Detecting simulated memory impairment with the Rey Auditory Verbal Learning Test: implications of base rates and study generalizability.

The accuracy of the Rey Auditory Verbal Learning Test (RAVLT) for detecting simulated neuropsychological impairment was investigated using discriminant function analyses and cross-validation procedures. In Study 1, a discriminant function analysis, using six RAVLT variables, correctly classified 80% of 50 simulating and 50 non-simulating participants. Base rates were changed and this discriminant function was tested in Study 2 with a 33% base rate of simulation. The discriminant function correctly classified 75% of 20 simulators and 90% of 40 non-simulators, yielding an overall classification rate of 85%. To determine the generalizability of this discriminant function to a clinical sample, Study 3 compared the RAVLT performance of 20 simulating participants to that of 40 motor vehicle accident (MVA) patients. The discriminant function from Study 1 incorrectly classified 60% of the MVA patients and yielded an overall correct classification rate of 48%. The limitations associated with generalizing from analog research and the implications of varying base rates on classification accuracy are discussed.