Modeling Bicyclist Exposure to Risk and Crash Risk: Some Exploratory Studies
Greg Lindsey, J. Wang, Steve Hankey, Michael Pterka
Report no. CTS 18-15
This report presents models for estimating bicyclist exposure to risk and crash risk. Direct demand models for estimating weekday PM peak-period bicyclist exposure to risk are estimated from a database of PM peak-period bicycle counts in Minneapolis and used to estimate exposure for the street network. Bicycle crashes in Minneapolis are described and crash risk is assessed. Probability models to assess crash risk at both intersections and along segments show that both bicyclist exposure and vehicular exposure are associated with the likelihood of a bicycle crash. Estimates of exposure at 184 roadway-trail crossings are used to apply warrants for traffic controls. The results show that warrants for traffic signals and pedestrian hybrid beacons are most likely to be met using weekend peak-hour traffic flows. Most locations that meet warrants already have controls, but site specific safety investigations may be warranted at 9% of all crossings. Count-based models of bicyclist exposure are estimated for Duluth using origin-destination centrality indices as explanatory variables. Although these indices correlate positively and significantly with bicyclist volumes, estimates of exposure do not correlate with bicycle crashes. Together, these analyses illustrate how measures of bicyclist exposure to risk can be used in assessments of safety and crash risk. The approaches can be used in planning-level studies where consistent measures of exposure or risk are needed. These results underscore the need to continue bicycle traffic monitoring and make available estimates of exposure for safety assessments.