ON THE SOCIAL INELUENCE OF EMOTIONS IN GROUPS 277 ally one mor his or her indicated that within the that arlicrating,adif essed by the majority(1 ou of or 57.1)than afte rated on ay-point scal 106 Wal's glmpeacpaseehmkadncaaiko b252p=01 25.ocase in the prototypical condition (anger)B-0.04. Wald's :=000 Results To investigate whether peripheral members conformed after a n be 3 aiority emotion ma as the ind dent ormity as the dependen 162 (84)=12 76.R2=Similarly.partic d that the tin han afte seein ns (M 60.SD=1.42 -10.000 resamples).As in trap r nples yie ng 9 ct ef that for p ripheral group memb ers.0 was not en d seen that is was not (M =3.18.SD=0.94).B=0.72 95%C1 the 02 58. on-tailed.whichis stent with the idea that peripheral mem the mani ed more rejection s expected, in the confidenc rating (M 2)than heyh BC. tions and 001.one-tailed.Wh P xplained by felt acceptance/rejection d in the model.neither a nain effect of prototypicality was found ot po alcula differen ure o ting that the exte ictpeoptfehaCcspledorrejected all pa As can be seen from Figure 4.peripheral 100- wewe able to use the raw scores for th nore alter an angry reactio ndCeLpobilregresionnieldedasgnicnEmoionXProto rd deviati focal na was ed by th on of the group.As can b inher ity t nity.B -0.98.t56 1.92p= 03.one-tailed ple effect d that,as p oe afte 49.SD19g Using the logit link functi n for these nly ch =051 e We 's T3Eindividual rating round), then two new paintings, followed by the focal painting, and finally one more familiar painting. To reduce the possibility that a participant would attempt to recall his or her earlier rating, a different rating scale was used. Instead of the slider that was used at T1 and T2, paintings were rated on a 9-point scale (from 1
ugly to 9
beautiful). After completing some more questions about art, participants were thanked and given a link to a full debriefing. Results Manipulation checks. Participants reported that their fellow group members had shown more anger when they had received angry emoticons from the majority (M
5.55, SD
1.60) than when they had received happy emoticons (M
1.84, SD
1.06),
1.62, t(84)
12.76, R2
.66, p  .001. Similarly, partici￾pants indicated that their fellow group members had shown more happiness after seeing the smiling emoticons (M
4.98, SD
1.25) than after seeing angry emoticons (M
2.60, SD
1.42),
1.33, t(84)
8.29, R2
.45, p  .001. Participants also indicated that they were more prototypical of their group when they had seen that their personality was similar to their fellow group members’ personalities (M
4.07, SD
1.33) than when they had seen that is was not (M
3.18, SD
0.94),
0.72, t(84)
3.58, R2
.13, p
.001. There were no cross-effects of one manipulation on the other check and no interactions. Hence, the manipulations were successful. Acceptance/rejection. As expected, participants felt more re￾jected when the majority had expressed anger about the partici￾pant’s rating (M
4.16, SD
0.82) than when they had expressed happiness (M
2.99, SD
0.83),
1.16, t(84)
6.60, R2
.34, p  .001, one-tailed. When the prototypicality manipulation was included in the model, neither a main effect of prototypicality,
0.02, t(84)
0.14, p
.89, nor an interaction of the two manipulations,
.27, t(84)
1.52, p
.13, was found, indicating that the extent to which people felt accepted or rejected was not dependent on the prototypicality manipulation. T2 conformity. As can be seen from Figure 4, peripheral members. conformed more after an angry reaction than after a happy reaction, but this was not the case for prototypical members. Indeed, probit regression8 yielded a significant Emotion  Proto￾typicality interaction on T2 conformity, B
1.02, Wald’s z
1.70, p
.04, one-tailed. Simple slopes indicated that within the peripheral condition, conformity was higher after anger had been expressed by the majority (12 out of 21, or 57.14%) than after happiness had been expressed (four out of 21, or 19.05%), B
1.06, Wald’s z
2.52, p
.01, one-tailed, whereas this was not the case in the prototypical condition (anger: five out of 20, or 25.00%; and happiness: five out of 21, or 23.81%), B
0.04, Wald’s z
0.09, p
.93. To investigate whether peripheral members conformed after an angry reaction because they felt rejected, we conducted a moder￾ated mediation analysis (Preacher et al., 2007; see also Study 3). The majority emotion manipulation was specified as the indepen￾dent variable, conformity as the dependent variable, felt accep￾tance/rejection as the mediator, and prototypicality as a moderator of the path from felt acceptance/rejection to conformity (see Figure 1). The strength of the indirect effect was estimated separately for prototypical and peripheral members by calculating BCa intervals after bootstrapping (R
10,000 resamples). As in Study 3, boot￾strap resamples yielding empty cells were discarded, leaving 9,704 resamples to estimate the strength of the indirect effect. The results showed that for peripheral group members, 0 was not enclosed in the 95% CI of the indirect effect of the majority emotion on conformity, B
0.54, 95% BCa CI [lower limit: 0.05], p
.02, one-tailed, which is consistent with the idea that peripheral mem￾bers conformed more after an angry reaction than after an enthu￾siastic reaction because they felt more rejected. For prototypical group members, 0 was enclosed in the confidence interval, B
0.24, 95% BCa CI [0.92, 0.45], p
.46, and any relation between majority emotions and conformity could therefore not be explained by felt acceptance/rejection. T3 conformity. Because in the posttest, we employed a 9-point scale instead of the slider that was used at T1 and T2, it was not possible to calculate difference scores as the measure of conformity at T3. As nearly all participants initially rated their focal painting within three scale points of the critical value (on the 100-point slider),9 we were able to use the raw scores for these participants. The absolute T3 ratings (N
60) were subjected to a square-root transformation to correct for nonnormality. Reported means and standard deviations are untransformed values. Analysis of the ratings of the focal painting at T3 showed that even after 3 weeks, participants’ evaluation of the focal painting was influenced by the emotional reaction of the group. As can be seen from Figure 5, the emotions of the majority had more effect on peripheral members’ conformity than on prototypical members’ conformity,
0.98, t(56)
1.92, p
.03, one-tailed. Simple effects confirmed that, as predicted, peripheral members conformed more after an angry reaction (M
4.79, SD
1.93) than after a happy reaction (M
3.40, SD
1.88) to their original 8 Using the logit link function for these analyses instead, the reported p values are virtually identical (deviations in the .005–.01 range in both directions). The only change to the interpretation is that the interaction becomes marginally significant (p
.05). 9 One participant had initially rated the focal painting at 9 scale points from the critical value. We chose to discard data from this participant, as we were not convinced that we could compare this participant’s T3 rating to the other participants’ ratings, nor that we could simply adjust this person’s T3 rating to make it comparable. 0 20 40 60 Peripheral Prototypical Conformity at T2 Majority Emotion Happiness Anger Figure 4. Proportion of participants conforming in each condition (Study 5). T2
Time 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. ON THE SOCIAL INFLUENCE OF EMOTIONS IN GROUPS 277