Our study should be interpreted in the context of several limitations. Although the age range covered encompassed 14 to 25 years, participants did not have data across this entire range. Since data collection was not age-standardized, each participant contributes to the portion of the overall trajectory for which they have data, but different time points have somewhat different individuals contributing. The number of individuals contributing to each age point is included in Figure 1. In addition, although we assume that the jump in drunkenness found between ages 18 and 19 is attributable to changing environmental circumstances associated with the transition to adulthood, we have not explicitly included any measured aspects of the environment in the model. Finally, the analyses reported here are limited to the European American subset of a study that selected families rich in alcohol dependent subjects, and it is not clear if similar results would be seen in other subgroups. There are known allele frequency differences between populations and we did not want this to lead to spurious findings, however, we note that the overall pattern of results was similar when the entire sample (of which European Americans were 62%) was analyzed jointly.
