Is a Confirmatory Tendency to Blame for Poor Diagnostic Decisions?

Abstract

In diagnostic reasoning, especially in a clinical setting, practitioners often widely disagree about the causal explanation for a given case. Most studies have assumed that such disagreements result from judgmental mistakes due to biased reasoning, especially the tendency to seek confirmation for a theory or hypothesis they already entertain. Alternative explanations for these mistakes are: confusion about the type of requested diagnosis and a difference in knowledge available to the practitioner. The present paper introduces a method to control the latter two factors, for the first time enabling a study of the influence of the first factor in isolation. In the experiment, eighteen advanced postgraduate practitioners made a diagnosis in interaction with a computer program simulating various reading impairments. We found two surprising results: First, the confirmatory tendency was not as prevalent as commonly assumed and second, more important, that where it was employed it was conducive to sound diagnostic decisions.