Analysis of Crash Severity Influencing Factors on Mountainous Expressways Based on Ordinal Logistic Regression
Year: 2026
Type: preprint
Abstract: Road traffic injuries remain a leading cause of death worldwide, and mountainous expressways in subtropical karst regions pose distinctive safety challenges related to complex horizontal–vertical alignments and freezing pavement. Using 7,739 crash records (2019–2023) from high-risk segments of expressways in Guizhou Province, China, we developed an ordinal logistic regression model to identify factors associated with crash severity, classified as property damage only, injury, or fatal. Six categories of explanatory variables were considered: holiday status, time of day, segment type, weather, vehicle combination, and a newly proposed 14-grade horizontal–vertical alignment coupling classification. The parallel lines test ( p = 0.078) supported the proportional odds assumption, and the likelihood ratio test ( χ ² = 272.978, p < 0.001) confirmed overall model significance. At the conventional α = 0.05 level, motorcycle involvement (OR = 24.31), truck–truck collisions (OR = 6.32), freezing pavement (OR = 1.95), midnight driving (OR = 1.29), and other vehicle configurations were associated with elevated severity, whereas holiday periods (OR = 0.61) and the G6 alignment (OR = 0.53) were protective. Closed-form probability equations were derived for scenario-based severity prediction. These findings provide quantitative benchmarks for safety governance of mountainous expressways.
Source: Research Square
Language: English
Topic: Traffic and Road Safety
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225888729
