Do We Relatively or Absolutely Overestimate Rare Events?

Abstract

How good are we at estimating the frequency of rare events? Previous evidence has been mixed: While some studies report well-calibrated frequency judgments, others found rare events to be overestimated. We examined whether the distinctiveness of a rare event fosters overestimation. Distinctiveness may result from exceptional valence, statistical infrequency, or the joint occurrence of two distinct features. Such paired distinctiveness has been found to be particularly salient. We conducted two experiments to investigate the impact of paired distinctiveness and valence on frequency estimations of rare events by combining two paradigms from the decision-making literature and the social psychology literature on stereotypes. The results indicate that rarity alone does not necessarily result in overestimation; rather, the combination of statistical infrequency and small sample sizes (i.e., paired distinctiveness) led participants to overestimate the frequency of a rare event. We were able to rule out alternative explanations such as regression to the mean, but the exact role of outcome valence needs further investigation.