TRB Pubsindex
Text Size:
x

« Prev Next »

Return to Results | New Search

Bookmark and Share

Title:

Crash Frequency Modeling Incorporating Time Trend with Panel Data

Accession Number:

01664158

Record Type:

Component

Abstract:

Standard crash models using longer time frames and aggregated data do not consider time effects on crash frequency which can result in biased regression parameter estimates due to unobserved heterogeneities, serial correlations, or time trends. An effective safety evaluation strategy is to develop predictive models that are capable of accounting for these different variations and properties which exist in crash data. In this study, the Negative Binomial distribution was used to model crash frequency where the temporal correlation was accounted for using a Generalized Estimating Equation (GEE) approach. Time was also included as an explanatory variable with linear and quadratic polynomials to investigate the crash trends across time. This analysis was based on 10 years of data for 174 four-legged signalized intersections in Wyoming. The study revealed that the Negative Binomial GEE (NB GEE) models outperformed traditional Negative Binomial (NB) and Random Effect Negative Binomial (RENB) models in terms of prediction capability. Quasilikelihood Information Criterion (QIC) statistics were used for assessing the choice of the working correlation structure in NB GEE models. This method also showed that NB GEE models incorporating “time” variables performed better than the models without time variables. Intersections with high traffic volume and large number of lanes in major and minor approaches, involvement of young drivers, and weather components came out to be significant contributing factors for high crash frequencies. Adding left-turn lanes at major approaches and presence of on-street parking contributed to reductions of all types of crashes. Intersection crashes also had an overall quadratic trend in time.

Supplemental Notes:

This paper was sponsored by TRB committee ANB25 Standing Committee on Highway Safety Performance.

Report/Paper Numbers:

18-05367

Language:

English

Authors:

Sharmin, Sadia
Ahmed, Mohamed
Wulff, Shaun S

Pagination:

4p

Publication Date:

2018

Conference:

Transportation Research Board 97th Annual Meeting

Location: Washington DC, United States
Date: 2018-1-7 to 2018-1-11
Sponsors: Transportation Research Board

Media Type:

Digital/other

Features:

References (8)

Geographic Terms:

Wyoming

Subject Areas:

Data and Information Technology; Highways; Safety and Human Factors

Source Data:

Transportation Research Board Annual Meeting 2018 Paper #18-05367

Files:

TRIS, TRB, ATRI

Created Date:

Jan 8 2018 11:21AM