Man or Machine as Diagnostic Tool: a Comparison between Clinical Psychologists and Discriminant Function Analysis

Abstract

Power et al. (1974) gave the Maudsley Personality Inventory plus a malingerer scale to three groups—neurotics, controls and students asked to malinger—and found, using discriminant function analysis, that 89 per cent were correctly classified on the basis of their test scores. When a second sample of the three types of subject was allocated using the weights established by the first analysis 78 per cent were correctly classified. Experienced clinical psychologists were given the norms established with the original three groups, and the scores of 20 of the individuals in the second set of groups, and were asked to state group membership of these individuals. The psychologists performed, on average, as well as did the discriminant analysis. This was better than expected, since several variables had to be considered at the one time. Some interesting discrepancies are discussed, as is the extent to which the results can be generalized.