ABSTRACT
Objective: Assessment of intellectual abilities in individuals with autism spectrum disorder (ASD) is a core component of a comprehensive diagnostic evaluation. However, relatively limited information is available regarding the validity of one of the most commonly-used measures of intelligence, the Wechsler Intelligence Scale for Children – 5th Edition (WISC-V) in ASD.
Method: We investigated the factor structure and measurement invariance of the WISC-V in a sample of 349 children aged 6–16 diagnosed with ASD using single- and multi-group confirmatory factor analysis. The comparison group was the WISC-V standardization sample.
Results: A four-index bifactor solution best fit the ASD group data. Measurement invariance analyses indicated support for configural and metric, but not scalar, invariance of the published 5-index structure, suggesting systematic differences in performance among some subscales in ASD. The 7-subtest FSIQ scale had partial scalar invariance after relaxing equality constraints on the Coding and Digit Span subtest intercepts, suggesting sources other than theorized IQ ability contribute to lower scores on these subtests within ASD. The Cognitive Proficiency Index (CPI) failed to demonstrate appropriate fit in baseline models. The General Ability Index (GAI) had full configural, metric, and scalar invariance.
Conclusions: Statistical bias on the WISC-V within ASD in processing speed and working memory subtests creates significant limitations for the use of FSIQ and especially CPI index scores in ASD populations. The GAI showed strong measurement properties and should be considered as the preferred indicator of overall intellectual functioning when assessing children with ASD using the WISC-V.
Acknowledgments
Standardization data from the WISC-V. Copyright © 2018 NCS Pearson, Inc. Data used with permission. All rights reserved. NCS Pearson, Inc.
We would like to thank the families who participated in the assessments used in this study as well as the psychologists and psychometricians for completing the assessments. Special thanks to Aaron Kaat, PhD for his help interpreting the study results and to Teresa Ibañez and Erin Given for their assistance in collecting the participant data.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1 SRMR has different cutoff values due to being more sensitive to noninvariance for metric vs. scalar invariance (F. F. Chen, Citation2007).