A configural analysis of psychiatric diagnostic stereotypes

Abstract

This article is the second report of a series of investigations concerned with psychiatric diagnostic classification, designed to increase the understanding of psychiatric diagnosis and eventually to contribute toward an increase in the effectiveness of diagnostic procedures through the application of objective quantitative methods. The first article (Overall & Gcrham, 1963) reported the use of a Bayesian probability model and computer techniques to classify rating profiles of typical or stereotype cases. It was demonstrated that the computer could identify with high degree of accuracy the particular diagnostic types being rated. The fact that some errors in classification did occur and that they were more likely among certain categories than among others suggested the need for a further investigation of relationships and differences among symptom profiles associated with different diagnostic concepts. The present article reports application of the method of multiple discriminant analysis to the study of configural relationships among symptom profiles associated with thirteen psychotic types.