Modeling social influences on public drinking.

This article combines the qualitative tradition of naturalistic observations with quantitative multivariate methods to quantify the relationships between drinker characteristics, group composition, duration of the drinking period and alcohol consumption. Employing observational data on over 6,000 drinkers collected in six on-premise licensed establishments in California, structural equation regression methods were used to estimate the simultaneous effects of duration of the drinking period and total consumption on each other, controlling for patron characteristics and group composition. Results from all six establishments indicate that consumption is higher in larger groups because of longer drinking times at the establishment and not because of any direct effect of consumption on length of the drinking period. Group composition effects on consumption are also indicated since females in mixed drinking groups tend to reduce the alcohol consumption of males, but not the group as a whole.