In order to test whether the jump in drunkenness was specific to the transition between ages 18 and 19, we also tested whether the genotypic effect was significant if the transition point was moved forward or backward one year in time. This allowed us to define the window across which the observed genotypic effect was evident. The effect was found to be specific to age 18–19; the genotypic effect associated with the homozygous major allele was not significant at age 17–18 (p = 0.14) or age 19–20 (p = 0.61).
