Abstract
In 1956 Meehl predicted that the relationships between MMPI scores and the psychosis-neurosis diagnostic classification should be highly configural in character, and therefore that no linear combination of MMPI scores should be able to differentiate neurotic from psychotic patients as accurately as either experienced clinical psychologists or configural actuarial techniques. The present paper summarizes the findings from ten years of research on this question. While the search for configural actuarial procedures has led to a moderator variable, neither clinical experts, moderated regression analyses, profile typologies, the Perceptron algorithm, density estimation procedures, Bayesian techniques, nor sequential analyses-when cross-validated-have been-able to improve on a simple linear function. The implications of these negative findings for investigations of configural relationships with other problems are discussed.

This publication has 20 references indexed in Scilit: