Abstract
Shame has been found to promote both approach and withdrawal behaviours. Shame theories have not been able to explain how shame can promote such contrasting behaviours. In the present article, the authors provide an explanation for this. Shame was hypothesised to activate approach behaviours to restore the threatened self, and in situations when this is not possible or too risky, to activate withdrawal behaviours to protect the self from further damage. Five studies with different shame inductions and different dependent measures confirmed our predictions. We therefore showed that different behavioural responses to shame can be understood in terms of restore and protect motives. Implications for theory and behavioural research on shame are discussed.
Notes
1All scenarios are translated from the original Dutch version.
2Reversed mediational analyses with Situational Choice as a mediator and with Restore and Protect as dependent variables did not show a mediating effect. The significant effects of Emotion condition on Restore, β=.35, t=5.48, p<.01, and on Protect, β=.14, t=2.12, p<.05, remained significant when Situational Choice was added as a mediator, β=.18, t=3.48, p<.01 for Restore and β=.28, t=4.81, p<.01 for Protect. When Willingness to Perform was added as a mediator, the effects of Emotion condition also remained significant, β=.24, t=5.14, p<.01 for Restore and β=.25, t=4.24, p<.01 for Protect.
3Reversed mediational analyses with Willingness to perform as a mediator and with Restore and Protect as dependent variables did not show a mediating effect. The significant effects of Emotion condition on Restore, β=.56, t=5.37, p<.01, and on Protect, β=.52, t=4.82, p<.05, remained significant when Willingness to perform was added as a mediator, β=.46, t=4.76, p<.01 for Restore and β=.56, t=5.06, p<.01 for Protect.
4Reversed mediational analyses with Achievement approach and Achievement avoidance as mediators and with Restore and Protect as dependent variables did not show mediating effects. When reported shame, Achievement approach and Achievement avoidance were added to the analyses, the effect of Achievement approach on Protect became nonsignificant, β = − .02, t=0.19, p=.85, and the effect of Achievement avoidance on Restore became nonsignificant, β = − .13, t=0.77, p=.45.
5As can be seen in , participants in Studies 2 to 5 also reported a heightened degree of guilt. However, guilt cannot explain the effects on the restore motive and on approach behaviours in any study. In all analyses, adding reported guilt as a covariate showed nonsignificant effects, while the effects of the Emotion condition in Experiments 2 and 5, and the effects of reported shame in Experiments 3 and 4, remained significant. In Experiment 2, the effect of Guilt on Restore was nonsignificant, F(1, 61) = 0.03, p=.87, η2<.01, whereas the effect of Emotion condition on Restore remained significant, F(1, 61) = 14.24, p<.01, η2=.19. The effect of Guilt on Willingness to perform was also nonsignificant, F(1, 61) = 0.18, p=.67, η2<.01, while the effect of Emotion condition remained significant, F(1, 61) = 3.33, p=.05, η2=.05. In Experiment 3, the effect of Guilt on Achievement approach was nonsignificant, β=.02, t=0.15, p=.88, while the effect of Shame remained significant, β=.29, t=2.03, p<.05. In Experiments 4 and 5, the effects of Guilt on Repair (Experiment 4), β=.09, t=0.97, p=.33, and on Willingness to perform (Experiment 5), F(1, 133) = 0.23, p=.63, η2<.01, were both nonsignificant.