The Use of Simulated Decision Makers in Information Evaluation.

Abstract This article deals with the expansion of the predictive ability criterion to consider the effects of information on decision-maker's (DM) behavior. The analysis involves two parts. In the first part, an information relevance criterion that simultaneously considers both the information and the DM's interpretation and use of it is developed. Under this relevance standard called prediction achievement, information alternatives are evaluated based upon the accuracy of the predictions made by the DM. Efficient implementation of this relevance measure requires that the effects of different information alternatives on the DM's behavior be predicted based upon descriptive mathematical models. The second part of this research consists of a field experiment designed to test three necessary conditions for the use of one particular model in the prediction of two-group classification decisions. E. Brunswik's Lens Model brings the relationship between predictive ability and information utilization into bold relief.