Citation

Lee J, Abdel-Aty MA, Xu P, Gong Y. Accid. Anal. Prev. 2019; 125: 116-123.

Affiliation

Department of Civil, Environmental, and Construction Engineering, University of Central Florida, Orlando, Florida, 32816, United States. Electronic address: gongyaobang@knights.ucf.edu.

Copyright

(Copyright © 2019, Elsevier Publishing)

DOI

10.1016/j.aap.2019.01.037

PMID

30739046

Abstract

In previous studies, the safety-in-numbers effect has been found, which is a phenomenon that when the number of pedestrians or cyclists increases, their crash rates decrease. The previous studies used data from highly populated areas. It is questionable that the safety-in-numbers effect is still observed in areas with a low population density and small number of pedestrians. Thus, this study aims at analyzing pedestrian crashes in a suburban area in the United States and exploring if the safety-in-numbers effect is also observed. We employ a Bayesian random-parameter Poisson-lognormal model to evaluate the safety-in-numbers effects of each intersection, which can account for the heterogeneity across the observations. The results show that the safety-in-numbers effect were found only at 32 intersections out of 219. The intersections with the safety-in-numbers effect have relatively larger pedestrian activities whereas those without the safety-in-numbers effect have extremely low pedestrian activities. It is concluded that just encouraging walking might result in serious pedestrian safety issues in a suburban area without sufficient pedestrian activities. Therefore, it is plausible to provide safe walking environment first with proven countermeasures and a people-oriented policy rather than motor-oriented. After safe walking environments are guaranteed and when people recognize that walking is safe, more people will consider walking for short-distance trips. Eventually, increased pedestrian activities will result in the safety-in-numbers effects and walking will be even further safer.

Copyright © 2019 Elsevier Ltd. All rights reserved.

Language: en

Keywords

Bayesian approach; Pedestrian safety; Poisson-lognormal model; Random-parameter model; Safety in numbers