Sarah Guarino

This paper investigates the influence of bicycle lanes on the nature and frequency of cyclist motor-vehicle collisions in the city of Toronto. This is accomplished through a binary logistical regression model with the presence or absence of a bicycle lane as the independent variable. The result of this analysis is a model that assesses the most important factors contributing to bicycle motor vehicle collisions on these two types of roads. Collision frequency is approximated based on the number of collisions per road segment.

The primary data source used in this investigation is a database ofbicycle-motor vehicle collisions reported to the police from 1998 to 2003. As a result of this investigation it has been determined that collisions on and off bicycle lanes do not differ significantly enough to enable a reliable differentiation of these two collision locations using a binary logistic regression model. Collision frequency however does differ significantly between roads with and without bicycle lanes, most likely due to a higher cyclist volume in areas where bicycle lanes are present.