01100673

Component

https://doi.org/10.3141/2061-06

http://scholar.google.com...Lord&publication_year=2008

In estimating safety performance, the most common probabilistic structures of the popular statistical models used by transportation safety analysts for modeling motor vehicle crashes are the traditional Poisson and Poisson–gamma (or negative binomial) distributions. Because crash data often exhibit overdispersion, Poisson–gamma models are usually the preferred model. The dispersion parameter of Poisson–gamma models had been assumed to be fixed, but recent research in highway safety has shown that the parameter can potentially be dependent on the covariates, especially for flow-only models. Given that the dispersion parameter is a key variable for computing confidence intervals, there is reason to believe that a varying dispersion parameter could affect the computation of confidence intervals compared with confidence intervals produced from Poisson–gamma models with a fixed dispersion parameter. This study evaluates whether the varying dispersion parameter affects the computation of the confidence intervals for the gamma mean (m) and predicted response (y) on sites that have not been used for estimating the predictive model. To accomplish that objective, predictive models with fixed and varying dispersion parameters were estimated by using data collected in California at 537 three-leg rural unsignalized intersections. The study shows that models developed with a varying dispersion parameter greatly influence the confidence intervals of the gamma mean and predictive response. More specifically, models with a varying dispersion parameter usually produce smaller confidence intervals, and hence more precise estimates, than models with a fixed dispersion parameter, both for the gamma mean and for the predicted response. Therefore, it is recommended to develop models with a varying dispersion whenever possible, especially if they are used for screening purposes.

01114754

English

pp 46-54

2008

Transportation Research Record: Journal of the Transportation Research Board

9780309113311

Print

Figures (7) ; References (29) ; Tables (4)

Confidence intervals; Crashes; Dispersion (Statistics); Mathematical models; Mathematical prediction; Poisson distributions; Regression analysis; Rural areas; Three leg intersections; Unsignalized intersections

Negative binomial distributions

California

Data and Information Technology; Highways; Safety and Human Factors; I81: Accident Statistics

TRIS, TRB, ATRI

Jan 29 2008 3:54PM

• Adjustment for Maximum Likelihood Estimate of Negative Binomial Dispersion Parameter
• Analysis of Naturalistic Driving Data: Prospective View on Methodological Paradigms
• Analysis of Road Crash Frequency with Spatial Models
• Crash Frequency Analysis with Generalized Additive Models
• Cross-Correlation Analysis and Multivariate Prediction of Spatial Time Series of Freeway Traffic Speeds
• Nonparametric Bayesian Estimation of Freeway Capacity Distribution from Censored Observations
• Variable Selection Issues in Tree-Based Regression Models