Pilot Evaluation of Risk Factors for Adverse Interactions between Drivers and Bicyclists


The study had two specific aims: 1) To demonstrate the feasibility of use of a simulator to measure vehicle/bicycle interactions under specific, replicable conditions; and 2) to evaluate potential risk factors for vehicle/bicycle crashes, and to assess the effects of infrastructure-based bicycle safety interventions on the margin of safety between vehicles and bicycles.

Final Report:

Bicycling is an active and sustainable alternative to automobile-based transportation, but may entail substantial injury risks. These risks are poorly understood due in part to the challenges inherent in measurements made in situ. Our study used an advanced automobile simulator to quantitatively evaluate vehicle-motor vehicle (BMV) crash risk associated with a range of road characteristics and specific infrastructure elements (e.g., roadway configurations and signage).

Thirty adult subjects (15 male and 15 female) ages 30-50 years old and having been licensed to drive for at least two years were recruited into the study. Subjects drove a predefined simulated route through urban, suburban, and rural environments. Participants were instructed to drive the virtual course as they would normally drive on an actual roadway under similar circumstances. Experimental scenarios encountered by subjects included nine unique combinations of the following conditions: presence or absence of bicycle lanes; presence or absence of bicycle signage; 35 or 50 MPH speed limit; presence or absence of a curb; and two or four lane roadway. Virtual bicyclists were encountered at various points in the course, with at least one bicyclist appearing in each of the experimental conditions. The appearance of bicyclists was triggered by driver behavior and course location, which resulted in the absence of certain bicyclists in some drives. In order to evaluate changes in driver performance related to the presence of bicyclists, for each segment of the course in which a bicyclist was present, another segment with comparable conditions but no bicyclist was used as a control segment.

The primary outcomes assessed were BMV separation and driver’s gaze direction as measures of crash risk, and vehicle speed as a measure of potential crash severity. Subjects also completed a survey that assessed demographics, driving experience, and driving- and risk-related knowledge, attitudes, and behaviors.

Our findings suggest that bicycle lanes and, to a lesser extent, bicycle signage may reduce crash risk by increasing BMV separation and reducing vehicle speeds. Four lane roads had substantially increased BMV separation (e.g., reduced crash risk) when compared to two lane roads, but also had increased vehicle speeds (e.g., increased crash severity). Conversely, vehicle speeds on roads with curbs were reduced, but BMV separation was diminished. Regression model results suggested that older and male drivers had decreased risk of BMV crash, while drivers with poor lanekeeping ability and a greater number of lifetime accidents were at increased risk of BMV crash. Overall, our results generally supported our first hypothesis, that roadways with narrow shoulders or curbs and roadways with higher posted speed limits were associated with greater risk of vehicle/bicycle crash. Our results also supported our second hypothesis, that bicycle lanes and bicycle signage reduced the risk of vehicle/bicycle crash.

Our study demonstrates that automobile simulator technology can be used to evaluate the efficacy of bicycle safety interventions and roadway configurations with limited human injury risk and without the necessity of expensive changes to real-world infrastructure.