Performance Measures for Prioritizing Highway Safety Improvements Based on Predicted Crash Frequency and Severity

Abstract

The goal of this research was to develop highway performance measures that could be used to prioritize safety improvement projects by utilizing results from predicted crash frequency and crash severity for roadways the state of Alabama. The models used were derived from collected crash specific, roadway infrastructure, spatially related socio-economic and demographic, and roadway demand data and analysis of the derived performance measures. The research also implemented a novel approach at modeling crash frequency by using an ordered probit model, which has seldom been used in previous crash frequency studies, although frequently applied in crash severity studies. Growth trends for applicable demographic and roadway demand factors from both models were computed and forecast for a horizon year of 2025 to predict crash frequency and crash severity fifteen years into the future. Two different performance measure calculations were made based upon most likely and probabilistic scenarios for safest, median, and worst case observed crash severity scenarios for both 2010 and 2025. Results from the performance measure calculations, and their derived cartographic and graphical representations, suggest that the results of this research can be used as practical and integral tools for all persons working in the roadway safety realm.