“Fury, us”: Anger as a basis for new group self-categories

Abstract

We tested the hypothesis that shared emotions, notably anger, influence the formation of new self-categories. We first measured participants' (N = 89) emotional reactions to a proposal to make university assessment tougher before providing feedback about the reactions of eight other co-present individuals. This feedback always contained information about the other individuals' attitudes to the proposals (four opposed and four not opposed) and in the experimental condition emotion information (of those opposed, two were angry, two were sad). Participants self-categorised more with, and preferred to work with, angry rather than sad targets, but only when participants' own anger was high. These findings support the idea that emotions are a potent determinant of self-categorisation, even in the absence of existing, available self-categories.

Keywords:

• Emotion
• Self-categorisation
• Social identity
• Attitudes

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary material is available via the ‘Supplementary’ tab on the article's online page (http://dx.doi.org/10.1080/02699931.2015.1023702).

Notes

¹ A number of other measures relating to intergroup orientations were included after the measures reported here. These included scales of opinion support (two items) and action support for (two items) and against (two items) the proposal, collective efficacy for (two items) and against (two items) the proposal, and expected validation from team members (four items). These measures are not analysed here because they are more relevant to hypotheses regarding intergroup behaviour, and collective action specifically; they have also not been reported anywhere else.

² Participants' own anger and sadness were significantly correlated (r = .715). Including own sadness as a covariate enabled us to study the effects of own anger controlling for its shared variance with own sadness. Regression analyses including condition, own anger and own sadness as predictors confirmed that no issues of multi-collinearity emerged when entering own anger and own sadness together; all tolerance statistics>.488.

³ The selection of two individuals without replacement arguably creates interdependence in the selections. For this reason, we also ran a bootstrapped (non-parametric) version of the analysis. This yielded almost identical results; specifically, the three-way interaction between targets' emotion, condition and own anger was still highly significant, B = − .21, standard error = .07, 95% confidence intervals = − .352 and − .062.