bottle for beer, one glass for wine, and one drink (shot) for hard liquor. Frequency was multiplied by the amount usually consumed – taking into account the serving sizes for beer, wine, and hard liquor due to differences across these drink categories in ethanol content – to create the approximate yearly volume scores for beer, wine, and hard liquor separately, and then these three values were summed to create the total yearly alcohol volume score. The dependent variables were log transformed prior to use in the modeling analysis to reduce skew and to improve scaling for the Mplus maximum likelihood estimation procedure.
