Effect of Input From a Mechanical Model on Clinical Judgment


Numerous studies have demonstrated that mechanical models outperform decision makers. Despite this evidence, mechanical models have rarely been used in practice. Because people appear to be unwilling to base important decisions entirely on the output of a mechanical model, the present study investigated the effects of presenting decision makers with estimates based on mechanical models of their own prior decisions, before they were required to make a prediction. In the task, 20 MBA students attempted to predict the number of games a National League baseball team won during the year based on the team's earned-run average, the team's batting average, and the average number of home runs hit per game during the year. The results indicated that input from a mechanical model improved the decision makers' predictions, although their predictions were not as accurate as the model. The decision makers' estimates were more extreme than their models' estimates. In addition, their estimates deviated most from the model when the three team statistics provided conflicting information. Outcome feedback increased the amount by which the decision makers' estimates deviated from the model. © 1986 American Psychological Association.

Document Type




Publication Date


Journal Title

Journal of Applied Psychology