An Objective Warrant System for Red Light Cameras
Author(s): Hildebrand, McBride
Slidedeck Presentation:
Abstract:
Background/Context: Photo-enforced traffic signals, or red light cameras (RLC), have been shown to be an effective countermeasure to address red light running (RLR). Unfortunately, RLCs are often installed at specific intersections in reaction to public pressure or based on recommendations from traffic engineers or enforcement officers without the benefit of rigorous analyses. There is no existing analytical methodology that can objectively determine where RLCs would prove to be most effective. As more jurisdictions modify legislation to allow the enforcement of RLCs, road authorities require an unbiased warrant system that can be used to identify appropriate treatment sites.
Aims/Objectives: The overarching goal of this research was to develop a concise and objective analytical methodology to determine if an intersection is an appropriate candidate for RLCs. This research evaluated the strength of relationships between the characteristics of local intersections and the frequency of RLR (dependent variable). A forecasting model was developed to allow the prediction of RLR frequencies at different intersections.
Methods/Targets: There were four main areas of focus involved in this research: selection of study intersections, collision data analyses, RLR field observations, and the development of an RLC warrant system that considers the cost-effectiveness of the treatment. A sample of 38 signalized intersections were selected in New Brunswick for inclusion in this study. In an effort to avoid any bias in the resulting warrant system, the study locations were not selected on the basis of any preconceived assumption that they might have a potential for high rates of RLR. In order to accurately capture the full extent of RLR at the study sites, a video camera was installed at each location for two-hour periods and the data analyzed in detail post-recording. During this study, a total of 22 independent variables were collected/computed for consideration in the RLR predictive model. A negative binomial model was the model format eventually chosen due to the over dispersion of the RLR count data. Collision data analyses were then completed to evaluate the existing performance of each of the sampled intersections in order that a relationship between RLR and preventable collision configurations could be developed.
Results/Activities: The resulting methodology combines the predictive model that forecasts RLR incidence, collision modification estimates associated with RLC installations, and a cost-effectiveness evaluation. Consequently, a network of intersections can be evaluated to determine where RLC installation is appropriate and relative priorities.
Discussion/Deliverables: The methodology is somewhat analogous to the development and adoption of traffic signal warrant systems that allowed road authorities to objectively decide which intersections should be signalized. The approach developed by this study is a first significant step toward the establishment of a technical process to decide whether RLC is appropriate for specific sites. By limiting their use to only those sites that stand to benefit the most, their judicial use should result in more consistent practice and overall safety improvement.
Conclusions: A robust methodology was developed to permit road authorities to make objective decisions about the installation of RLCs.