An Unlucky Feeling: Persistent Overestimation of Absolute Performance with Noisy Feedback

How does overconfidence arise and persist in the face of experience and feedback? We examine experimentally how individuals' beliefs about their absolute, as opposed to relative, performance on a quiz react to noisy, but unbiased, feedback. Participants believe themselves to have received `unlucky' feedback and they overestimate their own scores, but they exhibit no overconfidence in non-ego-relevant beliefs---in this case, about others' scores. Unlike previous studies of relative performance estimates, we find this to be driven by overconfident priors, as opposed to biased updating, which suggests that social comparisons contribute to biased information processing. While feedback improves performance estimates, this learning does not translate into improved estimates of subsequent performances. This suggests that people use performance feedback to update their beliefs about their ability differently than they do to update their beliefs about their performance, contributing to the persistence of overconfidence.