Selected variables.

This study aims to identify the key factors contributing to the severity of road accidents in Korea, analyzing more than 3,000 motor vehicle insurance records using the generalized ordered logit model (GOLogit). The model addresses the limitations of the parallel regression assumption, which ignores the differences between adjacent discrete levels of injury severity. The variable “Vehicle type (including pedestrian) with less liability”, which has been rarely examined in previous studies, demonstrated that individuals in the less responsible and more vulnerable position tend to suffer more severe injuries in South Korea. Consistent with this, the GOLogit estimates showed particularly high log-odds for severe injuries among pedestrians (4.912) and non-motorized cyclists (4.746), while speed-limit violations substantially increased the likelihood of fatal outcomes (2.456). In contrast, population density exhibited a protective effect, reducing injury severity (scaled log-odds = −1.055). This pattern is similar to broader societal trends, where economically disadvantaged regions tend to experience more severe traffic-related injuries. Specific road structures, such as the traditional right-angled crossroads, access roads to arterial roads, speedbumps on curved roads, and junctions between motor vehicle roads and sidewalks, pose significant safety challenges. Based on these findings, government policies on road safety should emphasize lowering the speed limits in residential areas, expanding the implementation of international pedestrian protection safety standards, and investing equitably in the safety of poor, low-population-density regions and older adults.