TRB Pubsindex
Text Size:

Title:

Before-After Safety Evaluations Using Parametric Bayesian Survival Analysis

Accession Number:

01519516

Record Type:

Component

Availability:

Transportation Research Board Business Office

500 Fifth Street, NW
Washington, DC 20001 United States

Abstract:

Before-after safety study is a key procedure to justify the performance of remedial treatments. The empirical Bayes (EB) and full Bayes (FB) methods have been widely used to assist such safety assessment. However, the practical use of EB and FB methods are often limited because of the extensive data needs from additional reference sites to develop safety performance functions (SPF). To overcome this issue, this study proposes a new safety evaluation approach by expanding the theory of survival analysis. A parametric Bayesian survival analysis model with a random effect term to address the heterogeneity across sites is developed. As a case study, the proposed model is used to evaluate the safety effectiveness of the recent red light running photo enforcement program in the state of New Jersey. As demonstrated in the case study, the proposed approach requires no data from reference sites. Using individual crashes as the analysis units, the survival analysis explicitly accounts for time-dependent covariates and makes it feasible to include up-to- date crash data into models. In addition, the survival analysis, takes an appropriate consideration of censoring effects, has the ability to address the regression-to-the-mean issue, and can be used to capture the probability of crash occurrence beyond a specified time for study sites.

Supplemental Notes:

This paper was sponsored by TRB committee ANB20 Safety Data, Analysis and Evaluation.

Monograph Accession #:

01503729

Report/Paper Numbers:

14-1466

Language:

English

Corporate Authors:

Transportation Research Board

500 Fifth Street, NW
Washington, DC 20001 United States

Authors:

Xie, Kun
Ozbay, Kaan
Yang, Hong

Pagination:

14p

Publication Date:

2014

Conference:

Transportation Research Board 93rd Annual Meeting

Location: Washington DC
Date: 2014-1-12 to 2014-1-16
Sponsors: Transportation Research Board

Media Type:

Digital/other

Features:

Figures; References; Tables

Uncontrolled Terms:

Geographic Terms:

Subject Areas:

Highways; Planning and Forecasting; Safety and Human Factors; I72: Traffic and Transport Planning; I80: Accident Studies

Source Data:

Transportation Research Board Annual Meeting 2014 Paper #14-1466

Files:

TRIS, TRB, ATRI

Created Date:

Jan 27 2014 2:33PM