| Both-at-fault crashes refer to crashes which both drivers are responsible for the accidents. Compared to single-at-fault crashes or other types of two-car crashes, both-at-fault crashes account for a small proportion though, lead to comparatively more severe consequences. Therefore, crashes of this type are worth relevant scholars’ attention. Presently, investigations on both-at-fault crashes are springing up at national wide and worldwide, thereby, there are few reference conclusions and literatures. Given the above situation, by researching and analyzing both-at-fault crashes based on the accident data of Michigan in 2009, this paper aims to figure out influential factors that affect severity of both parties in both-at-fault crashes significantly, and give suggestions and solutions in order to decrease the consequences and improve traffic security.Initially, having established the principles of choosing both-at-fault crashes data, this paper chooses appropriate data and preprocesses it, then gets a pure analytical data source. Based on it, this paper uses descriptive statistics to analyze the features and rules of this type of crashes as well as the relationship between those influential factors and the highest severity in the crashes and makes a basic judgment. Thereafter, this paper establishes a Bivariate ordered probit model which can simultaneously cope with both drivers’ severity. By substituting the model with potential influential factors, we suppose the significant influential factors are:both drivers’ ages, gender, pre-accident operation, hazardous driving behaviors, vehicle models, collision type, road performance level, road type, traffic control method and lighting conditions, etc. We calculate these factors’ marginal effect and make a research on their impact on drivers’ injuring severity and their effect mechanism. Finally, from the perspective of education, enforcement and engineering, we give appropriate solutions to these important influential factors with the purpose to prevent these crashes from happening or decrease their consequences. |