Mad enough to see the other side: Anger and the search for disconfirming information

Abstract

The current research explored the effect of anger on hypothesis confirmation—the propensity to seek information that confirms rather than disconfirms one's opinion. We argued that the moving against action tendency associated with anger leads angry individuals to seek out more disconfirming information than sad individuals, attenuating the confirmation bias. We tested this hypothesis in two studies of experimentally primed anger and sadness on the selective exposure to hypothesis confirming and disconfirming information. In Study 1, participants in the angry condition were more likely to choose disconfirming information than those in the sad or neutral condition when given the opportunity to read more about a social debate, and reading the disconfirming information affected their subsequent attitude. Study 2 measured participants' opinions and information selection about the 2008 US Presidential Election and their desire to “move against” a person or object. Participants in the angry condition reported a greater tendency to oppose a person or object, which resulted in the attenuation of the confirmation bias.

Keywords:

• Anger
• Action tendency
• Selective information
• Confirmation bias
• Argumentative

Notes

¹The mediating results also generalise when using the approach advocated by Baron and Kenny (1986) and Judd and Kenny (1981) for examining mediation in repeated-measure models. This entails including the covariate to examine whether the effect of the independent variable is reduced while controlling for the mediator. First, we found that participants in the Angry Condition (M=3.85, SD=1.77) reported having a greater desire to move against a person or object than those in the Sad Condition (M=1.53, SD=1.01), t(87) = 7.67, p<.001. Furthermore, results of a 2 Emotion (angry, sad)×2 Selection (confirming, disconfirming) ANCOVA revealed that when we controlled the tendency of moving against, the previously observed interaction between Condition and Selection on the selection of information was reduced to non-significance, F(1, 84) = 0.09, ns, η²=.00.