another, which makes straightforward Bonferroni correction inappropriate. The genetic analysis program Plink (Purcell et al., 2007) has implemented a permutation approach that gives a set-based empirical p value indicating the likelihood of observing the pattern of p values across the SNPs, taking into account their correlation structure. This program yielded an empirical p value of .039 for the observed interaction effect, suggesting that it was significant after correction.
