Online Program

College Safe Ride Programs: Bayesian Analysis of Ridership with Safe Ride Home Versus Anywhere Policies (309544)

Keywords: Bayesian, Safe Ride Program, Colorado State University

Safe ride programs are implemented by college campuses around the country as part of harm reduction efforts to prevent drunk driving by college students. These programs typically pick up patrons at any location but sometimes place limitation on the drop-off destinations serviced. The effects of these drop-off destination limitations on ridership has not been quantified. We evaluate a major policy change in Colorado State University’s safe ride program, RamRide, which transitioned from a safe ride home to a safe ride anywhere in Fort Collins, CO. We evaluate the effect of this change on ridership. We use a Bayesian fixed effects Poisson regression model to estimate the effect of changing from safe ride home to safe ride anywhere. The approach uses data matched by day of week and week of semester for years before and after the policy change. We use an elliptical slice sampler to simulate the posterior distribution from the Poisson GLM. An R package is developed to test this model and is posted on GitHub for full reproducibility.