneous opinion of a majority. Similarly, in her theory of the spiral of silence, Noelle-Neumann (1974, p. 43) observed that “to the indi￾vidual, not isolating himself is more important than his own judgment.” This may lead individuals holding deviant opinions to be reluctant to speak out in anticipation of negative reactions (i.e., conformity by omission; Cialdini & Trost, 1998). Several studies have suggested that conformity is especially likely if an individual feels motivated to seek acceptance from a group and if conformity can be observed by this group. For instance, Dittes and Kelley (1956) showed that participants who felt rejected by their group publicly conformed more to the judg￾ments of their groups than did participants who felt less rejected. In another study, peripheral group members (who experienced insecure status within their group) strategically exhibited greater conformity when their responses were made public to an ingroup audience than when their responses remained private (Jetten, Hornsey, & Adarves-Yorno, 2006). Similarly, DeWall (2010) showed that people who were led to expect that they would have a lonely future (Twenge, Catanese, & Baumeister, 2003) changed their attitudes to be congruent with the opinions of their peers (see also Lakin & Chartrand, 2003). Thus, conformity is likely if a This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 264 HEERDINK, VAN KLEEF, HOMAN, AND FISCHER

ON THE SOCIAL INELUENCE OF EMOTIONS IN GROUPS 265 The Present Research Studv 1 The theoretic Fr In Study I our aim wa to establish the hynothesized relatio een majority emotions and felt acceptance and rejection bya individual y ould feel re cted if the ma themselves in I his or no on.We pathsho subiecti ed f We fur expected that participants would report feeling more rejected r deviance.as it makes the devian on than after a r eutral reaction.V ncluded the di ual to seek ways to restore the sense of belonging.In ligh 1974 ts to fe angry reaction. the xtent to which (a)the devi ant is motiv ated t s t Method Participants and design.One hundred and fifteen individual e tested the basic on the which was ve to part of a test ry in which first-ve pu digm to te wheth actions from the majority:anger.happiness.disappoint erials and p Vigne vignette that des udy 3).In Studv ated.The t）had come ds to dis de on th tion nt d ted. asimulated group interaction.and w y ended with"When it's your tum,you tell the others where oto our n the introductions to the m(happiness cond and nd react neutral"(control condition).We used the word enthus all of ()b cause and 5)denenden than wed ting the 2002 put from the redu der made.Finally.with regards to hypothesis testing.weused on

person feels rejected, is motivated to seek acceptance in a given group, and if conformity is likely to elicit acceptance because it will be both perceived and appreciated by the group. The Present Research The theoretical model guiding this research is depicted in Figure 1. The first path in the model represents our hypothesis that a deviant individual would feel rejected if the majority expresses anger about his or her deviance, whereas the deviant individual would feel accepted if the majority expresses happiness. The second path shows how this subjective sense of acceptance or rejection, in turn, affects the behavior of the deviant individual. We expected that a happy reaction would not motivate behavioral change or would elicit further deviance, as it makes the deviant feel that his or her deviant behavior is acceptable. Feeling rejected after an angry reaction, on the other hand, might motivate the individual to seek ways to restore the sense of belonging. In light of the view of conformity as strategic behavior aimed at gaining acceptance in a group (e.g., Asch, 1956; Noelle-Neumann, 1974), we proposed that whether a deviant individual conforms to the majority position after an angry reaction from the group depends on the extent to which (a) the deviant is motivated to (re)gain acceptance in the group, and (b) conformity is a possible means to this end. We examined these ideas in five studies. In Study 1, we tested the basic idea that emotional expressions are interpreted as signals of an individual’s inclusionary status using a vignette approach. Then, we tested the influence of the motive to be reaccepted by manipulating the availability of alter￾native groups in another vignette study, Study 2. Next, we used a critical incidents paradigm to test whether happiness and anger are associated with differences in the perceived pressure to conform and whether this association is affected by the extent to which the situation is perceived as cooperative or competitive, as this deter￾mines whether conformity is an effective means to gain accep￾tance. We also tested if felt acceptance or rejection could account for this association (Study 3). In Study 4, we extended and repli￾cated these findings in a cooperative group task involving real interaction and a behavioral outcome measure. Finally, we tested the influence of the extent to which one’s status as a group member is secure, as another factor determining the motive to be reac￾cepted, using a simulated group interaction, and we tested whether emotional expressions produce conformity that lasts over time (Study 5). The specific hypotheses concerning these moderators will be developed in the introductions to the respective studies. On a statistical note, we use variants of regression analysis (in the statistical computing software R Version 2.15.1; R Core Team, 2012) for all of our analyses. There are two reasons for this choice. First, regression analysis can accommodate both dichotomous (Studies 2 and 5) and continuous (Studies 1, 3, 4, and 5) dependent variables and allowed us to do multilevel analysis (Study 4), thereby providing statistical consistency across studies. Second, regression coefficients were necessary for conducting the (moder￾ated) mediation analyses that were required to test our theoretical model (Preacher, Rucker, & Hayes, 2007). By focusing on regres￾sion output from the outset, we avoid reporting redundant statis￾tical analyses. To facilitate interpretation, we also reported means and standard deviations wherever comparisons between groups are made. Finally, with regards to hypothesis testing, we used one￾tailed tests to test directional hypotheses and two-tailed tests in all other cases. Whenever a one-tailed test is used, we noted it explicitly in the text. Study 1 In Study 1, our aim was to establish the hypothesized relation between majority emotions and felt acceptance and rejection by a deviant group member. Participants imagined themselves in a group in which a majority reacted with anger, happiness, disap￾pointment, or no emotion to their own deviant opinion. We mea￾sured the extent to which participants would feel accepted or rejected from the group as a result of this emotional expression. We expected that participants would report feeling more rejected after an angry reaction than after a neutral reaction, whereas participants were expected to feel more accepted after a happy reaction than after a neutral reaction. We included the disappoint￾ment condition to rule out the possibility that any effects of majority anger on feeling rejected could be attributed to the reac￾tion being generally negative in nature, and we expected partici￾pants to feel less rejected after a disappointed reaction than after an angry reaction. Method Participants and design. One hundred and fifteen individuals (26 men, 88 women, one individual missing demographic infor￾mation, Mage
