Linear regression was used to analyze the association between each of the SNPs and each of the genetic factor scores. The degree of relatedness (~50% for DZ twins and ~100% for MZ twins) was accounted for in the models using the GENMOD command in SAS 8.2 (SAS Institute, 2008). All p-value results from the association analyses were corrected for the number of independent tests conducted; the Nyholt correction indicated a significant threshold of p<0.005. Male and female data were collapsed in the genotypic analyses in order to maximize power to detect genetic association and to mirror the best fitting model from the twin analyses. Additionally, we conducted post hoc analyses of the association between DRD2/ANKK1 and the seven individual measures of alcohol consumption and problems in order to test whether using genetic factor scores would result in different conclusions than had we analyzed multiple individual measures of alcohol use/problems. When evaluating results for the seven alcohol phenotypes, the Nyholt correction indicated a significant threshold of a p<0.001 to take into account the additional tests.
