The Relationship of MMPI-A Item Effectiveness to Item Content, Diagnostic Category, and Classification Accuracy

Archer, Handel, and Lynch (2001) recently compared the item endorsement frequencies for the MMPI-A normative sample against two adolescent clinical samples. Results showed that the MMPI-A contains a substantial number of items that do not show a significant difference in item endorsement frequency between normative and clinical samples. The current study extends Archer et al.\u27s (2001) research in three ways: (1) it examines the item endorsement frequencies of the Supplementary scales, Harris Lingoes subscales, and subtle-obvious items; (2) it examines the Basic, Content, and Supplementary scales, and Harris Lingoes subscales with two homogeneous diagnostic criterion groups (as suggested by Archer, Handel, and Lynch); and finally (3) it reexamines and recalculates Basic scale data using only those items that were shown by Archer, Handel, and Lynch (2001) to effectively discriminate between the normative and clinical populations. The mean profiles of the normative and clinical groups were contrasted based on these “revised” Basic scales using a newly acquired independent clinical sample to evaluate the extent to which profile sensitivity and specificity is affected by these scale modifications. Results demonstrated that examining the Supplementary scales and Harris-Lingoes subscales, or subtle-obvious items, or when extended to homogeneous criterion groups led to no improvement in item effectiveness from the results of Archer and his colleagues\u27 study. However, results supported the hypothesis in that the overall effectiveness of Basic scale discrimination increased, in terms of sensitivity, specificity, positive and negative predictive power, and hit rate, when the Basic clinical scale items were removed that did not discriminate between normative and clinical groups of adolescents