Analysts from SAMHSA assisted in completing the analyses because state variables were not in the public domain. NSDUH generally requires at least two years of data for precise prevalence estimation at the state level. Because in 2009 and 2010 the Colorado medical marijuana industry experienced rapid growth, we conducted analyses pairing years 2003-4, 2005-6, 2007-8, 2009-10 and 2010-11. We decided a priori to conduct our analyses within three age groups (12-17, 18-25 and ≥26). We focused on an adolescent group (12-17) given that there has been concern about the effects of medical marijuana specifically on adolescents (Thurstone et al., 2011; Salomonsen-Sautel et al., 2012). We divided adults into those 18-25 years (young adults) and those 26 and older (other adults) because new-onset marijuana abuse and dependence is uncommon after age 25 (Stinson et al., 2006). Analyses were performed in SUDAAN based on weighted data to account for the complex survey design, following the established methodology previously described (SAMHSA, 2010b). Within-Colorado and within our pre-specified age groups our estimates are provided for rates or numbers for the perceived risk, availability of
