Abstract
This study explored the manner in which the desirability of an event influences its judged probability. Ss gave probability estimates for each of 5 events, only one of which could occur. Monetary payoffs, ranging from lose $5 to win $5, were contingent upon which event did occur. Desirability was found to bias probability estimates in a complex manner which varied systematically between Ss and between estimation trials. In general, it made estimates less reasonable. Rewards for accuracy did not reduce value biases. Instead, they encouraged ``risk-reducing pessimism.'' Individual differences were an important source of variance. Some Ss were consistently optimistic. Others were quite pessimistic.