Other notable limitations of our study include the possibility of retrospective recall bias; however as the sample is relatively young, we anticipate the effects of recall bias to be minimal. Second, as our sample consists of young women, results may not extrapolate to other demographic groups. Third, the binary indices used in this study reflect initiation and there is significant variability in the extent of cigarette and cannabis use that is not captured by them (i.e., used once or twice versus daily/problem users). Thus, we are uncertain about where in the spectrum of liability these race/ethnic differences may be occurring. Fourth, we did not assess whether participants were also smoking products that combined cannabis and tobacco. While the practice of adding tobacco to cannabis joints is uncommon in the present population (and rare in the United States; Ream et al., 2008), its role should be carefully examined in international samples where the practice is common (Belanger et al., 2011) even in individuals who do not report cigarette smoking (Belanger et al., 2013; Gage et al., 2014). We also did not
