occurs after that period. Differences in each of these parameters (slope of drunkenness across adolescence, transition point (“jump”), and slope of drunkenness across young adulthood) were tested by genotype. The model was fit in the framework of hierarchical linear modeling (Byrk and Raudenbush 1992) given varying numbers and times of assessments across individuals. Unlike conventional growth curve models using a structural equation modeling approach, hierarchical linear modeling does not have equality assumptions on the numbers and times of follow-up assessments of individuals (Mehta and West 2000). Model fitting and parameter estimation were conducted using the Mixed procedure, Version 9.3 of SAS.
