Evaluation Of The Safety Effectiveness Of The Conversion Of Two-Lane Roadways To Four-Lane Divided Roadways: Bayesian Versus Empirical Bayes

Abstract

This paper uses various observational before-after analyses to evaluate the safety effectiveness of widening urban and rural two-lane to four-lane divided roadways. The methods range from simple (naive) before-after, before-after with comparison group, empirical Bayes (EB), and Bayesian approach. The EB method requires safety performance functions (SPFs) to be calibrated; the simple SPF based on annual average daily traffic (AADT) is used widely. In this paper, two sets of negative binomial models are calibrated: The full SPF model, which uses various explanatory covariates, and the simple SPF, which uses AADT only. The preliminary results from the calibrated models indicate that the SPF is pivotal in the EB method; the more accurate the models, the more pragmatic the evaluation of the safety effectiveness of a treatment. The proposed method of using the full SPF in the EB method is recommended over the conventional EB observational before-after. To obtain more reliable estimates, the Bayesian before-after approach is performed. The Bayesian bivariate Poisson-lognormal approach provides comparable results and may have several advantages over the EB technique. The results from this paper indicate that the conversion from two-lane roadways to fourlane divided roadways results in a notable reduction in fatal and injury crashes of more than 63% on urban roadways and 45% on rural roadways. Conversion to a four-lane divided roadway produces a higher reduction in total and property damage only crashes in urban areas than it did in rural areas. In addition, the safety effects of the conversion appear to be more effective on roadway segments in urban areas with a high AADT value.

Publication Date

1-1-2015

Publication Title

Transportation Research Record

Volume

2515

Number of Pages

41-49

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.3141/2515-06

Socpus ID

84976273204 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84976273204

This document is currently not available here.

Share

COinS