Drinking-and-driving in the United States: 1983-2017

Drinking-and-driving remains a leading cause of preventable mortality and morbidity in the United States. For the past three decades, estimates of the prevalence of drinking-and-driving have come from survey-based instruments, most notably the National Roadside Survey (NRS), which seeks to ascertain the blood alcohol content (BAC) of a random sample of road users. Although subject to potential bias from non-random sample selection, the NRS field design overcomes the well-documented problem of systematic under-reporting of socially undesirable activity. Nonetheless, in 2015 the House of Representatives voted to prohibit the use of federal funds to plan or administer the NRS over concerns that drivers did not believe participation was truly voluntary. As a result, researchers and policy-makers currently lack reliable estimates of drinking-and-driving that can be compared to historic trends. Therefore, in this article, we apply methods introduced by Levitt and Porter (2001) to examine how the prevalence, relative risk, and externality of drinking-and-driving in the United States has evolved over the past three decades. Similar to previous versions of the NRS, we find that drinking-and-driving declined significantly between 1983 and 2007, but has since plateaued. Interestingly, the relative risk of drinking drivers has increased while the external social cost of their behavior has decreased, suggesting that ‘sober-biased technical change’ may have made all driving less dangerous, but had a smaller impact on drivers influenced by alcohol.