Executive functions (EFs) and intelligence are considered similar constructs. There is also a high degree of overlap when using psychometric assessment tools to measure the two constructs. The aim was to investigate the commonalities between EFs and intelligence by exploring the latent structure of the Wechsler Adult Intelligence Scale (WAIS-IV) and the Executive Functions Module from the Neuropsychological Assessment Battery (NAB). The study was also aimed at examining the correspondence of the two measures to the Cattel-Horn-Carroll (CHC) theory.
126 healthy participants, aged 18-88 years, were administered the NAB and WAIS-IV. Factor analytic approach combining principal component analysis (PCA) with the Schmid-Leiman (SL) procedure and confirmatory factor analysis (CFA) was employed.
The PCA with a promax rotation has resulted four factors, which accounted cumulatively for 63% of the variance. Within the SL procedure, first the maximum of variance was explained by a second-order factor (60%), and subsequently residual loadings on first-order factors were calculated. All subtests loaded most substantially on the second-order factor representing General Intelligence (g). When considering first-order factors, the correspondence to the CHC-broad abilities was in focus. Several subtests (WAIS-IV Similarities, Vocabulary, and Information; NAB Categories, Letter Fluency, and Word Generation) loaded on a factor representing Comprehension Knowledge (Gc). The WAIS-IV Block Design and Visual Puzzles subtests mostly loaded on a factor representing Visual Processing (Gs). The WAIS-IV Symbol Search and Coding subtests, and the NAB Mazes subtest mostly loaded on a factor representing Processing Speed (Gs). The WAIS-IV Digit Span, Matrix Reasoning, and Arithmetic subtests, and the NAB Judgment subtest mostly loaded on a factor representing two CHC broad abilities – Fluid Reasoning (Gf) and Short-Term Memory (Gsm). A higher-order model was specified within the CFA to examine the latent factor structure suggested by the PCA-SL solution. The broad ability factors exhibited substantial loadings on g; the highest g loadings were observed in the Gf + Gsm (β = .87) and Gc (β = .83) factors, followed by the Gv (β = .68) and Gs (β = .73) factors. The subtests loadings on the broad ability factors were all significant and slightly higher for the WAIS-IV than for the NAB. The PCA-SL model failed the χ2 test of perfect fit (χ2 = 125.36, df = 86, p = .004). However, a comparison with rival models reveled that this model best fitted the data. Only one nested model with one additional cross-loading (β = .31) from the Coding subtest on the Gf + Gsm factor yielded a statistically significant drop in χ2 (χ2 = 116.41, df = 85, p = .013). The worst model fit was represented by a one-factor model (χ2 = 224.17, df = 90, p = .000) and a model separating the WAIS-IV and NAB subtests (χ2 = 157.35, df = 85, p = .000).
The results demonstrate that EFs and intelligence have substantial overlaps; particularly regarding the CHC general ability. An EF factor being independent from the CHC abilities could not be confirmed. Instead, one factor combining Gf and Gsm and separate factors for Gc, Gv, and Gs grouped all subtests involved. The current findings suggest the confluence of EFs and intelligence within the CHC taxonomy.
