TRB Pubsindex
Text Size:
x

« Prev Next »

Return to Results | New Search

Bookmark and Share

Title:

The Effect of Model Specification on the Results of Pedestrian Injury Analysis

Accession Number:

01556990

Record Type:

Component

Availability:

Transportation Research Board Business Office

500 Fifth Street, NW
Washington, DC 20001 United States

Abstract:

Analytical models have been widely used in the field of pedestrian safety analysis. They are usually developed to measure the effectiveness of different safety improvement strategies and play a pivotal role in policy making process. Inappropriate model specifications affect the results of models, introducing inaccuracy, and bias in them. Since decisions and policies which are made based on inaccurate and biased results might lead to worsened safety issues, it is really important to correctly specify the models. In this study, several specifications of discrete-choice models are compared to find out the extent to which change in model specification can affect the results of pedestrian injury severity analysis in motor-vehicle crashes. Models are then evaluated based on their advantages and limitations to find the most appropriate specification. Police reported crash data between 2008 and 2011 from the State of North Carolina are used to develop various models. Estimates of applied models suggest several factors including pedestrian age, alcohol consumption, speed violation, lightening, and vehicle type affect pedestrian injury severity in accidents. A comparison of models indicates that the magnitude of effect of exogenous variables on injury severity is significantly different across diverse models. The differences prove the significant effect of model specification on the results of analysis. Evaluation of models suggests that a three-level nested logit analytical framework is the most appropriate model among developed models to be used in pedestrian injury severity analysis.

Supplemental Notes:

This paper was sponsored by TRB committee ANF10 Pedestrians. Alternate title: Effect of Model Specification on Results of Pedestrian Injury Analysis.

Monograph Accession #:

01550057

Report/Paper Numbers:

15-5352

Language:

English

Corporate Authors:

Transportation Research Board

500 Fifth Street, NW
Washington, DC 20001 United States

Authors:

Poorfakhraei, Amir
Samimi, Amir
Ermagun, Alireza

Pagination:

21p

Publication Date:

2015

Conference:

Transportation Research Board 94th Annual Meeting

Location: Washington DC, United States
Date: 2015-1-11 to 2015-1-15
Sponsors: Transportation Research Board

Media Type:

Digital/other

Features:

Figures; References; Tables

Geographic Terms:

North Carolina

Subject Areas:

Data and Information Technology; Pedestrians and Bicyclists; Policy; Safety and Human Factors; I84: Personal Injuries

Source Data:

Transportation Research Board Annual Meeting 2015 Paper #15-5352

Files:

TRIS, TRB, ATRI

Created Date:

Dec 30 2014 1:48PM