Citation

Galgamuwa U, Vahl CI, Dissanayake S. J. Transp. Saf. Secur. 2020; 12(6): 764-781.

Copyright

(Copyright © 2020, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439962.2018.1545714

PMID

unavailable

Abstract

The cross-sectional method has been widely used in research to estimate the effect of a parameter toward the target outcome. This method has been used to evaluate the safety effectiveness of countermeasures where the date of implementation of respective countermeasure is not known. However, in most cases the explanatory variables in such models are accounting only for the geometric and traffic-related characteristics; hence, the safety effectiveness estimated using the regression parameters may result in having either nonsignificant or under/overestimation of the safety effectiveness. Therefore, this research incorporates driver and environmental characteristics into the cross-sectional model to estimate the true safety effectiveness of the countermeasures. Generalized linear mixed models based on the Poisson distribution were used to develop cross-sectional models that incorporate driver and environmental characteristics as well as conventional cross-sectional models incorporating only road geometric and traffic-related characteristics. Akaike Information Criterion was used to compare the models. The results showed that the proposed models have better model fitness than the conventional models. Finally, it was found that the conventional models underestimate the safety effectiveness of most countermeasures on tangent road segments and overestimate the safety effectiveness of most countermeasures on the curved road segments.

Language: en

Keywords

crash modification factors; crash prediction; Cross-sectional models; environmental characteristics; generalized linear mixed models; human behaviors; traffic safety