Schorr Justin P, Hamdar Samer H
Department of Civil and Environmental Engineering, Center of Intelligent Systems Research, Traffic and Networks Research Laboratory, The George Washington University, 20101 Academic Way #201-I, Ashburn, VA 20147, USA.
Accid Anal Prev. 2014 Oct;71:93-105. doi: 10.1016/j.aap.2014.05.008. Epub 2014 Jun 2.
The objective of this study is to develop a safety propensity index (SPI) for both signalized and unsignalized intersections. Through the use of a structural equation modelling (SEM) approach safety is quantified in terms of multiple endogenous variables and related to various dimensions of exogenous variables. The singular valued SPI allows for identification of relationships between variables and lends itself well to a comparative analysis between models. The data provided by the Highway Safety Information System (HSIS) for the California transportation network was utilized for analysis. In total 22,422 collisions at unsignalized intersections and 20,215 collisions at signalized intersections (occurring between 2006 and 2010) were considered in the final models. The main benefits of the approach and the subsequent development of an SPI are (1) the identification of pertinent variables that effect safety at both intersection types, (2) the identification of similarities and differences at both types of intersections through model comparison, and (3) the quantification of safety in the form of an index such that a ranking system can be developed. If further developed, the adopted methodology may assist in safety related decision making and policy analysis.
本研究的目的是为信号控制和无信号控制的交叉口开发一种安全倾向指数（SPI）。通过使用结构方程建模（SEM）方法，根据多个内生变量对安全性进行量化，并将其与外生变量的各个维度相关联。奇异值SPI能够识别变量之间的关系，非常适合进行模型间的比较分析。利用高速公路安全信息系统（HSIS）提供的加利福尼亚州交通网络数据进行分析。最终模型中总共考虑了22422起无信号控制交叉口的碰撞事故和20215起信号控制交叉口的碰撞事故（发生在2006年至2010年之间）。该方法以及随后开发的SPI的主要好处包括：（1）识别影响两种交叉口类型安全性的相关变量；（2）通过模型比较识别两种类型交叉口的异同；（3）以指数形式对安全性进行量化，从而可以建立一个排名系统。如果进一步发展，所采用的方法可能有助于安全相关的决策制定和政策分析。
