Citation

Lee C, Abdel-Aty MA. J. Transp. Saf. Secur. 2009; 1(2): 121-134.

Copyright

(Copyright © 2009, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439960902735329

PMID

unavailable

Abstract

Crashes that occur on freeway ramps and in the vicinity of ramps were analyzed using a variety of statistical methods. The data used in this study are the 6 year records of crashes that occurred on 98 ramps of a 25 mile section of Interstate-4 freeway in the Orlando, Florida, metro area. Log linear models were used to identify the relationship among the frequency of ramp crashes, ramp geometry, and ramp traffic volume. Binary logit models and bivariate probit models were developed to observe the propensity of specific type of ramp crashes in terms of curvature of ramps, number of vehicles involved in crashes, and collision type. The binary logit model was also used to identify the factors correlated with injury severity of drivers involved in ramp crashes. The results of model estimates showed that crash characteristics differ by the location of crashes with respect to ramps: ramp related and ramp intersection related crashes. It was found that the presence of advisory speed signs potentially reduces the likelihood of the occurrence of crashes on off ramps and rear-end crashes on ramps. Finally, it was found that female drivers, single vehicle crashes, and shorter ramp length are correlated with higher probability of fatal and severe crashes on ramps. The findings suggest some important factors closely related to ramp crash risk and hidden relationships that have not been identified in the past.

Keywords: ramp; freeway crash; crash frequency; injury severity; log linear model; bivariate probit model