NORMATIVE AND SUBSTANTIVE EXPERTISE IN MULTIPLE HYPOTHESES EVALUATION

We distinguish two criteria for evaluating the judgments of trained pr ofessionals. One criterion is conformance with a theoretical model and the other is conformance with known external criteria. Further, we la bel judgments that depart from a theoretical model as errors and those that depart from known external criteria as mistakes, Following this distinction, we hypothesize that auditors' multiple hypotheses judgmen ts will be characterized by errors but not mistakes. This hypothesis w as tested by asking professional auditors to evaluate multiple hypothe ses. The results confirm our expectations. Auditors' judgments reflect ed ecological base rate information, and they appropriately ignored no ndiagnostic evidence. Moreover, auditors did not exhibit a perseveranc e bias and 84% of them identified the correct hypothesis. The absence of mistakes reflect substantive expertise, Conversely, auditors' proba bilities were not additive, and, when a hypothesis was eliminated, the y did not adjust beliefs for the remaining hypotheses. These errors re flect a lack of normative expertise. A second study employed inexperie nced subjects to establish whether the strong substantive performance of the professional auditors was attributed to their expertise or was an artifact of the task. The results from the second study were strong ly supportive of the substantive expertise explanation. Taken together , these results suggest that substantive expertise can help contain mi stakes but it is not sufficient to mitigate errors, Further, lack of n ormative expertise can lead to errors but these errors do not translat e into mistakes, The paper concludes that the distinction between norm ative and substantive expertise on one hand and errors and mistakes on the other is crucial to understanding when basic findings will genera lize to professional settings. (C) 1995 Academic Press, Inc.