| Abstract: | Family practice physicians read a case vignette describing a patient with a history of lung cancer, a new transient neurological disturbance, and a normal computerized tomographic (CT) scan of the head. They then estimated the probabilities of two diagnoses: transient ischemic attack (TIA) and brain tumor. Probability estimates of TIA were lower if the history of lung cancer was presented at the end of the case rather than at the beginning. This recency effect was found for both more and less experienced physicians and whether subjects were prompted for a single end-of-sequence probability judgment or multiple step-by-step judgments after each piece of information. These results are inconsistent with Hogarth and Einhorn's (1992) belief-adjustment model, which predicts a recency effect for the step-by-step condition but a primacy effect for the end-of-sequence condition. |
