B - 100 Principal Component Analysis of WAIS-IV Indices with the Serial Digit Learning Test.

Abstract

Previous studies have evaluated the construct validity of the Serial Digit Learning Test (SDL) using principal component analysis (PCA; Larrabee et al., 1995; Greenman & Moses, 2016). In this study, we sought to re-investigate the SDL construct using a similar methodological approach with the Weschler Adult Intelligence Scale-4th ed. (WAIS-IV) and a performance validity measure. Sample consisted of diagnostically diverse veterans referred for clinical neuropsychological assessment as part of their standard clinical care (N = 411). SDL-8-digit raw scores were submitted to PCA where a forced three factor solution was derived. Those derived factors were submitted to a second PCA using the four WAIS-IV index scores. The SDL8 solution explained 83% of the variance, and factors reflected early, middle, and late trials. The second PCA produced an unforced four-factor solution explaining 71% of the variance. Excluding loading coefficients below +/-0.3, Component 1 showed loadings on WMI (0.805) and early trials (0.813), Component 2 showed loadings on PSI (0.800) and middle trials (0.789), Component 3 loaded on PRI (0.685) and late trials (0.829), and Component 4 loaded on VCI (0.951) and PSI (-0.348). These results replicated earlier research demonstrating that earlier SDL8 trials load on attentional/working memory measures and middle trials on processing speed. Unexpectedly, late trials loaded onto the Perceptual Reasoning Index, whereas they had previously loaded onto Verbal Comprehension of the WAIS-III. As these results may reflect the updated theoretical model underlying the WAIS-IV, further research replicating results which incorporated earlier WAIS-III perceptual reasoning (POI) measures is warranted.