Test difficulty and judgment bias.

Two experiments tested the hypothesis that overconfidence in performance judgments is due to test- and person-driven errors. In Experiment 1, test difficulty accounted for the vast majority of variation in overconfidence when individuals judged items of varying difficulty within a homogeneous test. In Experiment 2, the severity of overconfidence did not differ between three unrelated tests once test difficulty was controlled. Both experiments supported the view that overconfidence is due largely to test difficulty. Some degree of overconfidence also occurred because individuals adopted a normatively high success criterion for judging their own test performance.