of what drives early and problem drinking behaviors in AAs as, although these behaviors are less common, they are not absent. Given the research suggesting that AAs who do use alcohol experience more severe AUD symptoms (Mulia et al., 2009) than their EA counterparts, understanding AUD etiology in this population remains an important step in the development of effective prevention efforts. We did not foresee the lack of association between trauma exposure and PTSD with alcohol outcomes in AA women. This lack of relationship could be due to limited AA sample size, but may also be explained by different patterns of choices of substances. AAs are more likely to use cannabis prior to alcohol (Sartor et al., 2013a), and recent findings in the same sample cohort did find trauma exposure increased the risk of cannabis initiation (but not the transition to cannabis problems) in AA women (Werner et al., 2016b). To address the possibility that cannabis use mediates the relationship between traumatic exposure and alcohol involvement, we re-ran all three models for AAs removing cannabis use from the model. Although hazard ratios associated with traumatic exposure slightly increased, the results did not reach statistical significance. Future research would benefit from
