Citation

Galgamuwa U, Dissanayake S. J. Transp. Saf. Secur. 2019; 11(4): 377-397.

Copyright

(Copyright © 2019, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439962.2017.1419523

PMID

unavailable

Abstract

This study investigates the safety effectiveness of commonly used lane-departure countermeasures on Kansas two-lane rural undivided road segments; 2-ft paved shoulders, centerline rumble strips, shoulder rumble strips, and shoulder & centerline rumble strips using crash modification factors (CMFs). Cross-sectional and case-control methods that are based on generalized linear regression assuming a negative binomial error structure and logistic regression models were used respectively. CMFs were calculated for all lane-departure and fatal and injury lane-departure crashes for tangent and curved road segments separately. Model validation of cross-sectional models was carried out using the mean of the absolute residuals and mean squared error (MSE). Classification tables were used to validate the models developed for the case-control method. Cross-sectional and case-control methods showed that all the countermeasures considered in this study were effective in reducing all lane-departure and fatal and injury lane-departure crashes except shoulder rumble strips on curved road segments for all lane-departure crashes. Even though similar results were obtained from both methods, CMFs developed using the cross-sectional method demonstrated a narrower range than the case-control method, indicating that the results from the cross-sectional method can be considered as more accurate and reliable than those from the case-control method.

Language: en

Keywords

case-control method; countermeasure evaluation; crash modification factors; cross-sectional method; lane-departure crashes