Binge Drinking and Violence among College Students: Sensitivity to Correlation in the Unobservables

ImpacTeen Research Papers

Binge Drinking and Violence among College Students: Sensitivity to Correlation in the Unobservables
Powell LM, Ciecierski CU, Chaloupka FJ, Wechsler H.

This paper examines the relationship between binge drinking and violence-related outcomes among college students. Drawing on data from the 1997 and 1999 waves of the Harvard School of Public Health Alcohol College Study, we examine four violence-related outcomes that include: arguing, damaging property, trouble with the campus or local police, and injurt to oneself. We estimate a bivariate probit model to undertake sensitivity analyses based on different assumptions on the correlations of disturbances between drinking and violence. The bivariate probit results show that once we control for endogeneity based on rho = 0.1, binge and frequent binge drinking significantly affect all four violence-related outcomes but the magnitude of these effects is smaller than those suggested by the single-equation probit model by a factor of almost two for the binge drinking and by a factor of five in the frequent binge model. Our sensitivity analyses reveal that in order to conclude that binge and frequent binge drinking have no effect on violence-related outcomes, the correlation between the unobservables would have to be very large, approximately 0.4 to 0.5.

Research Paper (PDF – 270KB)

Leave a comment

Your email address will not be published. Required fields are marked *