Causational Analysis Of Rear-End Crash Severity With Reginal Heterogeneity

With socio-economy and infrastructure constructions being developed,the motorized travel activities in China keep escalating year by year.However,the roadway safety issues in China have become substantially imperative,due to the following de facto conditions: certain theoretical deficiency in accident causation research,the absence of public safety awareness,the inadequacy in safety reinforcement measures.In order to mitigate the property damage incurred by roadway accidents and provide roadway management authorities with theoretical support,it’s critically significant to explore the causational nexus between rear-end crash severity and a medley of contributory factors.For the technical sake of quality data tractability and analytical accuracy,crash records(2010~2014)from North Carolina and relevant socioeconomic and demographic data were collected,integrated and screened,which gave rise to a sum of 11408 rear-end crash records.Through analyzing the regional elements in rear-end crashes and the statistical distributions of involved factors,the latent tendency of regional heterogeneity is preliminarily embodied and the relationship between rear-end crashes and factors is identified.Considering the ordinal feature of the accident severity and the existence of regional heterogeneity,a discrete choice model,mixed effects ordered Probit(MEOP)model,is established.The paper selects different random effects for comparative analysis,and the model of best goodness-of-fit was find out.In order to reduce the influence of multicollinearity,stepwise regression methods were used to eliminate certain insignificant factors.Finally,the comparison of the coefficient estimates,goodness-of-fit and prediction accuracy between MEOP and standard ordered Probit were conducted,and MEOP model was methodologically substantiated to be strong for unveiling rear-end crash causation mechanism by its exceptional interpretative caliber.The estimation results show that a couple of significant contributors,including driver’s physical condition,restraint usage,alcohol involvement,driver’s gender,driver’s age,vehicle type,roadway surface condition,lighting condition,traffic control condition,annual average daily traffic and road segment speed limit,regional education level and population density,significantly shape the regional heterogeneity.The marginal effect value indicates that fine physical condition of the driver,strict restraint usage can significantly reduce the severity of rear-end crash.The interaction between the speed limit value of the road section and the traffic control conditions and traffic volume of the road section jointly affects the potential severity of the rear-end collision accident.Based on the research result,some advices about enhancing roadway safety are put forward.