| China is a large country in the production and use of electric two-wheelers.Electric twowheelers not only bring convenience to residents’ daily commuting,but also cause many traffic safety issues.In recent years,there have been frequent collisions between electric two-wheelers and motor vehicles in China,seriously threatening the travel safety of vulnerable road users and restricting the smooth operation of the transportation system.Therefore,it is urgent to clarify the accident mechanism.In addition,there are often unobservable heterogeneity in traffic accident data,which can have a certain impact on the accuracy of research results.Therefore,this paper considered the heterogeneous effects of accidents and established various accident severity analysis models.This paper attempts to deeply analyze the factors that affect the severity of collision accidents between electric two-wheelers and motor vehicles,and propose reasonable and feasible accident prevention measures.This is of great significance for improving the traffic environment and reducing the risk of accidents.Firstly,the research background of this article was discussed,and the shortcomings of existing research on collision accidents between electric two-wheelers and motor vehicles were analyzed.The main research content of this paper was introduced according to the lack of existing research.Based on the data of electric two-wheelers and motor vehicle collision accidents in Shandong Province from 2012 to 2021,a sample database of accidents was established,and the data were screened and cleaned.The potential factors that affect the collision accidents between electric two-wheelers and motor vehicles were comprehensively analyzed from four aspects: people,vehicles,roads,and the environment.Through Pearson correlation analysis,16 influencing factors that are significantly related to the severity of the accident and unrelated to each other were identified.Secondly,in order to determine the necessity of using heterogeneity models and fully consider individual and group heterogeneity,standard multiple logit models that do not consider heterogeneity,mixed logit models that consider heterogeneity,latent category logit models,and mixed logit models based on latent category analysis were selected for analysis.The basic theory and advantages and disadvantages of the models were introduced in detail.The calculation methods of model parameter calibration,Goodness of fit test,prediction accuracy and average marginal utility were described in detail.Thirdly,the 16 factors obtained were used as independent variables and the severity of the accident was used as the dependent variable to construct an analysis model for the severity of collisions between electric two wheeled vehicles and motor vehicles.The effect of the model was evaluated from three aspects: prediction accuracy,goodness of fit and heterogeneity capture ability,and the optimal analysis model for the severity of the collision between electric twowheelers vehicles and motor vehicles was determined.The comparison results of the four models showed that the goodness of fit and prediction accuracy of the heterogeneity model are better than the accident severity model without considering the heterogeneity.In the heterogeneity model,the goodness of fit,prediction accuracy and heterogeneity capture ability of the mixed logit model based on latent class analysis are better than the mixed logit model and latent class logit model.Therefore,it was concluded that the mixed logit model based on latent class analysis is the optimal analysis model for the severity of electric two-wheelers and motor vehicle collision accidents.Finally,the impact of each factor on the severity of the accident is quantitatively analyzed by calculated the average marginal effect of the optimal model,and relevant preventive measures and suggestions were proposed based on the analysis results.The analysis results of influence factors showed that there are similarities and differences in the influence of various factors on accidents of different severity.Nine factors,including male rides and rides’ drinking and driving,have a significant impact on minor injury accidents and serious injury/death accidents.Among them,the impact of male rides on minor injury accidents and serious injury/death accidents has the opposite trend,while the impact of other eight factors on the severity of accidents has the same trend.Six factors,including adverse weather and non motorized vehicle lanes,only have a significant impact on minor injuries,while non side collisions and urban roads have a significant impact on serious injuries/fatalities. |