An Empirical Analysis of Bike Safety in Lawrence Using Road Geometry and Traffic Characteristics

Abstract

This study focuses on analyzing bike route safety in Lawrence using road geometry, infrastructure, and traffic characteristics. Bike crash incidence has been considered as a measure of bike route safety in this study. The independent variables considered for the bike route safety analysis are the number of lanes, route slopes (average), traffic volume (dummy variable based on functional road classification), the availability of bike routes, and posted speed limits. Bike crash data (the dependent variable), and Digital Elevation Model (DEM) data were collected from city of Lawrence. For the study purpose, streetwise bike route facilities, and traffic lane numbers have been updated based on city of Lawrence database. Finally, the average slope for each street in Lawrence has been calculated from DEM raster using ArcMap. As the data were characterized by over-dispersion and zero inflation, conventional negative binomial and zero inflated negative binomial models generate statistically significant variable coefficients. Interestingly, coefficients from both model have produced near identical bike compatibility maps for Lawrence. The study has found that bike route safety decreases with the increase of the traffic volume and lane numbers. In other words, collector and arterial roads are not the safest option for bicyclists in Lawrence, but the local neighborhood level streets are more suitable for biking. The route slope has no significant impact on bike route safety and the speed is negatively related with bike crash incidence. The unavailability of actual bike count data and bike speed data result in some flaws in the outcome of bike compatibility map. In a nutshell, complex statistical analysis adds some values in the current understanding of bike safety with the data available.