21.03, range 17–54 years)1 took part in the experiment, which was part of a test battery in which first-year psychology students participated to fulfill a course requirement. Participants were randomly assigned to receive one of four emo￾tional reactions from the majority: anger, happiness, disappoint￾ment, or neutral (i.e., no emotion). Materials and procedure. Vignette. Participants first read a short vignette that described a group situation in which the majority’s emotion was manipu￾lated. The protagonist (same sex as the participant) had come together with three same-sex friends to discuss and decide on their vacation destination. All three friends shared a preference for one destination, while the protagonist had picked a different destina￾tion. Thus, the situation resembled a standard conformity paradigm (e.g., Asch, 1956) with a majority of modal size (Bond, 2005). The story ended with “When it’s your turn, you tell the others where you’d like to go. Your friends don’t immediately agree with you . . .,” followed by “. . . and react with anger” (anger condition), “. . . but react with enthusiasm” (happiness condition), “. . . and react with disappointment” (disappointment condition), or “. . . and react neutral” (control condition). We used the word enthusi￾asm (enthousiasme in Dutch) instead of happiness (blij) because it was more ecologically valid in this situation. Although enthusiasm may imply slightly more arousal than happiness according to intrapersonal affect circumplex models (e.g., Russell & Barrett, 1 In all experiments reported in this article, we checked whether the genders were balanced over condition as a precondition to running our analyses. Although the low number of males in all studies did not permit us to conduct analyses using participant gender reliably, explorative inclu￾sion of this factor in the discussed models revealed only isolated indica￾tions of effects of participant gender, none of which challenged our main conclusions. We therefore do not discuss this variable in any of our studies. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 265

266 HEERDINK.VAN KLEEF.HOMAN.AND FISCHER soclalcontet 1999).they can be grouped into the same affiliative family of Acceptance/rejection.To test whether the maiority cmotion had the owhch participants e General discussion) the neutral control condition.As expected.after an angry reaction craecptedorrgjectcdusine ga four-item scale that was similar to 2000:Wili Sommer.197).The items were as follows:"Due to the group's "The makes me utral re andue to the groupseaction.I feel suppored(the -point scale (ranging ll pants in the angry condition.3)6.(onc-tailed) cate acceptan Discussion le was good (Cronbach's a=84).The aggregate score was ation in which wheter the manipulation of nger,wher cepted ifth the fi that on vith enthusiasm. ang s rejected after (from I =not at all to 7=very much). Results on. Manipulation check. a.Nowtht this basic effect s established. red with whenit 48 0.B Study 2 1.59)than when it had been de ribed as neutral (M 2.73 ejection may inspire negative views of the group,undermin 1610rd d00240B63 s.01(on-tailed)Finally.the disappointed (M=4.92.SD =2.15)con pared with when it had le to look for ways to restore belonging (Ma Wall,Baumeiste 2007,f y conform 31.sD=149.B= -0.84.al> 3.6.ps001 (one-tailed). membershipamative group is readily available form to the de Hence.the manipulation was successful

1999), they can be grouped into the same affiliative family of emotions based on their comparable interpersonal effects (van Kleef, De Dreu & Manstead, 2010; we return to this issue in the General Discussion). Acceptance/rejection scale. After participants had imagined themselves in the situation, we measured the extent to which they felt accepted or rejected using a four-item scale that was similar to other scales developed for this purpose (e.g., Wesselman, Butler, Williams, & Pickett, 2010; Williams et al., 2000; Williams & Sommer, 1997). The items were as follows: “Due to the group’s reaction, I feel rejected,” “The group’s reaction makes me feel alone against the rest,” “The group’s reaction makes me feel accepted,” and “Due to the group’s reaction, I feel supported” (the latter two items being reverse-scored). Items were answered on a 7-point scale (ranging from 1
not at all to 7
very much). Reflecting our bipolar conceptualization of this dimension, scores above the midpoint of the scale (4) indicate rejection and scores below the midpoint of the scale indicate acceptance. A principal factor analysis confirmed the unidimensionality of the scale (all factor loadings on the first factor above .69) and the reliability of the scale was good (Cronbach’s
.84). The aggregate score was calculated by averaging the items. Manipulation checks. To check whether the manipulation of majority emotion had been successful, at the end of the experi￾ment, participants indicated to what extent the group had reacted with enthusiasm, anger, and disappointment to their proposal (from 1
not at all to 7
very much). Results Manipulation check. Manipulation checks confirmed that the majority reaction was perceived as more angry when it had been described as angry (M
5.48, SD
1.96) compared with when it had been described as neutral (M
2.60, SD
1.48,
1.39), enthusiastic (M
1.90, SD
1.21,
1.73), or disappointed (M
2.77, SD
1.50,
1.31), all |t|s  6.4, all ps  .001 (1-tailed). Similarly, the majority reaction was perceived as more enthusiastic when it had been described as enthusiastic (M
5.53, SD
1.59) than when it had been described as neutral (M
2.73, SD
1.57,
1.41), angry (M
2.34, SD
1.56,
1.61), or disappointed (M
2.50, SD
1.10,
1.53), all |t|s  7.3, ps  .001 (one-tailed). Finally, the majority reaction was perceived as more disappointed when it had been described as disappointed (M
4.92, SD
2.15) compared with when it had been described as neutral (M
2.60, SD
1.45,
1.20), enthusiastic (M
2.17, SD
1.49,
1.43), or angry (M
3.31, SD
1.49,
0.84), all |t|s  3.6, ps  .001 (one-tailed). Hence, the manipulation was successful. Acceptance/rejection. To test whether the majority emotion had affected the extent to which participants felt accepted or rejected, we tested the effects of the emotion manipulations against the neutral control condition. As expected, after an angry reaction, participants reported feeling more rejected (M
