Group-Based Emotions: The Impact of Social Identity on Appraisals, Emotions, and Behaviors

Abstract

Because group-based emotions are rooted in the social identity of the perceiver, we propose that group-based emotions should be sensitive to changes in this social identity. In three experiments, young women reported feeling more anger, fear, and disgust toward Muslims when their identity as women had been made salient, in comparison with various control conditions where their identity as young adults, as social sciences students, their personal identity, or no identity had been made salient. These effects were mediated by appraisals of intergroup threats. In Experiment 3, the salience of the woman social identity also increased intentions to avoid Muslims.

ACKNOWLEDGMENTS

The present research was made possible by an ECRP grant from ESF and FRS/FNRS (grant 2.4531.06 FRFC). We thank Amy Cuddy, Stéphanie Demoulin, Thomas Kessler, Bernard Rimé, and Cátia Teixeira, as well as the members of the Center for the Study of Social Behavior of the Université catholique de Louvain, for their helpful feedback on earlier drafts of this article, and Bart Vanhaelewyn, Jan Claeys, Dieter Vandebroeck, and Bram Vanhoutte for their assistance in data collection.

Notes

¹The authors emphasized either participants’ American or their student identity, and they measured anger and respect toward Muslims and the police. A three-way interaction between social identity (American vs. student), target group (Muslims vs. the police), and emotion type (anger vs. respect) seems to provide evidence for the effect of social identity salience on group-based emotions. However, additional data presented by the authors suggest that emotions toward the police were affected by a recategorization of the police as ingroup members in the American condition. The remaining two-way interaction between participants’ social identity (American vs. student) and emotion type (anger vs. respect) for emotions toward Muslims was only marginally significant (p < .09), and none of the simple effects of categorization were significant (D. G. Ray, personal communication, May 18, 2009).

²The pity scale clearly lacks reliability. We therefore use only the item “pity” to represent this emotion (but conclusions are identical if the other item or the two-item scale is used).

Note. All appraisals and emotions were used as the criterion in a regression analysis with social identity condition represented by two orthogonal contrasts. The first contrast compared the woman identity to the other two conditions. The second contrast compared the no identity to the young adult identity condition. Parameters are standardized regression coefficients.

*p < .05. **p < .01. ***p < .001. ^† p < .10.

Note. All appraisals and emotions were used as the criterion in a regression analysis with social identity condition represented by two orthogonal contrasts. The first contrast compared the woman identity to the other two conditions. The second contrast compared the personal identity to the social sciences student identity condition. Parameters are standardized regression coefficients.

*p < .05. **p < .01. ***p < .001. ^† p < .10.

³For exploratory reasons, we also measured the intergroup behavioral tendencies from the BIAS map (Cuddy et al., 2007): active harm, passive harm, active facilitation, and passive facilitation. There were no statistically significant effects of the social identity manipulation for these measures.

⁴Note that Experiment 3 was conducted in French, whereas Experiments 1 and 2 were run in Dutch. A preliminary study in French showed that the translation of the two-item scale for disgust was not reliable, which is why we added a third item.

Note. Columns 2 and 3 present the means (standard deviations between brackets) for all dependent variables, by condition. The third column reports the standardized regression coefficient for the effect of the social identity manipulation (coded −1 for control and 1 for the woman condition), controlling for ethnocentrism.

*p < .05. **p < .01. ***p < .001. ^† p < .10.

⁵Ethnocentrism had a significant effect on most appraisals, emotions, and behavioral intentions, but because it is only a control variable we do not discuss these effects here.

⁶We thank an anonymous reviewer for suggesting this.