Building a Structural Model of Computer-Aided Decision Making

It is becoming clear that computerized automation in the form of expert systems can be a useful aid for a wide variety of decision tasks including those involving diagnosis. To optimally combine human talent and expert systems, we need knowledge concerning what variables affect human performance, what variables affect machine performance, and what variables affect the operator's use and acceptance of the judgments offered by the expert system. This paper presents preliminary work on a causal model that has been developed to describe the decision system for an expert-aided diagnostic inference task. It is suggested that covariance structure analysis techniques can provide a useful tool for assessing the validity of this type of complex system model. To illustrate the approach, laboratory data relevant to the model are presented including analysis via structural equation modeling.