Do In-Group Biases Lead to Overconfidence in Performance? Experimental Evidence

Abstract

Psychologists have long identified the tendency of humans to overestimate their skill relative to their peers (overplacement). We investigate whether this phenomenon is exacerbated by group affiliation: social identity theory predicts people evaluate in-group members more positively than out-group members, and we hypothesized that this differential treatment may result in greater overplacement when interacting with an out-group member. We tested this hypothesis with 301 US voters affiliated with either the Republican or Democratic party in the run-up to the 2020 Presidential election, a time when political identities were salient and highly polarized. We found there is a higher tendency for overplacement when faced with an out-group opponent than with an in-group opponent. Decomposition analysis suggests this difference is due to underestimating the opponent, as opposed to overestimating one's own performance to a higher degree. Moreover, any tendency to incur in overplacement is mitigated when faced with an opponent with the same political-identity relative to one with a neutral one. While group affiliation biases initial priors, such effect is unchanged when participants are asked to update their beliefs.