Demographic correlates were included in all models. Sex was coded with female as referent. Race/ethnicity was categorized using mutually-exclusive indicators for Black and Hispanic, with White as referent. Age was divided into 10-year intervals through 60+ years, with ages 18–29 as referent. Marital status was coded using indicators for never married, and separated/divorced/widowed, with currently married/living with partner as referent. Educational attainment was coded using three indicators for high school diploma/equivalent, some college, and four-year college/university degree, with less than high school as referent. Income was calculated in 2005 US dollars and coded using indicators for $20,001-$40,000; $40,001-$70,000; and $70,001 and up, with an extra indicator for missing income (approximately 13%), using income less than $20,000 per year as referent. Employment was coded with two indicators for unemployment and being out of the workforce (such as retired or homemaker), with full- or part-time employment as referent. Models also adjusted for whether the respondent reported their own religion discourages drinking; this was a dichotomous indicator, where all non-religious respondents were classified with those reporting that their own religion did not strongly discourage drinking (referent) (Michalak et al., 2007).
