A New Methodology for Before–After Safety Assessment Using Survival Analysis and Longitudinal Data

Abstract

The widely used empirical Bayes (EB) and full Bayes (FB) methods for before–after safety assessment are sometimes limited because of the extensive data needs from additional reference sites. To address this issue, this study proposes a novel before–after safety evaluation methodology based on survival analysis and longitudinal data as an alternative to the EB/FB method. A Bayesian survival analysis (SARE) model with a random effect term to address the unobserved heterogeneity across sites is developed. The proposed survival analysis method is validated through a simulation study before its application. Subsequently, the SARE model is developed in a case study to evaluate the safety effectiveness of a recent red‐light‐running photo enforcement program in New Jersey. As demonstrated in the simulation and the case study, the survival analysis can provide valid estimates using only data from treated sites, and thus its results will not be affected by the selection of defective or insufficient reference sites. In addition, the proposed approach can take into account the censored data generated due to the transition from the before period to the after period, which has not been previously explored in the literature. Using individual crashes as units of analysis, survival analysis can incorporate longitudinal covariates such as the traffic volume and weather variation, and thus can explicitly account for the potential temporal heterogeneity.

Association of human mobility with road crashes for pandemic-ready safer mobility: A New York City case study

“…In this study, we propose survival analysis methods to relate three types of crashes (motorists, pedestrians, and cyclists) to various factors: human mobility, social economy, and weather conditions during the COVID-19 pandemic. The survival analysis involves modeling the time-to-event data and applying the hazard function to capture the conditional probabilities (i.e., the likelihood of crash occurrence based on frequency patterns; Chang and Jovanis, 1990 , Mannering, 1993 , Xie et al, 2019 ). We took the time interval between two crashes as a dependent variable in the survival analysis model, which can model the temporal heterogeneity for individual crash risk differences ( Aalen, 1992 ).…”

Quantifying the Influence of Road Geometric Parameters on Road Safety (Case Study: Hawassa-Shashemene-Bulbula Rural Two-Lane Highway, Ethiopia)

“…This method simply compares the observed crash data (frequencies and rate) between before and after period was among the earliest practice. Despite easy to use, such comparison usually leads to inaccurate and potentially misleading conclusion because the method may be affected by several factors, especially in case when only short-term data, 2-3 years, are available as cited in several literatures [29] [30] [31]. One of them is its lack of flexibility to account for the effect of change in traffic volume of crash occurrence.…”

“…Another well-known shortcoming is its incapability of addressing regression-to-the-mean (RTM) bias. RTM bias would be occurred due to random fluctuation of accident occurrence at a given location [31]. This makes likely that road sites with high number of accidents in any one year may show a decreasing trend in the subsequent year even without treatment of the road site [30].…”

Safety analytics at a granular level using a Gaussian process modulated renewal model: A case study of the COVID-19 pandemic

“…These models study the crash intensities at the aggregated time intervals. Instead of studying the crashes at the aggregate level, survival analysis models (or duration models) study the crash occurrences as a sequential process ( Jovanis and Chang, 1989 , Chang and Jovanis, 1990 , Shankar et al, 2008 , Chen and Guo, 2016 , Xie et al, 2019 ) at the disaggregate level. Jovanis and Chang (1989) discussed the advantages of using survival theory principles to combine discrete crash data and aggregate exposure data.…”

“…The widely used survival models such as Cox proportional hazard (semi-parametric) and accelerated failure time (parametric) models, however, rely on the assumptions of proportional hazard functions over time, which can be hard to satisfy in practice. Xie et al (2019) proposed a Bayesian survival analysis model which accounted for the crash hazard functions ( Hensher and Mannering, 1994 , Kalbfleisch and Prentice, 2011 ) during the consecutive crash times across sites. While this study modeled individual crashes, covariates such as traffic volumes, however, were aggregated during the crash intervals.…”