Given the possible influence of demographic differences (i.e., sex, race, IQ, age) on the dependent variables, the inclusion of potential covariates was considered. Using bivariate correlations, we found that race was not significantly associated with any of the outcome variables, ps =.224, and therefore was not considered an appropriate model covariate. Sex was significantly related to outcome variables (p <.001). Follow-up bivariate correlations revealed that sex was not associated with any CANTAB variables (p = .330) but was significantly correlated with the WISC-IV and D-KEFS variables (p < .001) and therefore was considered a covariate in those analyses. Age was significantly associated with all outcome variables (p < .001). Follow-up bivariate correlations revealed that age was negatively associated with WISC (p =.032) and D-KEFS and positively associated with the CANTAB (p < .001) variables and was used as a covariate in those analyses. Although groups also differed on IQ, the extant literature illustrates the analytical, statistical and theoretical problems of covarying for IQ for neurodevelopmental disorders (Dennis et al., 2009). Furthermore, our interest in the components of IQ as variables in this study, supports not using it as a covariate.
