Exploring correlated effects on crash severity in Korea’s metropolitan areas: A random-parameter with interaction effect approach

초록

This study investigates the contributing factors to traffic crash severity in Korea’s metropolitan areas, focusing on Seoul and Gyeonggi, which account for approximately 41% of national crash occurrences. Using 10 years of crash data (2011–2020) from the Traffic Accident Analysis System (TAAS), a random parameter ordered probit model with interaction effects was applied to capture unobserved heterogeneity and the complex interplay among behavioral and environmental factors. Crash severity was categorized into three levels. The analysis identified that crash severity was significantly associated with speeding, right-curved road segments, rainy weather, and van-involved crashes. Moreover, interaction effects revealed that combinations of these factors further amplified crash severity under specific conditions, particularly at night or on downhill sections. The model incorporating interaction terms demonstrated improved goodness-of-fit over conventional specifications. The findings underscore the need for context-sensitive safety strategies in densely populated urban regions. Beyond their traffic safety implications, these results also contribute to sustainable transportation by supporting interventions that improve the safety, resilience, and inclusiveness of metropolitan mobility systems, particularly through more efficient safety resource allocation and enhanced protection of vulnerable road users. © 2026 Taylor & Francis Group, LLC.

키워드