Citation

Cunningham CM, Hummer JE. J. Transp. Saf. Secur. 2010; 2(4): 312-324.

Copyright

(Copyright © 2010, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439962.2010.508868

PMID

unavailable

Abstract

This research effort expanded on a previous research effort funded by the North Carolina Governor's Highway Safety Program. The original report, published in December 2004, looked into the safety effects of an automated red-light running (RLR) program in Raleigh, North Carolina. The primary shortcoming of the initial analysis was the limited data set used in the after period: only 5 months of data at 14 intersections. The initial findings indicated that RLR cameras decreased all five collision categories studied: total, RLR-related, angle, rear-end, and injury collisions. This research effort expanded on the data set used in the initial study by including 46 months of after data. The comparison group methodology was the primary analysis method used; however, the strength of this study versus others of its kind is in the selection of treatment and comparison sites that were done in a pseudorandom fashion. The authors show that the comparison sites chosen for this analysis should follow a very similar pattern to those in a controlled experiment because the city dispersed the sites across the entire city instead of choosing the highest angle collision sites. The dispersion of treatment sites amongst a pool of high-angle collision sites virtually eliminates the bias of installing red light cameras at the highest collision sites. Assuming any bias in selection of treatment sites is removed, the study should act like a controlled experiment, which would account for any of the possible historical, seasonality, time, and regression-to-the-mean phenomena that obscure most studies completed to date. This is important because no RLR enforcement program has completed a study where sites were chosen in such a manner. The findings from this follow-on study indicate that the enforcement program has successfully decreased angle- and injury-related crashed. There has been no significant change in total, RLR-related, or rear-end collisions. This was different from the original report of a very limited data set that showed significant decreases in four of the five collision categories. Last, a separate and less rigorous before?after study using causal factors was completed to emphasize the need for more rigorous studies of traffic control devices. This study is typical of a safety study conducted by most municipalities because it is easy to understand and requires less time and effort. The results from the causal factors study would not have changed the conclusions much; however, the emphasis would have likely been different. This is because the standard deviations were much smaller with the causal factor study, which would have led readers to think results for other collision groups significantly changed from the before period. This was actually not the case for many of the collision groups using the more rigorous comparison group study.