The Wechsler Memory Scale—Revised (WMS-R) malingering indices proposed by Mittenberg, Azrin, Millsaps, and Heilbronner [Psychol Assess 5 (1993) 34.] were partially cross-validated in a sample of 200 nonlitigants. Nine diagnostic categories were examined, including participants with traumatic brain injury (TBI), brain tumor, stroke/vascular, senile dementia of the Alzheimer’s type (SDAT), epilepsy, depression/anxiety, medical problems, and no diagnosis. Results showed that the discriminant function using WMS-R subtests misclassified only 6.5% of the sample as malingering, with significantly higher misclassification rates of SDAT and stroke/vascular groups. The General Memory Index—Attention/Concentration Index (GMI-ACI) difference score misclassified only 8.5% of the sample as malingering when a difference score of greater than 25 points was used as the cutoff criterion. No diagnostic group was significantly more likely to be misclassified. Results support the utility of the GMI-ACI difference score, as well as the WMS-R subtest discriminant function score, in detecting malingering.
Clinicians have long recognized the need to identify exaggerated cognitive deficits when evaluating neuropsychological functioning, especially when secondary gain is evident (Heaton, Smith, Lehman, & Vogt, 1978; Pankratz, 1983 and Rogers, 1997). Since examinees are unlikely to confess their deceit, researchers in this area have concentrated their efforts on increasing their abilities to detect simulated cognitive deficits. One primary method used to help achieve this goal is the identification of aberrant performances on existing neuropsychological measures (Bernard, 1991; Greiffenstein, Baker, & Gola, 1994; Millis, 1994). Research investigating differences in neuropsychological test performance patterns between simulators and truly impaired individuals is important because these tests are routinely administered as part of a typical neuropsychological evaluation; thus, administration is cost effective and time efficient. Further, these tests are not easily recognized as simulation measures.
The Wechsler Memory Scale—Revised (WMS-R; Wechsler, 1987) is the third most frequently administered neuropsychological test in the field of forensic clinical neuropsychology (Lees-Haley, Smith, Williams, & Dunn, 1996). As such, Mittenberg, Azrin, Millsaps, and Heilbronner (1993) proposed a method to detect simulated memory deficits using the WMS-R. These researchers hypothesized that individuals attempting to simulate symptoms of head trauma may perform better on the General Memory Index (GMI) than on the Attention/Concentration Index (ACI), a pattern that is typically not characteristic of individuals with postconcussive memory impairment or global amnesia (Butters et al., 1988 and Crossen & Wiens, 1988). They tested their hypothesis in a sample of 39 head trauma patients and 39 matched volunteers instructed to simulate symptoms of head trauma.
Mittenberg et al. (1993) first entered WMS-R subtest scores into a stepwise discriminant function analysis to generate decision rules for identifying malingered performances. They found that 91% of their sample was correctly classified, with a 7.7% false positive rate and a 10.3% false negative rate. Statistical cross-validation was achieved by deriving 78(n−1) discriminant functions with a different subject left out of each and classifying that subject with the function (i.e., jackknife procedure). Application of the discriminant function to two independent data sets published in the literature (i.e., a sample of head-injured patients and a sample of undergraduates, half of whom simulated memory impairment and half of whom performed their best) also resulted in acceptably accurate levels of classification.
Because the hypothesized pattern of GMI greater than ACI scores was observed in simulating participants, Mittenberg et al. (1993) utilized the difference score (GMI-ACI) as the independent variable in a discriminant function analysis, resulting in an overall correct classification rate of 83.3%, with a 10.3% false positive rate and a 23.1% false negative rate. Statistical cross-validation using the jackknife procedure achieved nearly identical results, as would be predicted.
Level of education and length of time since injury were not significantly related to either the GMI-ACI discrepancy score or the discriminant function score, while length of unconsciousness was significantly related to discriminant function scores but not GMI-ACI difference scores. The latter finding suggests more severely injured patients are less likely to be classified as malingering by the discriminant function score whereas the GMI-ACI discrepancy score is not influenced by head injury severity. Mittenberg et al. (1993) concluded that decision rules based on the GMI-ACI discrepancy and discriminant function scores may have clinical utility in identifying malingered head trauma symptoms. To help maximize diagnostic accuracy, they provided a probability table for accurate malingering classification listing several cutoff scores for both the GMI-ACI discrepancy and discriminant function scores. However, they cautioned that the provided cutoff scores may not be appr
