MMPI-2 Content Scales: How Much Content Do They Measure?

Abstract

The new Minnesota Multiphasic Personality Inventory-2 (MMPI-2) content scales, designed to assess a wide range of substantive dimensions of psychopathology, are evaluated in terms of their confounding with general factors associated with response styles, such as desirability responding. We developed a set of 21 scales that ranged from high to low desirability by blocking sets of items ranked in terms of desirability and arbitrarily keyed in terms of ‘true’ and ‘false.’ The MMPI-2 was administered to 221 university undergraduates and scored for the 21 desirability scales and the 15 content scales. Factor analysis with Varimax rotation of the desirability scales showed a clear circumplex structure, with one dimension reflecting desirability and the other separating true- and false-keyed scales. Dwyer extension methods were used to project this solution into the space of the 15 content dimensions. Correlations between the projected stylistic structure and that obtained from the content scales directly were .98 and .98 for the two largest dimensions. Further analyses revealed that coefficient alphas of MMPI-2 content scales were reduced sharply when variance attributable to response styles was removed. It was concluded that convergent and discriminant validity of MMPI-2 content scales are seriously compromised by the presence of substantial, confounding, general variance.