Hypothesis generation and confidence in judgment.

Ss asked to generate their own hypotheses expressed less confidence that they were true than did other Ss who were presented with the same hypotheses for evaluation. This finding holds across domains varying from prediction and social inference to general knowledge questions. Furthermore, Ss who generated their own hypotheses appeared to be more sensitive to their accuracy than were Ss who evaluated the hypotheses. The results are interpreted as evidence that the hypothesis generation task leads Ss to consider more alternative hypotheses than Ss who are asked to evaluate a prespecified hypothesis. This interpretation is supported by experiments demonstrating that the difference between generation and evaluation disappears if a closed set of alternatives is specified or if a delay is inserted between hypothesis generation and confidence assessment