The detection of fake-bad and fake-good responding on the Millon Clinical Multiaxial Inventory III

The purpose of this study was to examine the effectiveness of the three Modifying Indices of the MCMI-III in the detection of fake-bad and fake-good responding. The entire sample consisted of 160 outpatiant psychiatric patients, 80 with a primary diagnosis of a Mood Disorder and 80 with a primary diagnosis of Schizophrenia. 2 x 2 x 2 mixed model MANOVA's with discriminant function analyses were performed to examine the effects of Instructional Set (faking versus standard instructions), Diagnosis (mood disorders versus thought disorders), and Order of Instructions. Two identical analyses were utilized, one for faking bad and one for faking good. In differntiating faking from honest responding, there were no significant effects involving the order of instructions or diagnostic group. As hypothesized, Instructional Set did produce significant differences on Scale X (Disclosure Index), Scale Y (Desirability Index), and Scale Z (Debasement Index) in both fake-bad and fake-good analyses. To detect faking bad and faking good, cutting scores for Scales X (DIS), Y (DES), and Z (DEB) as single scales and in combination were established through various methods. Corresponding operating characteristics (sensitivity, specificity, positive predictive power, negative predictive power, overall diagnostic power) were then determined. Single scale cutoffs proved just as effective as the multiple scale cutoffs. The highest overall rate of successful classification in the fake-bad analysis was 74%, using an optimal Scale X (DIS) cutoff of BR...