关键词: 交通工程, 潮汐车道使用意愿, 潜类别 Ｌｏｇｉｔ模型, 心理潜变量, 平均边际效应

Abstract: Using tidal lanes is an effective traffic organization method to solve tidal traffic congestion. It is of great significance to study the influencing factors of tidal lane usage intention. The researches on tidal lanes mainly focus on the theoretical innovation of its setting mode, the investigation and analysis of its implementation effect, traffic safety and other aspects, while the researches on the intention of using tidal lanes from personal and psychological levels are few. In order to study the influencing factors of drivers’ willingness, the latent class Logit model is used to explore the unobserved heterogeneity in the willingness. The psychological latent variables of the drivers’ tidal lane setting are obtained through questionnaires, and a structural equation model is used to quantify the psychological latent variables. The thesis compares and analyzes the fitting effect and heterogeneity effect of the latent class Logit model considering only personal attributes and added latent variables. Then, an average marginal effect analysis is performed. First of all, the basic principle of the model construction is expounded. Secondly, this paper designs a questionnaire, obtains data through the questionnaire survey, and conducts descriptive statistical analysis and data tests. Thirdly, the significantly influencing factors are screened, and the fitting effect and heterogeneity effect of the latent Logit model, which only considers personal attributes and added latent variables, are compared and analyzed.Parameter estimation and average marginal effect analysis are then carried out to provide reference for improving the efficiency of tidal lane use. The results show that: (1) Data are obtained from personal attributes, psychological latent variable attributes and scene hypothesis attributes, and insignificant factors are eliminated. The latent Logit model with added psychological latent variables screens out the variables such as an age of 18-29 years old, those who are below a college degree, those who have children, those who own a car, driving for 0 years, driving for more than 5 years, frequent driving, robust driver, behavior attitude, perceived risk, social influence, travel time and isolation facilities having significant effects on dependent variables. (2) Compared with the Logit model which only considers personal attributes, the latent Logit model with psychological latent variables has better fitting effect and heterogeneity effect. For each additional level of attitude, the probabilities of choosing to use the tidal lane and taking a detour increase by 4.59% and 9.33%. For each additional level of social influence, the probabilities of choosing the tideway and detour increase by 16.00% and 11.92%. With the improvement of drivers’ behavior attitude and social influence on the tidal lane, more drivers will choose to use the tidal lane. (3) The latent Logit model divides data into three categories, accounting for 11.2%, 11.5% and 77.3% respectively. Owning a car, driving for more than 5 years and the perceived risk show significant heterogeneity. (4) With each increase of the perceived risk level, the probabilities of choosing to use the tidal lane and detour other roads decrease by 20.13% and 24.1%, while the probability of choosing to drive only in the conventional lane increases by 22.83%. As the perception of the risk of tidal lanes increases, more drivers will choose to drive only in the conventional lanes. It provides reference for improving the efficiency of tidal lane use.