4.65, SD
1.14) than after a neutral reaction (M
3.84, SD
1.04),
0.66, t
2.95, p
.002 (one-tailed). Also as predicted, after an enthusiastic reaction, participants felt less rejected (i.e., more accepted, M
2.91, SD
1.03),
0.77, t
3.45, p  .001 (one-tailed) than after a neutral reaction. Finally, a disappointed reaction (M
4.06, SD
0.95) did not arouse stronger feelings of rejection than a neutral reaction,
0.18, t
0.77, p
.44. Additional independent t tests revealed that, as expected, participants in the disappointed condition reported feeling less rejected than partici￾pants in the angry condition, t(53)
2.06, p
.02 (one-tailed). They also felt more rejected than participants in the happy condi￾tion, t(54)
4.31, p  .001. Discussion In Study 1, we showed that in a situation in which one disagrees with the majority, one feels less accepted (i.e., more rejected) if the majority expresses anger, whereas one feels more accepted if the majority expresses happiness. Furthermore, the finding that one does not feel more rejected if the majority expresses disappoint￾ment compared with a neutral reaction, and feels less rejected after a disappointed reaction compared with an angry reaction, suggests that not all negative emotional reactions lead to feelings of rejec￾tion. These findings support the basic assumption underlying the present research, namely, that happiness and anger are signals of one’s inclusionary status. Now that this basic effect is established, the question is how expressions of anger versus happiness influ￾ence the deviant individual’s behavior. Study 2 Being rejected is a painful experience, which may fuel two very different behavioral tendencies. On the one hand, feelings of rejection may inspire negative views of the group, undermine identification, and lead people to leave their group (Williams, 2007; see also Levine & Moreland, 1994; Moreland & Levine, 1989). On the other hand, being rejected also constitutes a threat to the sense of belonging (Baumeister & Leary, 1995), which moti￾vates people to look for ways to restore belonging (Maner, DeWall, Baumeister, & Schaller, 2007), for instance by conform￾ing to the majority position. Whether people leave the group or conform to the group in such cases likely depends on whether membership in an alternative group is readily available. Consistent Figure 1. General theoretical model. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 266 HEERDINK, VAN KLEEF, HOMAN, AND FISCHER

ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 267 the idca that tcetroup:williams ef a2w bal-throwinggam Results nt or prdicted that ouas there is described as angry (M5.50.SD=1.31)compared with wheni Based on these considerations,in Study 2 we aimed to invest 2.15, ed).Si1)B=0.73 ns exp to th For ad been pulate the avai M-194.D704.B-075.7 56.p001 (one-tailed).Finally. n430,D ejected)than for people a happy reaction (and na nipulations affected the check for the other manipulatio d choose to remain in the on any o the manipulation checks fiaccptanceandrejection anger .001 Method ffects involving As can be seen from available:yes or no)2(majority emotion:anger or happiness) n-s dure 1 the ns.As e s adapted from the one used in Study his in n we calculated simp of the m orityCpr on after ar yo haven't et n angry available .the met some fellow students whom like.and when you were ne ma 18 the i tives condition).The ontinued as in Study 1. me four-item scal in the ves were tha but when ater hat while abandonine one's own destination ("Abide by the feeling rejected ngand group.depending on the availabilty of aer cu m The manipulation of the of the indirect effec of varable on penden with three go on vac With the logit in obit link Cronbach's a =.82).Two more items checked to which extent the

with the idea that feeling rejected can prompt people to seek belonging in a different group, Williams et al. (2000) showed that people, after having been ignored by two other participants in a virtual ball-throwing game, conformed more to the unanimously incorrect decisions of an alternative group. Yet, if there is no viable alternative to the current group, we predicted that people would feel pressured to conform to their current group, as there is no other way to restore their sense of belonging. Based on these considerations, in Study 2 we aimed to investi￾gate whether emotions expressed by a majority influence the choice between conforming to the current group and leaving the group. For this purpose, the scenario from Study 1 was modified to manipulate the availability of alternative groups in addition to the emotion expressed by the majority. We hypothesized that this choice would depend on the availability of alternatives: If alter￾natives are available, the likelihood of exiting the group should be higher for people who receive an angry reaction (and therefore feel rejected) than for people who receive a happy reaction (and there￾fore feel accepted). When no alternatives are available, people should choose to remain in the group, regardless of whether they feel rejected. A further aim was to find out whether any influence of perceived majority emotions on behavior would be mediated by felt acceptance and rejection. Method Participants. Seventy-three participants (18 men, Mage
21.04, range 18 – 44 years) were recruited for the experiment in exchange for 7 euro or partial course credits. Participants were randomly assigned to one of the conditions of a 2 (alternatives available: yes or no)  2 (majority emotion: anger or happiness) between-subjects design. Materials and procedure. Vignette. The vignette was adapted from the one used in Study 1. In this version, the introduction explained that the protagonist had just started studying in a different city, where he or she hardly knew anyone. We manipulated the availability of alternatives by then including in the story either the statement “You hardly know anyone in your new study group, and you haven’t met any fellow students that you like so much that you’d like to go on vacation with them” (no alternatives condition) or the statement “You have met some fellow students whom you like, and when you were recently discussing vacations, you had the impression that every￾one would be interested in going on vacation together” (alterna￾tives condition). The story then continued as in Study 1. Acceptance/rejection scale. We used the same four-item scale as in Study 1 (Cronbach’s
.89). Conforming versus leaving the group. Participants were asked to choose between two alternatives: (a) conforming to the group, while abandoning one’s own destination (“Abide by the majority”), or (b) attempting to find other people to go on vacation with (“Go on vacation with others”). These options were presented as two buttons on the screen, forcing a choice between these alternatives. Manipulation checks. The manipulation of the availability of alternatives was checked with three items (e.g., “Except for my friends from high school, there is nobody I could go on vacation with,” rated from 1
