Episodic heavy drinking among college students: Methodological issues and longitudinal perspectives

Longitudinal data from 81 undergraduates (47 women and 34 men) were used for concurrent and predictive validation of binge drinking measures. Results suggest relative strengths,and weaknesses of different binge definitions. The conventional binge measure of greater than or equal to 5 drinks in a row (greater than or equal to 4 drinks in a row for women) yielded higher prevalence estimates and higher sensitivity but less specificity than other quantity-frequency measures using alcohol-related problems as the criterion. Alternative binge measures resulted in lower prevalence rates and sensitivity but higher specificity for alcohol-related problems. Only a subset of students exhibited heavy drinking patterns consistently over time. Such consistent heavy drinking was significantly more strongly associated with increased risk of adverse alcohol-related consequences.