Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, Florida, 32816-2450, United States.
Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, Florida, 32816-2450, United States.
Accid Anal Prev. 2019 Sep;130:91-98. doi: 10.1016/j.aap.2018.05.015. Epub 2018 May 31.
Pedestrians are considered the most vulnerable road users who are directly exposed to traffic crashes. With a view to addressing the growing concern of pedestrian safety, Federal and local governments aim at reducing pedestrian-involved crashes. Nevertheless, pedestrian volume data are rarely available even though they among the most important factors to identify pedestrian safety. Thus, this study aims at identifying surrogate measures for pedestrian exposure at intersections. A two-step process is implemented: the first step is the development of Tobit and generalized linear models for predicting pedestrian trips (i.e., exposure models). In the second step, negative binomial and zero inflated negative binomial models were developed for pedestrian crashes using the predicted pedestrian trips. The results indicate that among various exposure models the Tobit model performs the best in describing pedestrian exposure. The identified exposure-relevant factors are the presence of schools, car-ownership, pavement condition, sidewalk width, bus ridership, intersection control type and presence of sidewalk barrier. It was also found that the negative binomial model with the predicted pedestrian trips and that with the observed pedestrian trips perform equally well for estimating pedestrian crashes. Also, the difference between the observed and the predicted pedestrian trips does not appear as statistically significant, according to the results of the t-test and Wilcoxon signed-rank test. It is expected that the methodologies using predicted pedestrian trips or directly including pedestrian surrogate exposure variables can estimate safety performance functions for pedestrian crashes even though when pedestrian trip data is not available.
行人被认为是最易受伤害的道路使用者，他们直接暴露在交通事故中。为了解决行人安全日益受到关注的问题，联邦和地方政府旨在减少涉及行人的交通事故。然而，即使行人数量是识别行人安全的最重要因素之一，但它们的数据却很少。因此，本研究旨在确定交叉口行人暴露的替代度量。采用两步法实现：第一步是开发 Tobit 和广义线性模型来预测行人出行量（即暴露模型）。在第二步中，使用预测的行人出行量开发负二项式和零膨胀负二项式模型来预测行人事故。结果表明，在各种暴露模型中，Tobit 模型在描述行人暴露方面表现最佳。确定的与暴露相关的因素包括学校的存在、汽车拥有量、路面状况、人行道宽度、公交车乘客量、交叉口控制类型和人行道障碍物的存在。还发现，使用预测的行人出行量和观察到的行人出行量的负二项式模型在估计行人事故方面表现同样良好。此外，根据 t 检验和 Wilcoxon 符号秩检验的结果，观察到的和预测到的行人出行量之间的差异似乎没有统计学意义。预计即使在没有行人出行数据的情况下，使用预测的行人出行量或直接包含行人替代暴露变量的方法也可以估计行人事故的安全绩效函数。