strongly disagree to 7
strongly agree; Cronbach’s
.82). Two more items checked to which extent the group had reacted with happiness and anger to their proposal (rated from 1
not at all to 7
very much). Results Manipulation checks. As intended, participants reported hav￾ing received a more angry reaction when the reaction had been described as angry (M
5.50, SD
1.31) compared with when it had been described as happy (M
2.15, SD
1.31),
0.73, t(71)
9.12, R2
.54, p  .001 (one-tailed). Similarly, partici￾pants indicated that the reaction had been more happy after the reaction had been described as happy (M
4.44, SD
1.17) as opposed to angry (M
1.94, SD
1.04),
0.75, t(71)
9.58, R2
.56, p  .001 (one-tailed). Finally, participants indicated that there were more alternative groups that they could go on vacation with in the alternatives condition (M
5.98, SD
0.76) than in the no alternatives condition (M
4.30, SD
1.73),
1.06, t(71)
5.30, R2
.28, p  .001 (one-tailed). None of the manipulations affected the check for the other manipulation, and no interactions were found on any of the manipulation checks. Thus, the manipulations were successful. Acceptance/rejection. As in Study 1, participants felt more rejected after the majority had expressed anger (M
4.99, SD
1.02) than after the majority had expressed happiness (M
2.85, SD
1.01),
0.73, t(71)
8.99, R2
.53, p  .001 (one-tailed). There were no main or interaction effects involving alternatives. Conforming versus leaving the group. As can be seen from Figure 2, the choice between conforming to the group or leaving the group depended on both the availability of alternatives and the emotion expressed by the majority. Using probit regression,2 the choice between conforming and leaving the group was regressed on the manipulations. As expected, the interaction was significant, B
1.76, Wald’s z
2.68, p
.004 (one-tailed). To interpret this interaction, we calculated simple slopes of the majority emo￾tion manipulation within the alternatives and no alternatives con￾ditions (Aiken, West, & Reno, 1991). As expected, if an alternative group was available, fewer participants chose conformity after an angry reaction (six out of 16, or 37.50%) than after a happy reaction (16 out of 19, or 84.21%), B
1.32, Wald’s z
2.80, p
.003 (one-tailed). When no alternative group was available, the emotions expressed by the majority did not affect the choice between conformity and leaving the group (anger: 15 out of 18, or 83.33%; happiness: 14 out of 20, or 70.00%), B
0.44, Wald’s z
0.97, p
.33. Thus, when no alternatives were available, participants generally preferred staying in the group even if that meant yielding to the majority’s position, but when alternatives were available, anger expressed by a majority increased the chance that participants would prefer to leave the group. Mediation analysis. To investigate whether feeling rejected after an angry reaction could explain the choice between conform￾ing and leaving the group, depending on the availability of alter￾natives, we conducted a moderated mediation analysis (Preacher et al., 2007). A moderated mediation analysis estimates the strength of the indirect effect of an independent variable on a dependent 2 With the logit instead of the probit link function used for these analyses, the reported p values are virtually identical (deviations in the .005–.01 range in both directions). The interpretation does not change. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 267

HEERDINK.VAN KLEEF.HOMAN.AND FISCHER Discussion this perso stays in the group or eiving an angr 20 Afte reaction,participants in this study almost invari er leaving th group.As c ty wa ed desre torem rather tha e to rega h of thes on I ariable through a mediator separately at different values of the ountfor the behavior of participan who receive I a happy The choice between conforming (0)or leaving the dent of the choice betweer of the in the group or no switched toc ren ving the group relative to conformin The em which taps into naive th about the effectsof s (ct 3)and may therefore gh山 of the path from fecling reje ed t see rg ion to devian in groups.Thus.in the last three experiment rmined by bootstrapping (10,000 in effect,a divi Study 3 inducin feein of tance versus As nd all ed and accele I(BC Efror 1987)confi of showing that one isa good group ngs ps nd cating no mediation)and are th goal pursuit.This implies that especially in cooperative setings val of the tha the group 1998:Dirks,1999 has indicated that conformity is indeed mor -021. 95 BC.confid interdepend 1957).We therefore proposed that feeling rejected makes people that th ce between confo For this an 4 02 (one-tailed).It gy201 angry rea sided con onal hy eeling tween cor ming and leaving the group

variable through a mediator separately at different values of the moderator. The choice between conforming (0) or leaving the group (a) was the dependent variable and was specified so that higher coefficients of the predictor variables indicate a higher likelihood of leaving the group relative to conforming.3 The emo￾tion manipulation served as the independent variable, perceived rejection as the mediator, and we used availability of alternatives as the moderator of the path from feeling rejected to the choice between conforming or leaving the group (see Figure 1 for the general model that was tested). The significance of the indirect effects was determined by bootstrapping (10,000 resamples in total). Calculating a coefficient estimate based on the likelihood of one response category over another in case one of the cells in the design is empty is, in effect, a division by zero, which yields coefficient estimates that approach infinity. As such estimates are obviously incorrect, resamples of the data set with empty cells were dropped from the distribution of indirect effects. The esti￾mates are therefore based on the remaining resamples (R
9,067). In this and all remaining mediation analyses, both 95% bias￾corrected and accelerated (BCa; Efron, 1987) confidence intervals and nonparametric p values are reported. These p values are based on the proportion of the distribution of indirect effects on either side of 0 (indicating no mediation) and are therefore uncorrected. The results of the analysis indicated moderated mediation. When no alternatives were available, 0 was enclosed in the con￾fidence interval of the indirect effect, indicating that any effects of the majority emotion manipulation on the choice between con￾forming and leaving the group were not mediated by felt accep￾tance/rejection,
0.21, 95% BCa confidence interval (CI) [1.27, 0.81], p
.74 (one-tailed). When alternatives were avail￾able, however, 0 fell outside the confidence interval of the indirect effect, indicating that the choice between conforming and leaving the group was mediated by feeling accepted versus rejected due to the majority’s emotions,
