Incremental validity of the MMPI-2 content scales in an outpatient mental health setting.

Abstract

The incremental validity of the Minnesota Multiphasic Personality Inventory-2 Content Scales (MMPI-2; J. N. Butcher, J. R. Graham, C. L. Williams, & Y. S. Ben-Porath, 1990) was examined using clinical and content scale scores to predict conceptually relevant symptoms and personality characteristics of 274 male and 425 female mental health center outpatients. Regression analyses were performed to determine if the content scales contributed significantly beyond the conceptually relevant clinical scales in predicting therapists' ratings. Of the 10 content scales analyzed, incremental validity was demonstrated for 7 scales for men and 3 scales for women. A 2nd set of analyses indicated that incremental validity was demonstrated for 4 clinical scales for men and 6 clinical scales for women. The findings provide further evidence that the content scales aid interpretation of MMPI-2 scores by contributing additional information beyond the clinical scales. The clinical scales of the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway & McKinley, 1942) were constructed using an empirical-keying approach, resulting in heterogeneity of content for each of the scales. Because not all items of a given scale are related logically to the clinical syndrome that the scale was developed to assess, and the clinical scale score itself does not directly reflect which content areas are endorsed by a test-taker, heterogeneity serves to reduce the ease with which clinical scale scores are interpreted and their meanings are communicated (Butcher, Graham, Williams, & Ben-Porath, 1990). Consideration of item content became more important as researchers and clinicians recognized its usefulness in score interpretation (Graham, 1993). Several attempts have been made to devise an effective method of assessing the item content endorsed by test-takers. One of the earliest approaches involved critical items whose content was judged to be indicative of serious psychopathology (Caldwell, 1969; Grayson, 1951; Koss & Butcher, 1973; Lachar & Wrobel, 1979). Another approach was to create content-based subscales within some of the clinical scales (Harris & Lingoes, 1955/1968). Also attempting to create content scales, Tryon and several colleagues (Stein, 1968) used factor analysis to identify clusters of items from across the entire test. In the most thorough and comprehensive effort at creating content scales for the MMPI, Wiggins (1966) used a combined rational-statistical approach that resulted in 13 content scales formed from across the entire MMPI item pool. These scales were