Spatial econometric models to understand factors affecting older drivers at accident hotspots

Abstract

Road accidents involving older drivers tend to cluster geographically rather than occur randomly due to collective road conditions, traffic environment, and driver characteristics. This paper analyses the spatial distribution and causal factors of road accidents involving older drivers in the West Midlands region of the United Kingdom. Local Indicators of Spatial Association (LISA) was applied to accident data from 2006 to 2016 to identify accident hotspots and measure spatial clustering. LISA and Moran’s I (0.341) of accident data indicate that road accidents involving older drivers cluster in major urban centers within the West Midlands, particularly near complex junctions and dense traffic areas. This paper later applies the spatial lag and spatial error models to examine the existence of spillover effects and spatially autocorrelated errors, respectively. The spatial diagnostic tests, such as log likelihood, Akaike Information Criterion (AIC), Bayesian information criterion (BIC), and likelihood ratio test probability, indicate that the spatial error model best fits the observed accident data. The spatial error model identified that journey purpose, location, and types of junctions, poor lighting, road surface condition, weather conditions, gender, and time of day are the most significant predictors of accident risk involving older drivers. Noticeably, older drivers are exposed to accident risks during school runs, social trips, and navigating complex junctions due to slower reaction times, cognitive decline, and reduced ability to interpret dynamic traffic conditions. The findings of spatial models provide actionable insights for policy interventions at both local and national levels. Policymakers can improve mobility and safety for older drivers by focusing on environmental design, driver assessment, technology, and educational programs.