1.09, 95% BCa CI: [lower limit: 0.31],4 p
.002 (one-tailed). It can be concluded that although all participants felt more rejected after an angry reaction, this only led them to leave the group if an alternative group was available. When no alternatives were available, feeling rejected did not affect the choice between conforming and leaving the group. Discussion In this study, we replicated the finding that a deviant who receives an angry reaction from a majority feels more rejected than a deviant individual who receives a happy reaction. Furthermore, we showed that the availability of alternatives determines whether this person stays in the group or leaves after receiving an angry reaction. When no alternatives to the current group are available, showing good group membership by conforming is the likely option as this helps resolve the threat to belonging when experi￾encing feelings of rejection. If membership in an alternative group is available, the deviant is likely to leave the group after an angry reaction. Happiness, on the other hand, leads to feeling accepted, which appears to keep people committed to the group. After a happy reaction, participants in this study almost invari￾ably chose conformity over leaving the group. As conformity was contrasted with leaving the group, the preference for conformity after a happy reaction may reflect a heightened desire to remain in the group, rather than a desire to regain acceptance. Based on this research, we cannot determine which of these explanations can account for the behavior of participants who received a happy reaction. Therefore, in the last three studies, we employed mea￾sures of conformity that were independent of the choice between staying in the group or not. Additionally, we switched to different paradigms to overcome the limitations of the vignette paradigm, which taps into naïve theories about the effects of emotions (cf. Parkinson & Manstead, 1993) and may therefore produce slightly different results than actual reactions to emotions that surface in reaction to deviance in groups. Thus, in the last three experiments, we used more realistic settings to test how a majority’s angry and happy reactions to deviance shape conformity. Study 3 In this study, we investigated in which situations the majority’s emotional expressions can pressure deviant individuals to conform by inducing feelings of acceptance versus rejection. As argued in the Introduction, whether feeling rejected leads to conformity critically depends on the extent to which conformity is a mean￾ingful way of showing that one is a good group member. In cooperative settings, coordinated action is required for groups to achieve their shared goals, and deviance may threaten effective goal pursuit. This implies that especially in cooperative settings, good group membership may be communicated by showing com￾mitment to the group’s goals and a willingness to conform to further the group’s interests (Cialdini & Trost, 1998; Dirks, 1999; Mayer, Davis, & Schoorman, 1995). In line with this reasoning, previous research has indicated that conformity is indeed more likely to the extent that (positive) interdependency (i.e., coopera￾tiveness) is perceived among the group members (Berkowitz, 1957). We therefore proposed that feeling rejected makes people 3 For this analysis, R was programmed according to Preacher et al.’s (2007) recommendations and checked in personal communication with Andrew F. Hayes (May 2011). 4 This notation indicates a one-sided confidence interval that is used to test a directional hypothesis. The notation [lower limit: X] indicates that there is a 95% confidence that the test statistic falls between X and positive infinity. If the hypothesis predicts a positive relation, and X is greater than zero, the hypothesis is supported. 0 20 40 60 80 100 No Alternatives Alternatives Available % choosing leaving the group over conformity Majority Emotion Happiness Anger Figure 2. Proportion of participants choosing leaving the group over conforming (Study 2). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 268 HEERDINK, VAN KLEEF, HOMAN, AND FISCHER

ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 269 rative.but not in they just described.The scale was reversed for use in the analyses this person would feel p nform ir sing a list of 26 affect scales (from conforming,as there respective emotion had been wn by the stionaire using principal facte sis showed that th etta&).which may motivate him or her to ructure wa y depende andanger.which weretw with anger na situation that this person perceived as competitiv that emerged co n by asking the optimal number of factors.Focusing on these tw Pin They wer Conformity was measured by asking people to which extent they onbach'sg =83)and the second as anger r=20 001) his measur notion cluste gated by averagin people rally reluct red using three items Tw of these items ("To wha group pres experienced during the experiment. oth ded) out the goal st formity to make it oint scales (from 1 not atall to ? competiti (Be &Noussair.2010)that is ultim the 3md5= e).The scale co ed of these item by conl would therefore ny effects of ma rity emotions on conformity pre individ ere mediated by perceived reiection res Method Participants. Sixty-gh participants Results unable to r groups consensus. an d of and 46 female participants M 22.11.range 18-50 years). h course credits or 7 eurd Overal.many situations that we described ritical inciden prompt. I and 2. oles here ctin a number of unrelated personality neasures.the de in which sked to describe as many detai anted a cheap car though,so we wouldn't have to worry abou cceptance/rejection. After participants had described the sit (ro SThe full list is jealousy.disappointm 1-rejected to 7-accepted)how they had felt in the situation

conform in situations they perceive as cooperative, but not in situations that are perceived as competitive. We expected that when a deviant individual perceived the situation as cooperative, this person would feel pressure to conform in case the majority reacts with anger to their deviance. In competitive settings, on the other hand, an individual cannot show commitment to a group goal by conforming, as there are conflicting goals in the group. The majority’s anger may even signal that the individual is reaching his or her goals at the expense of the pursuit of other people’s goals (Lanzetta & Englis, 1989), which may motivate him or her to stay the course. Thus, we expected that a deviant individual would be less likely to feel the pressure to conform if the majority responded with anger in a situation that this person perceived as competitive. We investigated the role of the perceived cooperativeness of the situation by asking participants to recall a situation in which their opinion had differed from that of other group members. They were then asked to report the emotions expressed by the majority and to reflect on the