Cross-Validation of a Psychological Test Battery to Detect Faked Insanity

Attempts were made in these 2 studies of simulated insanity to enhance ecological validity: Experimental fakers were given potent monetary incentives to conceal their deception; psychiatric inpatients and convicted felons were used in criterion groups; and subjects took a battery of psychological tests that varied in task demands. In Experiment 1, 8 predictor variables drawn from the MMPI, Bender Gestalt, and a newly developed Malingering Scale were used to derive a single discriminant function that best differentiated 20 prison inmates faking insanity from 40 nonfaking criterion and control group subjects. The optimal function consisted of 3 variables, yielded a canonical correlation of .823, and correctly classified 93.3% of subjects. Experiment 2 served to cross-validate the derived equation on 22 substance abusers instructed to fake insanity and 20 inpatients from a schizophrenia treatment unit who received standard instructions. On cross-validation, 97.6% of subjects were correctly classified. Reasons for the unexpected lack of predictive "shrinkage" are discussed.