Background: Traffic signals are often seen by the public and elected officials as a cure-all for operational and safety problems at intersections. Although signals have been used for many years, very little is actually known about their impact on safety. The accident experience warrant in the Manual on Uniform Traffic Control Devices (MUTCD) (one of 11 warrants that set minimum thresholds for considering installation of a traffic signal) is not well supported and does not consider accident severity. The MUTCD specifies that an engineering study "should indicate a traffic signal will improve the overall safety and/or operation of the intersection" before a signal is installed; however, there are no tools to help the traffic practitioner determine the likely impact on safety from installing a traffic signal. Past studies have yielded contradictory results and suffered from a number of serious deficiencies. Sometimes changes in traffic conditions can eliminate the need for an existing traffic signal. Practitioners need a way to analyze the safety impact of removing such a signal. This information can be used to alleviate the public concern that usually blocks signal removal.

Objective: The objectives of this project were to develop an improved accident warrant for traffic signals suitable for use in the Manual on Uniform Traffic Control Devices and to provide a model(s) to estimate the safety impacts of installing or removing traffic signals.

Status: The report has been published as NCHRP Report 491.

Tasks: Research includes the following tasks: (1) Review previous research evaluating accident experience at signalized intersections and those controlled by stop signs. Identify the issues and challenges in assessing the safety impacts of traffic signals and developing an accident warrant. Such challenges include accounting for unreported accidents, jurisdictional differences in accident reporting requirements and procedures, inaccuracy in reported accidents, definition of the types of accidents correctable by a traffic signal, high accident levels that do not accurately reflect the mean accident rate, and data-collection sample sizes needed for accurate estimation. Identify accident data sources that may be suitable for this project. (2) Considering the issues and challenges identified in Task 1, develop a data-collection and analysis plan suitable for the Tasks 5 and 6 development work. (3) Prepare an interim report that includes the information developed in Tasks 1 and 2 and a revised work plan and budget for the remaining tasks. (4) Execute the approved data-collection plan. (5) Develop a model(s) to estimate the number, severity, and types of accidents expected at signalized and stop-controlled intersections and the changes expected from installation or removal of a traffic signal. Estimates should include confidence limits. The model(s) should consider different intersection characteristics (e.g., approach and turning volumes, intersection geometry, proximity to other signalized intersections, control type, signal phasing, regional differences, and area type). (6) Using the Task 5 model, identify the conditions under which signal installation or removal is likely to improve or degrade safety and the types of accidents most affected. Recommend an improved accident warrant and related material suitable for inclusion in the MUTCD. The recommendations must be credible, definitive, and defensible. (7) Prepare a final report documenting the research results, presenting the Task 6 model and showing how it could be used with other analysis tools to assess the desirability of signal installation or removal. The recommended revisions to the MUTCD and justification for them should be included as an appendix.