Man versus mean: the exploitation of group profiles for the construction of diagnostic classification systems.

To investigate the comparative advantages of data from group as compared to individual profiles, 233 group MMPI profiles were used to construct both sequential and simultaneous classification procedures for psychiatric diagnosis. Virtually perfect separation of the group profiles was accomplished by both procedures, the results ranging from 93% correct (psychotic versus neurotic) to 99% correct (psychiatric versus sociopathic). Moreover, a predictor index initially developed on individual profiles produced a cross-validity coefficient of .83 when applied to the group profiles. These findings suggest that group data appear to contain such a high signal-noise ratio that they become extraordinarily efficient indicators of underlying processes—processes which are normally obscured by the unreliability inherent in individual profiles. Moreover, since these processes appear to be unusually well captured by a general linear model, the rather jaundiced views towards group profiles held by most of the clinical and psychometric community may well be unfounded.