We used the variance inflation factor (VIF) to test for collinearity among our potential covariates. Due to the high VIFs (approximately 100–300), race/ethnicity distribution, religious adherents distribution, and percent living in rural areas were not included in the models. Their collinearity with state fixed effects is not surprising, as many of these covariates may be relatively constant over time. Age distribution, per capita income, and the alcohol excise tax variables also had relatively high VIFs (range 13 to 54). However, their VIFs were much less than the other covariates and were included in the models because they were significant for at least one alcoholic beverage type. The final models adjusted for state unemployment rate, per capita income, age distributions, and alcohol excise taxes in addition to state and year fixed effects.
