By Peter Leasure and Robert Kaminski

The current study used a time series design with Ohio State Highway Patrol and Ohio Department of Public Safety data to explore whether sales of recreational marijuana increased incidents of operating a vehicle while impaired (OVI) in Franklin County, Ohio and vehicle crashes in Columbus, Ohio. Two different statistical models were used to ensure the robustness of the results (a generalized linear model and Linden's itsa). In other words, we would want to see the same results from both models for moderate support of a particular finding. The results were mixed regarding the impact of recreational marijuana sales on total OVIs in Franklin County, Ohio. The generalized linear model showed a statistically significant increase in OVIs after recreational sales began, but the itsa model failed to replicate that finding. The results were also mixed regarding the impact of recreational marijuana sales on marijuana OVIs in Franklin County, Ohio. The generalized linear model showed a statistically significant increase in marijuana OVIs after recreational sales began, but the itsa model failed to replicate that finding. However, both models failed to find that recreational sales increased vehicle crashes in Columbus, Ohio, meaning that there is modest evidence that recreational sales did not increase vehicle crashes in Columbus, Ohio. A sensitivity analysis that used a balanced time series (91 days before the intervention and 91 days after) was also utilized, and those results largely confirmed the above findings.

Ohio State Legal Studies Research Paper No. 915

Columbus, OH Ohio State University (OSU) - Michael E. Moritz College of Law