Overconfidence, base rates and outcome positive/negative of predicted events.

Overconfidence is said to occur when a person’s confidence in a series of predictions exceeds the level of accuracy achieved. In this experiment, questionnaire items were selected according to the objective base rates of occurrence of a series of events and their outcome positivity/negativity for subjects in a pilot study. The 98 subjects in the main experiment predicted whether they would experience each event within the next week and rated their confidence in their predictions. Predictions were compared with responses to a follow-up questionnaire a week later to determine accuracy. Significant overconfidence occurred, but it was greater for positive-outcome than negative-outcome items, and the results revealed a curvilinear relationship between base rates and overconfidence, with maximum overconfidence at intermediate base rate levels and underconfidence at both extremes of the base rate range. Subjects tended to overestimate base rates below 40 per cent and to underestimate higher base rates.