College of Psychology: Faculty Articles


Cross-Validation of Predicted Wechsler Memory Scale-Revised Scores in a Normative Sample of 25- to 34-Year-Old Patients

Document Type


Publication Date


Publication Title

Archives of Clinical Neuropsychology








Equations for prorating Wechsler Memory Scale-Revised General Memory (GM) and Delayed Recall (DR) index scores (Woodard and Axelrod, 1995) were confirmed in the Mittenberg et al. (1992)normative WMS-R sample of 50 subjects between the ages of 25 and 34, and confirmed in a separate clinical sample of 30 patients with closed head injury who were age, education, and gender matched with 30 subjects from that normative sample. Predicted GM and DR index scores fell within 6 points of the obtained scores for 98% of the Mittenberg et al. (1992)sample and 93% to 100% of the matched head injury and normative samples, despite statistically significant differences between these groups in obtained GM, DR, and percent retention of LM I and II and VR I and II. Six points is well within the standard error of measurement of these index scores. These findings confirm the earlier cross-validation results reported by Axelrod et al. (1996)in a mixed sample of traumatic brain injury and other neurological insult, and suggest that this method of estimating weighted score sums for WMS-R General Memory and Delayed Recall indices may be safely used in normative samples of patients in this age range as well as neurologically compromised patients without significantly impacting index score accuracy. © 1997 National Academy of Neuropsychology. Published by Elsevier Science Ltd

The revision of the Wechsler Memory Scale (the Wechsler Memory Scale-Revised; Wechsler, 1987) included measures of nonverbal attention, learning, and short-term and delayed recall in an attempt to address criticisms that the original Wechsler Memory Scale (WMS) was primary sensitive to verbal memory only (Black, 1973; Larrabee et al., 1983; Prigatano, 1978; Skilbeck and Woods, 1980). Recent confirmatory factor analytic studies of the WMS-R, however, have demonstrated unitary composite indexes of only General Memory and Delayed Recall (Woodard, 1993). In addition, although previous investigations of patients with lateralized lesions have revealed differences in the expected direction on the WMS-R Verbal Memory Index (Chelune and Bornstein, 1988), subsequent investigations of modality-specific and material-specific memory changes (Chelune and Bornstein, 1988; Naugle et al., 1993) have primarily relied upon comparisons of the short term and delayed scores from the Logical Memory and Visual Reproduction Subtests of the WMS-R, using a methodology similar to that employed with the WMS by Russell (1975).

Woodard and Axelrod (1995)derived equations for prorating the General Memory (GM) and Delayed Recall (DR) Indexes of the WMS-R using raw scores from the Logical Memory I and II, Visual Reproduction I and II, and Verbal Paired Associates I and II subtests. This methodology would seem to reduce the administration time of the WMS-R without sacrificing the clinical efficiency of the test by retaining the composite indexes of GM and DR as well as the scores from Logical Memory I and II and Visual Reproduction I and II that have proved useful in investigating material-specific memory.

The Woodard and Axelrod (1995)regression equations were originally derived using a mixed sample of 308 psychiatric and nonpsychiatric patients referred for neuropsychological evaluation, and these equations were confirmed with the subtest means from the WMS-R standardization sample. The normative data for individuals ages 18 to 19, 25 to 34, and 45 to 54 provided in the Wechsler Memory Scale-Revised manual (Wechsler, 1978) and used for this confirmation, however, was derived by interpolation from adjacent-aged samples rather than empirically derived. Mittenberg et al. (1992)normative study of WMS-R performance of a stratified sample of 50 volunteer subjects 25 to 34 years old demonstrated substantial differences between the interpolated norms of the WMS-R standardization sample and the norms of their empirically derived sample, including score differences as great as 15 to 18 points on the Visual Memory and Delayed Recall indexes at the upper end of the distribution.

The Woodard and Axelrod (1995)regression equations have been cross-validated in a clinical sample of 258 patients (Axelrod et al., 1996), primarily traumatic brain injury patients of varying injury severity. To date, however, they have not been cross-validated in an empirically derived normative sample, limiting their applicability in clinical practice.

The present study investigates the efficacy of the Woodard and Axelrod (1995)regression equations in the Mittenberg et al. (1992)normative WMS-R sample of 50 subjects between the ages of 25 and 34. This methodology is also confirmed in a separate clinical sample of 30 patients with closed head injury similar to the Axelrod et al. (1996)cross-validation sample, who were age, education, and gender matched with 30 subjects from the Mittenberg et al. (1992)normative sample. Confirming the validity of this WMS-R memory index estimation procedure is especially crucial for head injury patients in the age range of 25 to 34. Traumatic brain injury accounts for greater rates of disability for individuals under the age of 35 than all other causes combined (National Institute of Neurological Disorders and Stroke, 1989), with peak rates of incidence occurring between the ages of 15 and 24.