|
Abstract
|
Introduction
We are just beginning to be able to detect the policy implications and public health effects of changes in marijuana laws and consumption in the United States on motor vehicle related fatalities. Most studies to date use classical regression methods to study these and are thus susceptible to both model and parameter uncertainty. This study examines the associations between marijuana and motor vehicle fatality rates taking these two issues of uncertainty into account using Bayesian sturdy-values, i.e., s-values.
Method
This study utilizes a new balanced panel dataset across all states and Washington, D.C. for the period 2010 to 2016 in the context of linear models using Bayesian s-values. It addresses the association between marijuana and alcohol consumption along with the legal environment across states and through time on crash fatalities. Other important factors such as the distractive influence of cell phones are studied. The s-value approach considers a vast number of model specifications and provides robust policy guidelines.
Results
A strong association between marijuana and alcohol use on motor vehicle crash rates is found. The statistical results are both substantial and robust, i.e., non-fragile. Other important variables include cell phone use, seat belt use, speed limit laws, and fleet modernization.
Conclusion
Our results have found strong evidence of a life-taking relationship between marijuana use and vehicle crashes. This suggests that policy makers recognize that legislation liberalizing marijuana use may have tragic ramifications regarding motor vehicle fatalities.
Keywords: Cannabis impaired driving
Language: en |