J. Wang, Y. Li, X.J. Ji, S.E. Fang
In this research a binary logistic model is used to identify potential factors related to motorcycle and electric bicycle accidents at arterial road access points. Motorcycle and electric bicycle accidents are considered as dependent variables in the modeling process. Accident severities (death and injury), speed of vehicles, vehicle drivers’ behavior, cyclists’ behavior, cyclists’ characteristics, weather, road characteristics, types of vehicles, collision types and time of accidents are used as variables in the model. Akaike Information Criterion (AIC) is used as a criterion to optimize the model. Each variable is selected step by step. Odds ratio (OR) shows the change of significant factors on the motorcycle or electric bicycle accidents. Model estimation results indicate that the speed of vehicles (<40km/h) and cyclists’ gender have the most significant impact on accidents. The interesting findings are that electric bicyclists are more often involved in accidents than motorcyclists. Speed under 40km/h results in more motorcycle accidents at access points than high speed. Electric bicyclists are more likely to be involved in head-on accidents compared with motorcycles. More motorcycle accidents happen in the evening than electric bicycle accidents, as well as on 6-lane roads without traffic control and green barriers between vehicle and cycle lanes.
Keywords: binary logistic model; motorcycle accidents; electric bicycle accidents