Noli Brazil, Assistant Professor, Department of Human Ecology, University of California, Davis
David Kirk, Professor, Department of sociology, nuffield college at the university of oxford
Ride-sharing services started 10 years ago in select cities in the United States and has rapidly dispersed across the country. Some have argued that ridesharing could curtail the extensive amount of drunk driving that occurs in the country. Indeed, Uber, the largest ridesharing company in the United States, claimed on its website that a city with its services has fewer drunk drivers on the streets. This claim went largely untested until recently. While some studies found evidence that ridesharing presence is associated with a decline in drunk-driving related fatalities, others found either no decline or decreases in some cities but not in others. Two important factors may be contributing to these contradictory results. First, studies have examined a period (late 2000s-early 2010s) during which ridesharing was in its earliest stages of adoption for most cities. Second, studies estimated average effects rather than testing the moderating influence of various city-level factors such as population density and the quality of public transportation. The current project updates the first national study of ridesharing’s influence on drunk-driving fatalities by including recent years of data (2015-2017) and testing various factors that might explain differential associations between ridesharing presence and drunk-driving fatalities. The project will answer the following research questions: First, what is the association between Uber deployment and drunk-driving fatality rates in US metropolitan areas since 2014, the last year of data in the first study? Second, what city-level factors moderate the relationship between Uber deployment and drunk-driving fatalities? I will test various demographic and socioeconomic characteristics, the reliance on and quantity and quality of transportation options, and spatial aspects of various built environment variables.