Sex differences in anger-related behaviour: Comparing expectancies to actual behaviour

Abstract

Expectancies about and the reactions of men and women to a provocation were compared to self- and observer ratings of the actual behaviour in response to the provocation. In Study 1, men and women were provoked by the experimenter during a laboratory task. In Study 2, participants were presented with video clips of the provocation and estimated how a man or a woman would likely react to it. The man was expected to react more aggressively than the woman. In their actual behaviour, however, men were more likely to react with humour or by distancing than were women. The woman was expected to react more submissively than the man, which was inaccurate with respect to both the observer ratings and the self-ratings.

Acknowledgements

This research was supported by Grant We 1779/5–1 from the German Research Foundation (DFG) awarded to HW.

The authors would like to thank Juliane Kranzpiller and Alexander Kaps for their help with the data collection. We are very much indebted to Rainer Reisenzein for his many insightful comments on an earlier version of this paper. We are also grateful to Ursula Hess and three anonymous reviewers for their helpful suggestions and comments on an earlier version of this article.

Notes

¹We are grateful to Wilhelm Glaser for allowing us to manipulate the computerised Stroop test that he has developed. The program manipulation was provided by Conrad Schimke and Alexander Kaps.

²The goals measured by the ARGI reflect the objectives that people pursue when responding to anger-eliciting interpersonal events—for instance, regulating affect or asserting social norms. In the present report, only the anger-related reaction subscales of the ARGI were used, since the two studies focus on observable behaviour (including observer ratings), not intentions.

³Additional analyses were performed to study the relationship between observer-rated anger-related reactions and the duration of the actual interaction between participant and experimenter, as measured from the participants’ response to the experimenters’ first provocation until the end of his or her response to the experimenters’ final provocative remark. The range of the duration of interactions was 41 to 213 seconds, with a mean length of 81°s (SD=24.6). The results revealed no significant correlations between duration of interaction and anger intensity, neither for the self-ratings nor for the observer ratings. However, small but meaningful correlations were found between the duration of the interaction on the one hand and self- and observer ratings of anger-related reactions on the other. Specifically, the duration of the interaction correlated significantly with self-ratings of feedback (r=.23, p<.05). All other correlations between duration and self-rated reactions were nonsignificant. When outliers (n=5, designated as scores that fell outside two standard deviations) were omitted from the analyses (leading to a compressed range of interaction length between 41 and 131°s), there was only a significant negative correlation between duration and distancing (r=−.29, p<.01). With respect to the observer-ratings, there were significant positive correlations between duration of interaction and both venting (r=.23, p<.05) and feedback (r=.32, p<.01), as well as a significant negative correlation between duration and submission (r=−.35, p<.01). When the five outliers were omitted from the analyses, there was a significant positive correlation between duration and feedback (r=.35, p<.01), and a significant negative correlation between duration and submission (r=−.31, p<.01). To examine whether the outliers influenced the results of the analyses for the self- and observer ratings of the participants’ anger-related reactions, all analyses were repeated without the outliers. This did not lead to different results with respect to the significant findings.

⁴For the self-ratings, correlations between the situational reactions scales and post-induction state anger revealed significant positive correlations between venting and anger (r=.49, p<.01), and rumination and anger (r=.36, p<.01). Significant negative correlations were found between post-induction state anger and distancing (r=−.37, p<.01), and between state anger and humour (r=−.27, p<.01). These results are in line with previous findings showing positive associations between anger and rumination, and negative correlations between anger and distraction (Rusting & Nolen-Hoeksema, 1998).

⁵In addition to the significant effects for sex of participant and sex of experimenter, which were the main questions of the present study, significant main effects for sex of observer were found for venting, F(1, 119) = 15.13, p<.01, η²=.11, feedback, F(1, 119) = 15.03, p<.01, η²=.11, distancing, F(1, 119) = 42.46, p<.01, η²=.26, humour, F(1, 119) = 8.17, p<.01, η²=.06, and submission, F(1, 119) = 30.17, p<.01, η²=.20, respectively. For submission, male observers gave higher ratings than did female observers. In all other cases, female observers gave higher ratings than did male observers. Significant sex of participant × sex of observer interactions were obtained for distancing, F(1, 119) = 11.17, p<.01, η²=.09, and humour, F(1, 119) = 4.45, p<.05, η²=.04, indicating that for both reactions, the significant differences between men and women were more pronounced for the female observer ratings, t(121) = 5.01, p<.01, d=0.90 (distancing), and t(121) = 3.02, d=0.55 (humour), than for the male observer ratings, t(121) = 3.02, p<.01, d=0.65 (distancing), and t(121) = 2.34, p<.01, d=0.42 (humour), as indicated by the smaller effect size. A significant sex of experimenter × sex of observer interaction was found for feedback, F(1, 123) = 4.88, p<.05, η²=.04, indicating that female observer ratings did not differ as a function of sex of experimenter, t(121) = − 0.60, ns, whereas male observer ratings were higher when the participants interacted with the female compared to the male experimenter, t(121) = − 2.17, p<.05.

⁶For the observer ratings, significant positive correlations were found between observer ratings of the participants’ state anger and venting (r=.84, p<.01), and between anger and feedback (r=.23, p<.01). There were significant negative correlations between anger and submission (r=−.52, p<.01), distancing (r=−.78, p<.01), and humour (r=−.32, p<.01). These findings (which are similar to those found for the self-ratings) suggest that from the observers’ perspective, anger intensity was positively associated with the open expression of anger, and negatively with reactions that implicated disengagement.

⁷In self-reports, trait anger is typically moderately to highly correlated with a higher tendency for anger-out and a lower tendency for anger-control (e.g., Deffenbacher et al., 1996; Spielberger, 1988). When faced with an ambiguous situation (such as predicting the behaviour of an anonymous target), there may be a tendency to impose one's own reaction tendencies on the target's anticipated reactions. Therefore it can be assumed that participants with higher anger-proneness may be more likely to expect the open expression of anger as a more likely response and controlling anger as a less likely response to the provocation. In fact, there was a positive correlation between the participants’ trait anger and anticipating venting as a likely response (r=.21, p<.05). However, all other correlations between participants’ trait anger and expected reactions were nonsignificant. To control for possible effects of anger-proneness on the expected anger-related behaviour of another person, all 2 (sex of target) × 2 (sex of experimenter) × 2 (sex of participant) ANOVAs were repeated with the participants’ trait anger as the covariate; these analyses yielded similar results.