Despite having clinically relevant and significant findings, our results should be considered in the context of important limitations. We did not examine the relationship between neuropsychological performance and IQ scores. Unfortunately, the options for dealing with expected IQ differences (e.g., matching groups or covarying IQ in analyses), result in additional analytical and theoretical problems, such as producing meaningless results if the covariate is intrinsic to group membership, creating statistically overcorrected results, and decreasing the generalizability of results (Dennis, et al., 2009).