type of situation in terms of cooperation/competition. Conformity was measured by asking people to which extent they experienced a pressure to conform in the situation. We preferred this measure over asking participants whether they actually con￾formed, because people are generally reluctant to overtly admit their conformity to a group. For instance, Asch’s (1956) partici￾pants blamed their conformity on their own vision, rather than on the group pressure experienced during the experiment. Further￾more, there is evidence that people distort their memories of an act of conformity to make it appear as though they initially agreed (Griffin & Buehler, 1993). We assumed that this pressure to conform would reflect the subjective experience of threat or anx￾iety (Berns, Capra, Moore, & Noussair, 2010) that is ultimately resolved by conforming to the group and that it would therefore be a good proxy of conformity in the situation. Finally, we tested whether any effects of majority emotions on conformity pressure were mediated by perceived rejection. Method Participants. Sixty-eight participants were recruited for a study on disagreement in groups. Four participants indicated that they were unable to recall and describe an incident in which their opinion had differed from a group’s consensus, and their data could therefore not be used. The final sample consisted of 18 male and 46 female participants (Mage
22.11, range 18 –50 years). They were compensated with course credits or 7 euro. Materials and procedure. Critical incident prompt. Upon arriving in the laboratory, participants were seated individually behind a computer, which was used for presenting all instructions and recording answers. After completing a number of unrelated personality measures, the critical incident prompt was displayed on screen. Participants were asked to recall an episode in which a group decision had to be made, and their opinion had differed from that of the group. They were asked to describe as many details of the situation as they could. Acceptance/rejection. After participants had described the sit￾uation, the experiment continued with the display of a prompt asking the participant to indicate on a bipolar 7-point scale (from 1
rejected to 7
accepted) how they had felt in the situation they just described. The scale was reversed for use in the analyses, such that higher scores indicate stronger feelings of rejection. Majority emotions. The emotions expressed by the majority were measured using a list of 26 affective states.5 The items were presented in random order, and the participant was asked to indi￾cate on 7-point scales (from 1
not at all to 7
very much) how much of the respective emotion had been shown by the majority. An initial attempt to reduce the number of emotions measured by this questionnaire using principal factor analysis showed that the factor structure was highly dependent on which items were in￾cluded in the analysis. Therefore, we restricted our analysis to the six items related to happiness and anger, which were two clusters that emerged consistently in all factor analyses. Both the point of inflexion in the scree plot and Kaiser’s criterion agreed on two as the optimal number of factors. Focusing on these two factors, maximum likelihood factor analysis using varimax rotation re￾sulted in a clear distinction between enthusiasm, happiness, con￾tentment, and amusement on the one hand and anger and irritation on the other (see Table 1). The first factor was labeled as happiness (Cronbach’s
.83), and the second as anger (r
.70, p  .001). The emotion clusters were aggregated by averaging. Cooperativeness. Perceived cooperativeness of the situation was measured using three items. Two of these items (“To what extent did you pursue personal goals that differed from the group’s goals,” and “To what extent did your goals conflict with the group’s goals,” both reverse coded) asked about the goal structure without directly referring to cooperation and competition and were answered on 7-point scales (from 1
not at all to 7
very much). A third item directly asked how cooperative or competitive the situation had been on a bipolar 5-point scale (1
competitive, 3
neutral, and 5
cooperative). The scale composed of these items was internally consistent (Cronbach’s
.68), and the average of these items was calculated after z-transforming the individual items to correct for the different response scales. Conformity pressure. At the end of the experiment, we asked participants to what extent they had felt pressure to change their opinion or behavior in line with the group (from 1
none at all to 7
very much). Results Only four of the 68 participants were unable to recall an instance in which their opinion had differed from the majority’s, which suggests that the kind of situation under investigation is quite common. We found that a broad variety of situations was reported. Overall, many situations resembled the situation that we described in the vignettes used in Studies 1 and 2. To give an impression of the kind of stories that our participants wrote, we give two exam￾ples here: We wanted to buy a car to go on vacation with. I wanted a somewhat more expensive car, so we could sell it for more or less the same value after our vacation. I also liked the luxury and comfort of a better car. And the risk of a car breakdown would be smaller as well. The others wanted a cheap car though, so we wouldn’t have to worry about 5 The full list is jealousy, disappointment, shock, suspicion, disgust, tense, anger, boredom, contempt, sorry, guilt, nervousness, enthusiasm, happiness, surprise, compassion, relaxation, contentment, fear, relief, irri￾tation, shame, amusement, schadenfreude, indifference, and interest. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 269

270 HEERDINK.VAN KLEEF.HOMAN.AND FISCHER Table anger:B and simple cfiect The inde endent rement of anger and happiness allowed Emotion Happiness Anger otctie he ons jointly motions ent was at intermediate levels.Thus istent ed as d finally that's the one w with it. Study 2) .The indirect effects of majority emotions on conformity When I was in high school,we had to doa project in eness of th situation was tested as the moderator of the path bet ativeness of the ist h t is p rvals after bootstrapping (R=10.000)at mo lues on in th es thu ery a ion pported for both 0.19 The exter nt to which participan eously.When acc ion wa tailed.In situations of ange 004.0.26 00 =20.on conformity pressure was me by felt acceptance Together,these findings indicate that in situations per -0.206 6.0 the were found.Irs 55. Conformitypr 0.35,58)=-1.82.p= 35 1 SD),perceivec omcesitive (-1 SD)Cooperative (+1 SD) creased simple effect B =051 (58)=2 03=02 one-tailed. ettings (-1 SD),no relation ween m sure(Sud女y3)

