Evaluations of the impact of medical and recreational marijuana laws rely on quasi- or natural experiments in which researchers exploit changes in the law and attempt to determine the impact of these changes on outcomes. This chapter reviews three key issues of causal inference in observational studies with respect to estimating of impact of medical or recreational laws on marijuana use—intervention definition, outcome measurement, and random assignment of study participants. We show that studies tend to use the same statistical approach (differences-in-differences) and yet find differential impacts of medical marijuana laws on adult use in particular. We demonstrate that these seemingly conflicting findings may be due to different years of analysis, ages of the study sample in each year, and assignment of jurisdictions to the control group versus treatment group.