Anotace:
Lane-changing behavior can significantly affect many aspects of traffic flow including capacity, shock waves, and safety. Therefore, it is imperative to understand the determinants behind lane change behavior. This paper investigates the determinants of lane-changing in congested traffic using video-recording as well as a survey approach. A mixed logit model was estimated to account for unobserved heterogeneity in lane-changing behavior across drivers. Estimation results show that all categories of explanatory variables including socioeconomic, driving style, and road environment have a significant effect on lane changing behavior. Besides, unobserved heterogeneity and taste variation among drivers with regards to the lateral distance of the target vehicle from the left car has been observed. Among the non-random parameters, speed of target vehicle, being a law-evading driver, disregarding yellow traffic signals at intersections, lateral distance of target vehicle from right/left cars, and experiencing at least two accidents are positively associated with a higher likelihood of lane changing when a driver encounters a downstream queue. The aforementioned interesting findings can significantly help to improve the performance of traffic flow models for the purpose of replicating and predicting traffic flow.