Due to the potential decrease in available funds for highway improvements, it is crucial to correctly prioritize safety improvement projects in order to realize the maximum possible return on investment. In this research, safety performance functions (SPFs) for two-lane roads in Virginia were developed for use in conjunction with SafetyAnalyst, a computerized analytical tool that can be used for prioritizing safety projects. A SPF is a mathematical relationship between frequency of crashes by severity and the most significant causal factors, on a specific type of road. Although the SafetyAnalyst Users Manual presents six SPFs for two-lane segments, they were developed using data from Ohio, North Carolina and Minnesota. It is therefore necessary to calibrate these SPFs or develop valid SPFs for individual states using appropriate data from each state. SPFs for two-lane roads in Virginia were developed using annual daily traffic and corresponding crash data. These SPFs were developed for both total crashes and combined fatal plus injury crashes through generalized linear modeling, using a negative binomial distribution. Models were developed for urban and rural areas while accounting for the different geometric and topographical characteristics that exist in the different parts of Virginia. A comparative analysis based on the Freeman-Tukey R2 coefficient (RFT2) and the mean squared prediction error (MSPE) was then carried between the relevant Ohio SPFs and those for Virginia, to determine which set of models better fitted the Virginia data. In general, the results indicated that the SPFs developed with the Virginia data fitted the Virginia data better than the Ohio SPFs.
Abstract