Comparison of Methods for Identifying Intersections with High Crash Risks Using Field Data

This paper describes how the ability to identify intersections with high crash risks is crucial for the safety of the transportation system and the agencies responsible for traffic control and transportation planning. Various methods have been developed to identify intersections with high crash risk over the years that range from a simple counting method to sophisticated methods based on advanced statistical models. Accident counts or rates at specific locations may vary with time due to statistical fluctuations or to changes in the mean. This paper presents the research results from a study that uses field data to: (1) examine the performance of current methods; (2) determine the validity of the various underlying assumptions made within these methods; and (3) determine the stochastic property of accident data.