Development and cross-validation of a specail scale of the MMPI-168 to predict psychiatric recommendations of incompetency to stand trial

Although the MMPI is commonly used in the assessment of individuals undergoing evaluations for competency to stand trial, there have been few studies in the literature that have examined the relationship of this instrument to pretrial competency. Moreover, these studies have focused upon differences among the basic scales and have produced inconsistent results. The present study investigates the usefulness of the MMPI (and MMPI-168) in differentiating competent and incompetent defendants through both a profile and item analyses of these inventories. A sample of 522 pretrial evaluations performed at the Middle Tennessee Mental Health Institute in Nashville, Tennessee was randomly divided into scale-construction and cross-validation subsamples. A discriminant function analysis of the basic MMPI scales from the scale-construction subsample found only one scale (7) which accounted for all significant variance. Although it significantly differentiated between these two groups in this subsample, it classified Ss significantly poorer than base rate prediction. Moreover, this scale failed to even reach significance between the competent and incompetent groups in either the split-half cross-validation subsample or a second cross-validation subsample of 104 pretrial defendants from Lakeshore Mental Health Institute in Knoxville, Tennessee. An item analysis was performed on the first 168 items of each MMPI of the scale-construction subsample using a 2 X 2 chi-square design; 35 items were derived which significantly differentiated competent and incompetent offenders. A discriminant function analysis was performed on these items to eliminate redundancy and to obtain the best scale. Five items accounted for all significant variance and comprised the final scale. Rescoring the scale-construction subsample with this scale showed that it significantly discriminated between the two groups; comparable results were obtained with both cross-validation subsamples. However, correct classification of Ss using this scale was not significantly better than simple base rate prediction. Excluding Ss whose I.Q.s were below 70 and whose MMPIs were of questionable validity did not significantly enhance the scale's classification rate. It was concluded that the MMPI (and MMPI-168) are not particularly useful instruments in predicting psychiatric recommendations of pretrial competency.