Realism in eyewitness confidence as a function of type of event witnessed and repeated recall.

This study investigated how the type of event witnessed and a repeated test schedule for confidence influenced the realism in confidence judgments. The experimental design contrasted 2 film versions (a violent and a nonviolent scenario) and 3 tests of confidence (immediate, repeated, and delayed). On average, for all single items, participants were highly overconfident in their judgments. However, the same participants severely underestimated their own performance when they, at the end of the test session, were asked to provide an estimate of how many questions they thought they had answered correctly. Whereas the effects on realism in confidence for the 2 different film versions were small, the realism in witnesses' confidence judgments increased when participants repeated their confidence ratings. The theoretical and forensical implications of these findings are discussed.