keeping it neat and would maybe be able to keep it. We had quite some discussion, and finally we decided to go for the cheaper car, so that’s the one we have now. All of us, including myself, are pleased with it. Quite a few situations were much less cooperative and friendly, as this story shows: When I was in high school, we had to do a scientific project in groups of four students. I was in a group with three of my friends, and we were talking about the topic for our project. I didn’t find the topics they proposed very interesting. I really wanted a topic that I found interesting and that I could find out more about. So when I proposed my topic (something related to parapsychology), they just laughed at me. Then, I tried to persist because I wanted to show them it is not such a stupid topic. They really didn’t want my topic, and I didn’t really want theirs . . . In the end, one of the other group members agreed with me, and the group split. Even outside class, we got into arguments about nothing. I was angry about their reaction, and I found them childish. In the end, the subgroups worked almost in isolation on two topics that were only minimally related. Acceptance/rejection. The extent to which participants felt rejected was predicted by anger and happiness both separately and simultaneously. When acceptance/rejection was regressed on both emotions simultaneously, the results indicated that, as expected, the more the majority reaction had been angry, the more partici￾pants felt rejected,
0.47, t(61)
4.32, Rp 2
.23, p  .001 (one-tailed). Also as expected, the happier the reaction from the majority had been, the less participants had felt rejected (i.e., more accepted),
0.20, t(61)
1.86, Rp 2
.07, p
.03 (one￾tailed). No interactions with the cooperativeness of the situation were found, |t|s  .58, ps  .55. Conformity pressure. As expected, the relation between ma￾jority emotions and conformity pressure depended on the per￾ceived cooperativeness of the situation. When conformity pressure was regressed on the interactions between majority emotions and the perceived cooperativeness of the situation, both the Anger  Cooperativeness,
0.35, t(58)
1.82, p
.04, one-tailed, and the Happiness  Cooperativeness,
0.32, t(58)
1.97, p
.03, one-tailed, interactions emerged as significant predictors. In cooperative settings ( 1 SD), perceived conformity pressure increased as the amount of anger expressed by the majority in￾creased, simple effect:
0.51, t(58)
2.03, p
.02, one-tailed, and the amount of happiness expressed by the majority decreased, simple effect:
0.40, t(58)
1.89, p
.03, one-tailed. In competitive settings (1 SD), no relation between majority emo￾tions and perceived conformity pressure was found, simple effect anger:
0.19, t(58)
0.85, p
.40; and simple effect happiness:
0.24, t(58)
1.19, p
.24. The independent measurement of anger and happiness allowed us to interpret the effects of both emotions jointly. Summarizing the previously described information, conformity pressure was highest in cooperative situations in which the majority expressed anger and little happiness, and the least conformity pressure was found in cooperative settings in which the majority expressed happiness and little anger (see Figure 3). Conformity pressure in situations perceived as competitive and in situations in which both emotions were equally present was at intermediate levels. Thus, consistent with the hypotheses, it is in situations perceived as cooperative that more anger is associated with more conformity pressure, and more happiness is associated with less conformity pressure. Mediation analysis. To find out whether felt rejection could account for the relation between majority emotions and conformity pressure in situations perceived as cooperative, we conducted a moderated mediation analysis (Preacher et al., 2007; see also Study 2). The indirect effects of majority emotions on conformity pressure through acceptance/rejection were estimated separately for anger and happiness. The perceived cooperativeness of the situation was tested as the moderator of the path between accep￾tance/rejection and conformity pressure (see Figure 1 for the general model). Because the cooperativeness of the situation was a continuous variable, we adopted an approach similar to simple slope analysis (Aiken et al., 1991). We calculated BCa confidence intervals after bootstrapping (R
10,000) at moderator values one standard deviation above and below the mean. These values thus reflected relatively cooperative and competitive situations, respec￾tively. Our hypothesis that felt acceptance/rejection would mediate the path from majority emotions to conformity pressure in situa￾tions seen as cooperative was supported for both anger, B
0.19, 95% BCa CI [lower limit: 0.08], p
.002, one-tailed, and happi￾ness, B
0.17, 95% BCa CI [upper limit: 0.04], p
.008, one-tailed. In situations seen as more competitive, neither the effect of anger, B
0.09, 95% BCa CI [0.24, 0.04], p
.15, nor the effect of happiness, B
0.07, 95% BCa CI [0.04, 0.26], p
.20, on conformity pressure was mediated by felt acceptance/ rejection. Together, these findings indicate that in situations per￾Table 1 Factor Loadings of Reported Majority Emotions on Two Principal Factors After Varimax Rotation (Study 3) Emotion Happiness Anger Happy .79 Content .79 Enthusiastic .73 Amused .63 Irritated .91 Anger .75 3.0 3.5 4.0 4.5 5.0 Competitive (-1 SD) Cooperative (+1 SD) Pressure to Conform Majority Emotion No Emotion Happiness Anger Both Figure 3. The relation between the majority’s anger and happiness and conformity pressure in situations perceived as relatively cooperative and competitive. Note that two independent effects were combined to illustrate the additive effects of anger and happiness in relation to conformity pressure (Study 3). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 270 HEERDINK, VAN KLEEF, HOMAN, AND FISCHER

ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 271 ween participants would hinder the effectiveness of the emot task to a partic a die roll de Discussion The results of this study replicate our finding that majority's the end of the(scale ranging from Never nd nfirmed that otivational nersnective on conformity we showed that in situa that are per nity pre Groups were nd le our all thre ship,which is congruent with the idea that ormity is not a tion to this random ment was that the ever aligned with salient Although the and a of our ses based on p and procedure. nent cons mes.1974). which is a p ask that Studv 4 hey may help es of survival if a pe n the s.we showed tha aiority's anger leads ianrindividualofelrejected.wh ch in tur c items the with the as inpu to which th study in which ltem generation. laboratory.the thre tionnaires that were lated to the pre ent h lesent que the nany items as possible that could be aseful in in such a circum outcome of the reiected.whichin u eads oneto .the influence of the individual ide hase had ended particinant I be re ely high ng the individual lists allowed,but Method ion of ideas,the instructions also m Thirty-th (99 partic duplicate,useless.or dangerous

