Citation

Washington SP, Metarko J, Fomunung I, Ross R, Julian F, Moran E. Accid. Anal. Prev. 1999; 31(1-2): 135-146.

Affiliation

School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta 30332-0355, USA. simon. washington@ce.gatech.edu

Copyright

(Copyright © 1999, Elsevier Publishing)

DOI

unavailable

PMID

10084628

Abstract

The southeastern United States, particularly those states representing the National Highway Traffic Safety Administration's and Federal Highway Administration's Region IV, have consistently had among the highest number of fatal crashes and fatal crash rates compared to the other five regions in the US. These states--Alabama, Florida, Georgia, Kentucky, Mississippi, North Carolina, South Carolina, and Tennessee, are suspected of sharing traits in common that lead to their systematically poor crash record. Inter-regional and intra-regional comparisons, such as the comparison between southeastern and non-southeastern states, raises interesting and challenging research questions that are applicable to similar comparisons. First, is there a sound practical and theoretical justification supporting an inter-regional comparison framework? Second, is there a way to construct a meaningful statistical hypothesis and test to determine whether one region, which is comprised of numerous entities, suffers from a characteristically better or worse crash record? This paper addresses each of these questions. After providing a brief summary of the Southeast's safety record, we discuss the issues pro and con surrounding inter-regional comparisons, illustrating the statistical strategy for such an analysis approach. Then, a simple generalizable statistical procedure is used for testing the hypothesis that southeastern states have a poorer crash record than non-southeastern states. Finally, we explore possible relationships between safety belt use, roadway functional class, vehicle miles of travel, and driver age on fatal crash occurrence. The data used in the analyses are from the Fatal Analysis Reporting System (FARS), primarily data compiled for 1995. The analysis suggests that regional differences in fatal crashes may indeed exist, that these differences are related in part to seat-belt use, VMT by functional classification, and speed limit differences, and that more detailed studies are needed to quantify the effect of these and other factors. The approach developed here lends insight as to where future in-depth studies may reveal causal factors of fatal crashes, and illustrates the relative safety performance records of US regions.