Inferences based on nondiagnostic information

Abstract

Nondiagnostic samples of evidence were presented sequentially in an inference task. These were samples which logically should not have influenced inference responses, e.g., by Bayes theorem. Experiment I, however, demonstrated that three different types of nondiagnostic samples (neutral, irrelevant, and null) consistently resulted in less extreme inference judgments. Experiment II indicated that the size of this “nondiagnostic effect” depended on serial location and was less with aggregate than sequential samples. Moreover, the effect was shown not to be a response artifact due to a central regression tendency. Several theoretical accounts were considered, such as representativeness and expectancy. But, the most plausible was that inference judgments are based on an averaging combination rule. Finally, some practical implications of this nondiagnosticity finding were discussed.