Does high scatter on the Wechsler Adult Intelligence Scale–Fourth Edition general ability index reduce validity in predicting Wechsler Memory Scale–Fourth Edition indexes?

Abstract

We examined whether significant scatter in WAIS–IV GAI will reduce its validity to predict performance on WMS–IV indexes. Participants were 330 individuals with neurological, psychiatric, or neurodevelopmental disorders and 59 referrals who were found to be free of a diagnosable disorder. For VCI > PRI, 59.32% were significant at p < .05 and 12.29% were >22 points. For VCI < PRI, 48.37% were significant at p < .05 and 7.19% were >22 points. Inter-subtest scatter across GAI subtests indicated 82.26% of individuals had a significant scatter range and 13.88% had an unusually large range (≥8). For the VCI, 49.10% had significant scatter (≥3) and 12.08% had an unusually large scatter range (≥5). On the PRI, 43.19% had a significant scatter range (≥4) and 12.85% had an unusually large degree of scatter (≥6). Moderation analyses revealed GAI was a significant predictor of each WMS–IV index. The interaction term of GAI with GAI scatter was not significant for any indexes, indicating that regression equations for predicting WMS–IV scores from GAI did not vary significantly across levels of scatter. Estimation of WMS–IV indexes from the GAI is justified even when significant VCI–PRI discrepancies are present and there is unusual variability across the GAI subtests.