ceived as cooperative, the higher conformity pressure experienced when the majority expressed more anger and/or less happiness was due to the fact that the group member felt rejected. In situations perceived as competitive, the majority’s angry and happy reactions to deviance were still associated with felt acceptance/rejection, but no relation with felt conformity pressure was found. Discussion The results of this study replicate our finding that majority’s emotions are associated with the extent to which a deviant indi￾vidual feels accepted or rejected. Furthermore, consistent with our motivational perspective on conformity, we showed that in situa￾tions that are perceived as cooperative, more conformity pressure was experienced to the extent that more anger and less happiness were expressed. Furthermore, supporting our general model (see Figure 1), this relation was mediated by feelings of rejection. In situations perceived as competitive, we did not find this relation￾ship, which is congruent with the idea that conformity is not a meaningful way of showing good group membership in a compet￾itive setting. Most important, we found these results across a wide range of social situations, which increases confidence in the gen￾eralizability of the findings from Studies 1 and 2. Although the critical incidents approach allows for high ecolog￾ical validity and a test of our hypotheses based on people’s recollections of actual situations, it also has several drawbacks, the most important of which is that it yields only correlational data. To address these limitations, we set out to replicate and extend these findings in two behavioral experiments. Study 4 In the previous studies, we showed that a majority’s anger leads a deviant individual to feel rejected, which in turn may lead this individual to experience a pressure to conform. In Study 4, we aimed to extend these findings by investigating whether anger, expressed in a cooperative setting, can lead to behavioral confor￾mity by inducing feelings of rejection. For this purpose, we con￾ducted an experimental group study in which groups, consisting of three participants, worked on a group problem-solving task. We manipulated the emotion expressed by the majority by instructing two of these participants to express either anger or happiness in response to ideas voiced by the third participant. A nonemotional condition, in which participants were instructed not to show their emotions, was also included as a reference condition. In this study, conformity was operationalized as the relative influence of this third participant (faced with either a happy or an angry majority) on the outcome of the group task. We reasoned that if being faced with an angry majority leads one to feel rejected, which in turn leads one to conform, the influence of the two angry group members should be relatively high relative to the third participant’s influence. Thus, we expected participants who were faced with an angry majority to have relatively less influence in their group than participants who were faced with a happy majority or those in the nonemotional condition. Method Participants and design. Thirty-three groups (99 partici￾pants, 22 men, Mage
20.99, range 15–29) participated in the experiment, which was advertised as a group creativity task. In exchange for their participation, participants received either course credit or 10.50 euro. Because we were concerned that familiarity between participants would hinder the effectiveness of the emotion manipulation, we invited four participants for each session, which allowed us to assign an individual backup task to a participant who coincidentally knew another participant (otherwise, a die roll de￾cided which participant would receive different tasks). A check at the end of the experiment (scale ranging from Never have seen this person to Best friend) confirmed that in the final sample, no pair of participants indicated more mutual familiarity than Have seen but have not spoken to this person. Groups were randomly assigned to one of three conditions: majority angry, majority happy, or majority nonemotional. Within groups in the former two conditions, two randomly selected par￾ticipants received an emotion instruction, and the remaining par￾ticipant (the “focal participant,” described later) received instruc￾tions to show no emotion. In the nonemotional condition, all three participants received the no emotion instruction. The only excep￾tion to this random assignment was that the manipulation was never aligned with salient demographic characteristics to avoid creating a salient diversity fault line (Homan, van Knippenberg, van Kleef, & De Dreu, 2007b). For instance, if the group consisted of one female and two male participants, it was always one of the male participants who received the no emotion instruction. Materials and procedure. The experiment consisted of a modified version of the desert survival task (Lafferty, Eady, & Elmers, 1974), which is a problem-solving task that is used in group research. In the original version of the task, the goal is to rank a list of items (e.g., a knife) according to the extent to which they may help promote chances of survival if a person is stranded in the desert. We used the Homan, van Knippenberg, van Kleef, and De Dreu (2007a) version of this task, in which participants do not receive a pre-existing list of items, but generate items them￾selves. By having participants first generate items individually, followed by group-wise selection of the best ideas with the indi￾vidual lists as input, we could estimate the extent to which the emotions expressed in the group influenced individual contribu￾tions to the group product. Item generation. Upon arrival in the laboratory, the three group members were seated separately and filled out a number of personality questionnaires that were unrelated to the present hy￾potheses. These questionnaires were followed by an introduction to the desert survival situation and an instruction to generate as many items as possible that could be useful in in such a circum￾stance. The only constraint was that it should be possible for one person to carry the item. Participants were given 10 min to gen￾erate items. Emotion instruction. After the individual idea generation phase had ended, participants received written instructions for the group task. These explained that the goal of the group task would be the group-wise creation of a list of as many ideas as possible, using the individual lists as input (generating new ideas during the group interaction was explicitly allowed, but too few groups made use of this possibility to analyze this variable). As an encourage￾ment for critical evaluation of ideas, the instructions also men￾tioned that a bonus of 75 euro (25 euro for each participant) would be awarded to the group that generated the most ideas that were not duplicate, useless, or dangerous. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 271