How Rural Is Rural? Agreement Between the FARS Rural/Urban Label and 2020 Census Urban Areas for U.S. Traffic Deaths, 2015–2024

Emmanuel Oti Nani 

U.S. rural road safety statistics, including the National Highway Traffic Safety Administration’s annual rural/urban fatality reports, classify each fatal crash by the rural/urban variable (RUR_URB) in the Fatality Analysis Reporting System (FARS). Population research, in contrast, usually defines rural with the Census Bureau’s urban areas. We measure how well the two definitions agree for every fatal crash with a usable location from 2015 to 2024 (355,591 crashes, 386,140 deaths), placing each crash on the 2020 Census urban-area map, held fixed across years, and comparing the result with its FARS label. The two definitions agree on 88.4% of deaths (Cohen’s κ = 0.77). Disagreement runs mainly in one direction: 17.5% of deaths in crashes outside Census urban areas carry an urban FARS label, against 5.8% in the opposite direction. These crashes cluster at urban edges, with a median distance of 0.65 km to the nearest Census urban area, compared with 9.76 km for crashes that both definitions call rural. The gap widened over the decade. The FARS rural share of deaths fell from 50.1% in 2015 to 41.1% in 2024, while the fixed-map share moved from 52.0% to 48.1%. The divergence is concentrated in a few states and, in New York and California, appears as a single step between 2017 and 2018. Rural road safety trends based on the FARS label therefore mix changes in where crashes occur with changes in how they are labeled. Analyses should state their rural definition and, for trends, hold it fixed. The county panel, crash-level crosswalk, and code are openly available.