Analysis on the heterogeneity of factors influencing the rear-end crash severity on highways
To explore the differences in the mechanisms leading to highway rear-end crash severity during day-time and nighttime,the correlated random parameter order Probit model with heterogeneity-in-means(CROPHM)was constructed to capture unobserved heterogeneity and explain the correlation between random parameters.A sample of 8 664 highway rear-end crashes was extracted from 2012-2018 Shanxi highway acci-dents dataset.With time variables,environmental variables and accident attributes as input indicators and the maximum injury severity as the output variable,the factors influencing the severity of rear-end crashes during the daytime and nighttime were explored separately.Firstly,likelihood ratio(LR)tests were conducted to vali-date whether rear-end crashes classified by daytime and nighttime should be modeled separately.Subsequently,two CROPHM models were established to examine the temporal differences in factors influencing the rear-end crash severity.Finally,through marginal effect analysis,corresponding recommendations were presented.The study findings indicate that rear-end crashes between 00:00 and 06:00 significantly increase the probability of fa-tal accidents by 0.063 7 compared with 18:00-24:00.Moreover,the probability of fatal crashes increases by 0.015 6 during nighttime multi-vehicle rear-end collisions compared with single-vehicle crashes.When large ve-hicles are involved in rear-end collisions,the probability of fatal accidents significantly increases,both during the daytime and nighttime.
traffic engineeringtraffic safetyrear-end crash severityheterogeneityordered Probit model
万方数据
维普
