Keeping Intersections Safe for Vulnerable Users: Contributory Factors to Pedestrians’ and Bicyclists’ Crossing Safety

Abstract

Pedestrian-and cyclist-involved crashes at intersections remain a major concern in China due to high ratio of fatal injuries and fatalities. This paper develops the random parameters negative binominal model to determine the relationship between potential contributory factors and crash frequency involving pedestrians and bicyclists at intersection locations and identify the significant risk factors, using 4,066 observations over the period 2008-2013 from Zhengzhou and Xi'an, China. The use of random parameters' model is utilized as a mean to accommodate possible unobserved heterogeneity across intersections, and a maximum likelihood method with Halton draws is proposed to estimate the maximum likelihood of random parameters. Additionally, marginal effects are applied to assess the impact of expected instantaneous change of each dependent variable on crash frequency while keeping all the other covariates constant. Among many potential factors, passenger countdown timer, passenger push button, number of left turn lanes, left turn traffic, ratio of major direction AADT to minor direction AADT and driver's disregarding 'give-way' sign behavior,… etc. have significant association with passenger crash frequency, but a less impact on bicyclist crashes. The findings of current study could provide an important insight into traffic safety improvement projects for passengers and bicyclists at both planning and operational levels.