Citation

Li J, Li C, Zhao X, Wang X. Transp. Policy 2024; 149: 21-35.

Copyright

(Copyright © 2024, Elsevier Publishing)

DOI

10.1016/j.tranpol.2024.01.021

PMID

unavailable

Abstract

Dockless bike sharing, also known as the shared bicycle industry, is booming, especially in China. Since cyclists are more vulnerable than motor vehicle drivers in traffic crashes, it is necessary to investigate shared bicycle traffic safety. Road network patterns and the proportion of different points of interest (POIs) are two critical macro-level factors influencing bicycle crashes. Therefore, it is necessary to consider bicycle traffic safety in road networks and land use policy in road traffic planning. This study investigated risk exposure, demographic data, proportion of different POIs, land use, road network features, and bicycle crashes in 124 census tracts in Beijing's Sixth Ring Road area. The betweenness centrality was calculated for the census tracts to classify the road network patterns. A negative binomial conditional autoregressive (NB-CAR) model was developed for bicycle total crashes, and bivariate negative binomial CAR (BNB-CAR) models were developed for bicycle single-vehicle (SV) and multi-vehicle (MV) crashes, property damage only (PDO) and injury crashes. The results show the following. 1) The BNB-CAR model had a better fit than the NB-CAR model. 2) The census tracts with parallel, mixed, and loops & lollipops patterns were associated with higher bicycle crash frequency than those with a grid pattern. The difference in the bicycle SV crash frequency between the mixed and loop & lollipop patterns was larger than that in the bicycle MV crash frequency. 3) Census tracts with higher proportions of POIs for subway and bus stations (T-POI) were associated with fewer bicycle crashes. 4) Census tracts with higher arterial proportions were associated with more injury crashes. This study provides a theoretical basis for formulating road network and land-use policies to ensure road traffic safety.

Language: en

Keywords

Bivariate negative binomial CAR model; Points of interest; Road network patterns; Shared bicycle crashes; Traffic safety planning