Date of Award
Doctor of Philosophy
Reading and intelligence tests are commonly used as instruments to decide on special educational services for students. The construct validity and predictive validity of the Woodcock Johnson IV Tests of Cognitive Abilities (WJ IV; Schrank, McGrew, & Mather,2014) has not been examined for bias across racial/ethnic groups. This study extends Keith's (199) examination of bias for general and specific cognitive ability latent variables predicting basic reading skills, reading comprehension, and reading rate using a cognitive model and a reading achievment model adapted from Niileksela, Reynolds, Keith, and McGrew (2016). Using Black, Hispanic, and White samples of 9- to 13-year-olds from the WJ IV norming sample, the cognitive model demonstrated minimal evidence of variance across groups at the metric level, and the reading achievment model was invariant across groups. Multisample structural equation modeling resulted in no significant differences across racial groups. Gcand Gs were consistent predictors for all reading factors. All predictive models for this study demonstrated a poor fit. Overall, no predictive bias was found in the distribution of underlying latent variables that produce reading skills and inherent differences across racial/ethnic groups are not supported.
dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.
Woods, Isaac L. Jr., "Do the WJ-IV Tests of Cognitive Abilities Predict Reading Equally across Groups?" (2017). Electronic Theses and Dissertations. 1692.