Optimizing measures of drinking to predict alcohol-related consequences.

Abstract

Alcohol use surges in late adolescence, particularly among college students, and is associated with a wide variety of negative socio-emotional, academic, legal, and health consequences. Although most studies assess risk for these consequences through self-report of drinking quantity and frequency, ample evidence indicates that youth who drink alcohol can be inaccurate in reporting their drinking behavior and that estimates become increasingly biased as drinking increases. In the current study, we examined two approaches for optimizing self-reported alcohol consumption estimates to assess associations with drinking consequences. The primary sample included 572 college students (53.9 % female; 18-23 age; 61.4 % White, 20.5 % Black, and 18.1 % additional racial-ethnic groups) who completed surveys and a standard lab-based free-pour knowledge task. We tested associations between consumption and consequences (as moderated by body mass index) using both multiple regression with covariates and instrumental variable analyses. Indicators of inaccuracy in self-reported consumption met requirements for serving as instrumental variables. In all models, consumption and consequences were strongly associated, though predictive validity measures differed in which technique they favored. Consumption rates only predicted consequences when body mass index was less than ~34 (obesity class 1). Our results demonstrate that including assessments of reporting inaccuracy, alcohol knowledge, and body mass index when studying alcohol consumption may be important for reducing bias due to measurement error and predicting consequences of alcohol use.