All analyses were conducted using the Statistical Analysis System (SAS). Association analyses were run using an additive model with sex, age, and race incorporated as covariates. In addition, the correlated nature of some observations (e.g., children from the same family) was taken into account using the survey option in SAS. We report p-values uncorrected for multiple testing, since the analyses are intended to be an exploratory examination of patterns of association across the impulsivity variables. The GABRA2 SNPs are in fairly high linkage disequilibrium (LD), and applying a Nyholt correction (Nyholt, 2004) to the data suggests just under two independent signals, which suggests a significance threshold of 0.026. We conducted tests on nine primary phenotypes, and therefore the final Bonferroni-corrected p-value for any single association would be .003, although this is conservative, as it does not take into account the correlated nature of the phenotypes. We regard it as suggestive evidence for association when multiple SNPs yield p < .05, and there is a generally consistent pattern across the SNPs. We present all results in order that the reader can evaluate the consistency of evidence across SNPs, phenotypes, and samples.
