Pedestrian safety at urban intersections: lighting conditions is the question

Sustainable and livable cities and non-motorized transportation modes are being increasingly promoted around the world. A considerable number of studies have investigated various factors associated with the severity of pedestrian crashes. However, little attention has been paid to the particular impacts of lighting conditions. This paper aims to differentiate between different lighting conditions based on solar altitude by developing separate Generalized Ordered Probit (GOP) models for the four distinct lighting conditions (i.e. daylight, twilight, dark, and lighted) analysing pedestrian injury severity at urban intersections. The models are tested using a set of log-likelihood ratio tests, which show there are key differences in the significance of the estimates, magnitude, and even the direction of the effects of explanatory variables, in which such differences may not be revealed by calibrating one aggregate model. Notably, built environments, age, gender of both pedestrian and driver, and speed limits were found to have different impacts on the severity of the crash.