"Another paradigm used to assess crash risk is to use cross-sectional surveys of reported nonfatal accidents that can be related to the presence of risk factors, such as alcohol and cannabis consumption. Such a methodology was employed in a provocative dissertation by Laixuthai (1994). This study used data from two large surveys that were nationally representative of high school students in the United States during 1982 and 1989. Results showed that cannabis use was negatively correlated with nonfatal accidents, but these results can be attributed to changes in the amount of alcohol consumed. More specifically, the decriminalization of cannabis and the subsequent reduction in penalty cost, as well as a reduced purchase price of cannabis, made cannabis more appealing and affordable for young consumers. This resulted in more cannabis use, which substituted for alcohol consumption, leading to less frequent and less heavy drinking. The reduction in the amount of alcohol consumed resulted in fewer nonfatal accidents."
Laberge, Jason C., Nicholas J. Ward, "Research Note: Cannabis and Driving -- Research Needs and Issues for Transportation Policy," Journal of Drug Issues (Tallahassee, FL: School of Criminology & Criminal Justice, Florida State University, 2004) Volume 34, Number 4, pp. 974-5.