RESPONSE STYLES ON THE MMPI: COMPARISON OF CLINICAL AND NORMAL SAMPLES1

Abstract

ABSTRACTIn an effort to appraise the degree to which response consistencies on the MMPI are attributable in widely differing populations to the general stylistic dimensions of acquiescence and desirability on the one hand, and to particular content consistencies on the other, two studies on two diverse samples were performed. By applying factor analysis to correlations among MMPI clinical and validity scales scored separately for “true” and “false” items, along with five specially constructed reference scales, to measure tendencies to endorse items at varying levels of desirability, relatively precise estimates of the proportion of variance attributable to response style and to content were obtained. The two samples employed–hospitalized neuropsychiatric patients and college students–yielded results consistent with those obtained from a previous study of prison inmates, in that the dominant role of acquiescence and desirability response styles was. emphasized. In all three samples two very large factors appeared, identifiable as acquiescence and desirability. On the average, approximately three‐quarters of the common variance and over half of the total variance was attributable to the two stylistic dimensions, involving in the hospital sample a somewhat larger and in the college sample a somewhat smaller proportion of the variance as compared with the previous prison sample. All MMPI sub‐scales employed in these three analyses elicited substantial response style effects. Furthermore, in each of the three analyses the factor loadings of the scales on the acquiescence and desirability dimensions arrayed the scale vectors in a circular pattern, suggesting that acquiescence is differentially elicited by items at varying levels of desirability.In addition, several quite small factors were obtained in each of the analyses, some attributed to item overlap and some to consistent content responses. Four factors in the hospital sample and three in the college sample were identified tentatively in terms of item content. Three factors more or less common to the two groups were interpreted respectively as: (1) endorsement of items reflecting anxiety symptoms; (2) naive test‐taking defensiveness; and (3) social brashness and impulsivity. One factor, appearing only in the hospital sample, was identified as reflecting somatic complaints.The massive response style effects on the MMPI, while contributing to convergent validity in grossly separating normals from mental patients, are interpreted as reducing the possibilities for discriminant validity, as in the differential diagnosis of psychopathological syndromes. Where the latter is important, experimental controls aimed at sharply reducing response biases are recommended.