Facing uncertainty in the game of bridge: A calibration study

Abstract Previous studies have shown a strong tendency toward overconfidence in peoples' probability assessments. Even experts are often poorly calibrated. The present paper suggets that quality of calibration is largely determined by the extent to which the cognitive processes required for repeated probability assessments are similar. One task that satisfies this condition is the game of bridge in which accurate probabilistic assessments are required for good playing. Two experiments were conducted in a natural setting of a tournament in which subjects were asked to assess the likelihood that a final contract (reached during the bidding phase) would indeed be made. Expert players were almost perfectly calibrated whereas amateurs were overconfident. The differences between expert and amateur players are discussed, and some guidelines for training procedures for calibration in general are proposed.