Effects of intersection control types on driver yielding behavior to cyclists using mixed logit modeling

Cycling safety at intersections is a growing concern as both cycling activity and motor vehicle traffic continue to rise. Intersections pose heightened risks for cyclists due to complex traffic patterns, ambiguous right-of-way rules, and insufficient signaling, often leading to collisions. This study investigates how intersection control types and operational characteristics influence driver failure-to-yield behavior toward cyclists. Using ten years of Michigan crash data involving single motor vehicle-cyclist collisions, we apply a Mixed Logit Model to account for unobserved heterogeneity in driver behavior. The analysis focuses on three types of intersection control: traffic signals, stop/yield signs, and uncontrolled intersections, examining their impact on various driver-cyclist interaction scenarios. Key findings indicate that driver age, day of the week, vehicle type, and speed limit consistently affect yielding behavior across all control types. Impairment due to alcohol or drugs significantly increases the likelihood of hazardous driver actions. Drivers are more prone to fail to yield in straight-ahead scenarios, though they are less likely to be deemed at fault in non-yield crashes. Intersection control effectiveness also varies by maneuver type; signalized intersections reduce failure rates in straight-travel scenarios, while stop/yield signs are more effective during left turns. This research addresses a critical gap by linking infrastructure features with driver yielding performance, offering evidence-based insights for improving intersection safety. The findings support targeted interventions in roadway design, driver education, and the integration of advanced technologies such as cyclist detection systems and vehicle-to-vehicle communication to enhance cyclist protection.