Construct Validity of the German Wechsler Intelligence Scale for Children–Fifth Edition: Exploratory and Confirmatory Factor Analyses of the 15 Primary and Secondary Subtests

Abstract

<jats:p> The latent factor structure of the German Wechsler Intelligence Scale for Children–Fifth edition (German WISC-V) was examined using complementary hierarchical exploratory factor analyses (EFAs) with Schmid and Leiman transformation and confirmatory factor analyses (CFAs) for all reported models from the German WISC-V Technical Manual and rival bifactor models using the standardization sample ( N = 1,087) correlation matrix of the 15 primary and secondary subtests. EFA results did not support a fifth factor (Fluid Reasoning). A four-factor model with the dominant general intelligence ( g) factor resembling the WISC-IV was supported by EFA. CFA results indicated the best representation was a bifactor model with four group factors, complementing EFA results. Present EFA and CFA results replicated other independent assessments of standardization and clinical samples of the United States and international versions of the WISC-V and indicated primary, if not exclusive, interpretation of the Full Scale IQ as an estimate of g. </jats:p>

Revisiting the Factor Structure of the German WISC-V for Clinical Interpretability: An Exploratory and Confirmatory Approach on the 10 Primary Subtests

“…Except for PS, however, reliability and replicability coefficients for the group or first-order factors appeared to be too low to suggest that the according unique cognitive dimensions are sufficiently well represented by the WISC-V primary indexes. Similar results have not only been found for the German WISC-V as described earlier (Canivez et al, 2021) but had already been observed in studies focusing on international versions of the WISC-IV (Watkins, 2010;Canivez, 2014; and on other Wechsler scales as well (Canivez and Watkins, 2010;Golay and Lecerf, 2011;Watkins and Beaujean, 2014). Even though results of the present study indicated meaningful interpretation of an overall measure for g, this is only true if the corresponding composite score is based on all 10 primary subtests.…”

“…Their aim was to compare their EFA and CFA results with the model solution provided in the test manual and those models proposed for other versions of the WISC-V. Additionally, the authors compared bifactor model and higher-order model solutions as rival explanations and provided detailed information about all sources of decomposed factor variance and the according model-based reliability coefficients. In line with the aforementioned studies on standardization and clinical samples of international WISC-V versions, findings of Canivez et al (2021) again supported a four-factor model rather than the five-factor model solution that was proposed by the test publishers. Furthermore, their CFA results additionally suggested a bifactor model with four group factors to be the best structural representation of the German WISC-V.…”

“…These changes included the translation and adaptation of the entire set of verbal items, the exclusion of some complementary scores, and several modifications to verbal contents (see Wechsler, 2017a , for a detailed description). In a recently published study, Canivez et al (2021) were the first to analyze the structural validity of the 15 WISC-V primary and secondary subtests based on the correlation matrix of the German standardization sample ( N =1,087). Their aim was to compare their EFA and CFA results with the model solution provided in the test manual and those models proposed for other versions of the WISC-V. Additionally, the authors compared bifactor model and higher-order model solutions as rival explanations and provided detailed information about all sources of decomposed factor variance and the according model-based reliability coefficients.…”

Measurement Invariance of the Intelligence and Development Scales–2 Across Language Versions, Gender, and Age

“…Model-based reliability analyses with McDonald's omegas revealed a strong general intelligence factor and weak broad ability factors for the IDS-2 intelligence and basic skills domains. This is in line with previous findings from Grieder and Grob (2020) and with findings on other major intelligence tests (e.g., Canivez et al, 2017Canivez et al, , 2021Cucina & Howardson, 2017). Although there was some variation in omegas across the Dutch and German IDS-2 versions, sex, and age, the main conclusions are the same: The true score variance due to general intelligence seems sufficient for interpretation of a full-scale IQ, while the little true score variance due to the broad abilities casts doubt on the justification of broad ability composites, with the possible exception of ASM, whose ωs sometimes exceeded the threshold of .50.…”

“…Our results provide further evidence against the separability of visual processing and fluid reasoning abilities, as the theoretical IDS-2 factors VP and AR-corresponding to the CHC broad abilities Visual Processing (Gv) and Gf-collapsed to one factor, AVR, in the present study. This is in line with results from Grieder and Grob (2020) and from previous studies on other intelligence tests (e.g., Canivez et al, 2017Canivez et al, , 2021Keith & Reynolds, 2010), and is in contrast to the CHC model. The collapse of these two factors may be due to the content of Gf subtests often being exclusively visual (e.g., matrices).…”