With the increasing demand for sustainability, walking is being encouraged as a main active mode of transportation. However, pedestrians are vulnerable to severe injuries when involved in crashes, which can discourage road users from walking. Therefore, studying the factors that affect the safety of pedestrians is important. This paper investigates the relationship between pedestrian-vehicle crashes and various zone characteristics in the city of Vancouver. The goal is to assess the impact of socio-economics, land use, built environment, and road facility on pedestrian safety using macro-level collision prediction models. The models were developed using generalized linear regression and full Bayesian techniques. Both walking trips and vehicle kilometres travelled were used as the main traffic exposure variables in the models. The safety models showed that pedestrian-motorist crashes were non-linearly positively associated with the increase in traffic exposure. The crashes were also found positively associated with the socio-economic variables (i.e., employment and household densities), some built environment variables (transit stop, traffic signal, and light pole densities), commercial area density, and arterial-collector roads proportion. On the other hand, the models revealed a decline in the pedestrian-motorist crashes associated with the increase in the proportions of pedestrian-actuated signals and local roads, as well as the increase in the recreational and residential areas’ densities. The spatial effects were accounted for in the full Bayes models and were found significant, which imply the importance of considering spatial correlation when developing macro-level pedestrian safety models.

With the increasing demand for sustainability, the use of cycling as an efficient active mode of transportation is being encouraged. However, the vulnerability of cyclists to severe injuries in crashes can discourage road users from cycling. Therefore, the study of the factors that affect the safety of cyclists is important. This paper describes an investigation of the relationship between cyclist–motorist crashes and various traffic zone characteristics in Vancouver, British Columbia, Canada. The goal was to assess the impacts of socioeconomics, land use, the built environment, and the road facility on cyclist safety through the use of macrolevel collision prediction models. The models were developed by generalized linear regression and full Bayesian techniques. An actual bike exposure indicator (the number of bike kilometers traveled) and the number of vehicle kilometers traveled were used as exposure variables in the models. The safety models showed that cyclist–motorist crashes were nonlinearly associated with an increase in bike, vehicle, and transit traffic as well as socioeconomic variables (i.e., population, employment, and household densities), variables related to the built environment (transit stop, traffic signal, and light pole densities), commercial area density, and the proportion if arterial–collector roads. The models revealed, however, a decline in cyclist–motorist crashes in association with an increase in the proportions of local roads and off-street bike links and an increase in recreational and residential area densities. The spatial effects were accounted for in the full Bayes models and were found to be significant; such a finding implies the importance of consideration of the spatial correlation in the development of macrolevel cyclist safety models.