The genetic variance accounted for by non-imputed variants did not differ significantly from zero; however, we do not interpret this result as indicating that genetic factors are not relevant for alcohol problems. First, the current results are consistent with twin studies demonstrating that environmental factors remain strongly influential on alcohol-related phenotypes into early adulthood [9]. Genetic influences remain relevant but account for less of the variance. Second, previous studies that have derived heritabilities based on measured genetic factors have also reported lower estimates than those derived from biometric structural equation modeling [57–59], inspiring the term “missing heritability” [10]. Our prediction of 0.5–0.6% of the variance in the split-sample analyses is comparable to the r2 estimate reported by Kos et al. [60], who used common alleles to predict risk for alcohol dependence in a European-American sample. Unmeasured common and rare variants likely contribute to missing heritability in the current and previous studies.
