“…Wakefield, Gerlinger, Domina, Bruckner, and Carpenter (2016) show that a host of detailed state demographic and political variables-which have been shown in prior work to correlate strongly with the timing of state adoption of criminal justice policies such as three strikes laws (Karch & Cravens, 2014) and other public health policies such as clean indoor air laws (i.e., smoking bans; Shipan & Volden, 2006)-cannot readily explain the timing of adoption of a state ATOD education policy, consistent with our core identifying assumption that the variation in when states began to require such education is plausibly unrelated to the unobserved determinants of youth substance use in a state. For a content analysis of ATOD educational standards, see Bruckner et al (2014). race, and Hispanic ethnicity), a dummy for being not married, dummy variables for living in a large standard metropolitan statistical area (SMSA) and any SMSA other than a large SMSA, father's education (high school or less, some college, bachelor's degree or more, and an indicator for father education missing), mother's education (same variables as for the father), and dummy variables for missing data on sex and marital status.…”