Citation

Shahdah UE. Int. J. Traffic Transp. Eng. (Belgrade) 2020; 10(1): 48-60.

Copyright

(Copyright © 2020, City Net Scientific Research Center, Faculty of Transport and Traffic Engineering, University of Belgrade)

DOI

10.7708/ijtte.2020.10(1).05

PMID

unavailable

Abstract

An integrated empirical Bayesian (EB) model based on crash-conflict safety performance function (SPF) as a prior is presented for estimating treatment effects at treated sites. This model has the advantage of accounting for changes in before and after factors not associated with the treatment, such as changes in volume and other site-specific attributes. The model also has the potential to reduce the need for large reference sites to develop the crash prediction safety performance function (SPF) in the EB formulation. The integration of traffic conflicts in EB before-and-after analysis can be achieved in two steps: 1) Develop a SPF to relate crashes to simulated conflicts for untreated (reference) sites, and 2) Simulate traffic at the treated sites with and without the treatment. The former serves as the prior in the EB formulation, while the latter accounts for changes in conflicts resulting from the treatment being considered and serves as the EB data likelihood. To evaluate the proposed crash-conflict model, an EB before and after analysis was performed for a sample of treated signalized intersections from Toronto, Canada. The treatment being considered at these intersections is changing the priority of left turn signals from permissive to protected-permissive control. Treatment effect is obtained in terms of Crash Modification Factors (CMFs) for rear-end and left-turn opposing (LTOPP) simulated traffic conflicts. CMFs obtained from the proposed crash-conflict model were compared with estimates of CMF from a convention.

Language: en