This research was undertaken to develop an evaluation procedure to identify high-risk four-legged signalized intersections in VDOT’s Northern Virginia district by traffic movements and times of day. By using the developed procedure, traffic engineers are expected to be able to identify signalized intersections where the traffic crash occurrences under different traffic conditions for different times of day are more frequent than would normally be expected. Using generalized linear models such as negative binomial models, one safety performance function was estimated for each of nine crash population reference groups formed by three traffic crash patterns (crash patterns 1, 4, and 6) and four times of day (A.M. peak, mid day, P.M. peak, and evening off peak). Crash pattern 1 is a same-direction crash (rear-end, sideswipe or angle crash) that occurs after exiting the intersection; crash pattern 4 is a right-angle crash between two adjacent straight-through vehicle movements in the intersection; and crash pattern 6 is an angle or head-on or opposite sideswipe crash between a straight-through vehicle movement and an opposing left-turn vehicle movement in the intersection. The procedure developed in this study is based on the empirical Bayes (EB) method. Additional data do not need to be collected in order to use the EB procedure because all the data required for applying the EB procedure should be btainable from VDOT’s crash database and from Synchro input data that are already available to traffic engineers for traffic signal phase plans. Thus, the EB procedure is cost-effective and readily applicable. For easy application of the EB procedure, an EB spreadsheet was developed using Microsoft Excel, and a users’ guide was prepared. These are available from the author upon request. (Author/publisher)
Abstract