IQ scores are often used to predict school performance. However, for children with learning disabilities, the predictive validity of IQ scores appears to be low. In the fourth and fifth versions of the Wechsler Intelligence Scales for Children (WISC), new ancillary indices were introduced. The General Ability Index (GAI) is a broad measure of fluid reasoning with verbal, visual-spatial, and figural test items. By contrast, the Cognitive Proficiency Index (CPI) combines different executive functions known to be frequently affected in children with dyslexia, ADHD, or combined learning disorders. To date, there is little evidence to demonstrate that these measures improve the prediction of school performance beyond the Full-scale IQ (FSIQ). We therefore used lasso regression to explore the predictive validity of these measures for school grades. The analyzed samples were taken from the German standardization samples of the WISC-IV and the WISC-V. In most cases, the prediction of school performance was not considerably improved by taking the GAI or the CPI into account. However, when the individual discrepancy between the CPI and the GAI was high, the FSIQ lost its predictive validity in elementary school. In this subgroup of children, reading and writing skills were best predicted by the CPI, whereas math skills were limited by the lowest score (i.e., the minimum out of the CPI and the GAI).
Keywords: CPI; GAI; IQ; learning disorders; linear regression; school grades; school performance.