ON DETECTING "FALSE NEGATIVE" MMPI PROFILES: A QUESTION OF VALIDITY
Abstract
A serious weakness of any structured personality questionnaire which, a priori, assumes subject frankness and accuracy, is its susceptibility to the effects of malingering and self-deception. That this is an important failing, particularly with respect to validity, has been amply demonstrated in the literature. The development of the Minnesota Multiphasic Personality Inventory (MMPI) in the early 1940's represented a major improvement in objective self-report personality testing. Its four "validity" scales, used singly or in combination, have been found of some value in detecting the "faked bad" or malingered profile. Experimental attempts to locate "faked good" profiles, however, typically have met with unsatisfactory results. The present study, following the works of Wales & Seeman (1968), investigates the behavior of the "subtle"/"zero" (Wiener & Harmon, 1946) items of the MMPI K scale (K("0")) and the prospects for establishing a "fake good" index by examining the relationship of these items and those of the MMPI F or "frequency" scale (F("X")). The subjects were 43 female third-year occupational therapy students. Their mean age in years was 21.2, the range 20 to 29. Each subject volunteered to take the MMPI under "honest" and "fake good" instructions. An analysis of the experimental effects of the independent variable revealed a significant difference for MMPI scales L, K, K("0"), Mf, and Si. These differences also were found to be in the predicted direction. A correlational analysis (phi) of the F("X") minus K("0") index raw scores indicated that "cutting scores" of -3 and -5 would produce the phi coefficient having the greatest value (r(,(phi)) = .396). This compared to a phi value of r(,(phi)) = .256 for the K scale raw scores and a value of r(,(phi)) = .588 for the K("0") items. The results of the study indicate that the proposed index is of questionable practical utility but suggest that the "zero" items may play an important role in future investigations of test dissimulation. Of greater significance for these experimental inquiries, however, may be the influence of the demand characteristics of the setting on the subjects' MMPI scores (viz., the type of instructions given; experimental versus practical). A discussion of these protean influences is presented as well as appropriate cautions with respect to experimental design.
This paper has been withdrawn.