Safety Evaluation of Red-Light Indicator Lights (RLILs) At Intersections

Chapter 5. Data Collection

FDOT provided the majority of the data for this strategy, including treatment sites, information regarding the enforcement lights (e.g., manufacturer specifications and law enforcement practices), screening for construction activity for many treatment sites, and access to its mainframe computer. This allowed the research team to query crash, roadway, and traffic data. Several city traffic engineering departments and police departments also provided information. Additional details about the design, installation, and maintenance of RLILs, as well as lessons learned, are provided in the appendix of this report.

Installation Data

The FDOT central office undertook an initiative to deploy enforcement lights at signalized intersections. It provided the FDOT districts with enforcement light equipment (e.g., mounts and bulbs) and lists of intersections for potential installations. Districts 1, 2, and 5 were able to implement this treatment widely. The engineering departments of these districts provided the following data for this evaluation:

• Treatment locations.
• Installation dates.
• Movements monitored (e.g., northbound through, westbound left).
• Enforcement light type (i.e., white incandescent or blue LED).

After receiving lists of all the potential treatment sites from the districts, the research team selected treatment sites appropriate for this study. Initially, there were more than 300 potential treatment sites with installation dates varying from 2003 to 2010. Only those treatment sites with an installation date on or after 2004 allowed retrieval of before-period data from the available data in Florida’s CAR system. This excluded a number of potential sites from the study. The CAR did not contain complete crash data for sites located on other route types, thus excluding sites not located on U.S., State, or county routes. The research team identified the number of approaches using aerial imagery of the potential treatment sites and observed that only a small proportion were three‐legged intersections. Such a small sample size would not likely produce statistically robust results; consequently, the research team removed the three‐legged treatment intersections from further consideration in this study. The final treatment group for this strategy was composed of 108 four‐legged intersections located in three districts in Florida (districts 1, 2, and 5).

The study also solicited information on law enforcement practices and public awareness campaigns regarding the enforcement lights. A number of local agencies provided feedback regarding the enforcement practices and public awareness, particularly in district 1. The appendix of this report provides the responses by city and county agencies tasked with red-light enforcement.

Reference Sites

As noted in the study design, the potential for spillover or crash migration effects existed for this strategy. To address this potential issue, the research team formed two reference groups. The first group, a potential spillover reference group, consisted of signalized four‐legged intersections immediately adjacent to a treatment site. In other words, a selected spillover reference site would not have a signalized intersection between it and a treatment site. The second group, a non‐spillover reference group, consisted of signalized four‐legged intersections that had one or more untreated signalized intersections separating them from a treatment site. If the research team detected no crash migration or spillover effects, then it pooled the two reference groups to form the reference group for the EB method. If crashes increased or decreased significantly at the spillover sites, then these sites would not help to identify the safety effectiveness of the strategy at treated sites reliably. This would also indicate effects of the strategy on adjacent intersections. In all, the research team selected 30 reference sites. Sixteen were spillover reference sites, and the remaining 14 were non-spillover sites. For each reference site, the research team queried all available crash records between January 1, 2003, and December 31, 2012.

Roadway Data

Using crash report data, FDOT provided roadway data for the installation and reference sites. Selected roadway data elements included the following:

• District.
• County.
• Shoulder type.
• Functional classification (which also defined area type).
• Surface width.
• Median width.
• On-curve presence.
• Posted speed limit.

Crash Data

FDOT’s crash database, the CAR system, provided the crash data for the treatment and reference sites using a two-stage process. First, the project team determined the node number for each site. This number was a unique number that FDOT assigned to each intersection within a county for identification purposes. The crash database also used these node numbers for location identification. After compiling the node numbers, the second stage was to query crash records. The research team queried records using the “around a node” option within the CAR system and a defined influence area of a 250-ft radius around each intersection.

Traffic Data

The crash reports that the CAR system generated also contained traffic volume data. Most important was the average annual daily traffic (AADT) values for the mainline and cross-street roadway sections for the intersections. In addition, the 2012 Florida Transportation Information (FTI) DVD provided traffic volumes for State roadways where no crashes occurred.⁽⁷⁾ Finally, the research team estimated traffic volumes for cross streets with missing data based on nearby traffic counts, surrounding land use, roadside development, and interconnectivity.

Treatment Cost Data

FDOT provided estimates of the costs and service lives of the treatments for use in conducting a B/C analysis of the treatment. Red-signal enforcement light sets (including housing) cost approximately $50 to $150 for traditional (rather than LED) bulbs. Generally, four to eight indicator sets were needed for a major intersection to cover all signals and intervals for a total intersection cost of $200 to $1,200.Cost depended substantially on whether LED or traditional bulbs were used; LED bulbs cost approximately three times as much initially but required much less power. The power cost was generally negligible in comparison with the power used by the signal, into which the indicator was directly wired. Therefore, the analysis ignored the power cost.

In addition, installation cost and service life data were explored through vendors,^[2] State transportation departments, and newspaper articles. (See references 1 and 8–12.) The cost of lights was dependent on color and hardware required for installation. The research team found that the lights cost between $77 and $300 per light. In addition, the installation of the indicator lights required manpower and equipment, and the associated cost depended on where the lights were mounted (i.e., on the back of signal head or on a pole). The cost per intersection ranged from $1,000 to $3,000 for installation. The State transportation departments reported the life span of the indicator lights was between 5 and 10 years.

Data Characteristics and Summary

Table 5 defines the crash types used by Florida. Table 6 provides summary information for the data collected for the treatment and reference sites. Installations dates ranged from 2004 to 2010. The before and after periods varied by location with the installation year marking the change from before to after. Before periods started as early as 2003 and ended as late as 2009. After periods started as early as 2005 and ended in 2012. The information in table 6 should not be used to make simple before–after comparisons of crashes per site-year because it does not account for factors other than the strategy that might cause a change in safety between the before and after periods. Such comparisons require an EB analysis, as presented in chapter 7.

MV = Moving vehicle.

Note: Crash rates are presented as crashes/site/year.
Avg. = Average.
Min. = Minimum.
Max = Maximum.

^[2] Melvin Barrios of Industrial Traffic Solutions, e-mail correspondence, December 3, 2014, and http://lanecontrols.com/tattle-tale/.