The study consisted of a major effort designed to determine the effects of altering speed limits on selected experimental sections, and a secondary effort to examine the effects that speed limit changes on the experimental sections have on nearby roadways.

The experimental method used was a before and after with comparison group design. The experimental or treatment sites were selected from a sample of sections where State and local jurisdictions planned to make speed limit changes as part of their routine activities. The comparison sites were selected by the research team after the experimental sites were identified.

The experimental sites selected for study were locations where State and local highway jurisdictions planned to change the speed limits as part of their routine engineering and traffic investigations. It is important to note that the speed limit changes were not made for the purposes of experimentation. Consequently, the researchers were not involved in determining the speed limit change, nor was a study of an individual State's method of setting speed limits undertaken. All speed limit changes were reported to the public in the routine method used by the local jurisdictions. As no special enforcement campaigns were initiated to enforce the speed limit at any of the study sites, it was assumed that normal enforcement was used throughout the study period.


Initially, 33 States agreed to participate in the study by notifying the researchers of sections where they planned to make speed limit changes. Using input from the participating jurisdictions, experimental sites were selected during the period of May 1986 through April 1987. Sites were selected based on the following criteria: Shown in figure 2 is a typical roadway in a small urban area where the speed limit was lowered from 35 to 25 mi/h (56 to 40 km/h). Depicted in figure 3 is a rural location where the speed limit was raised from 50 to 55 mi/h (81 to 89 km/h).

The criteria for selecting the comparison sites was based on matching as closely as possible the geometric, volume, and speed characteristics of the experimental sites. In some cases, one comparison site was matched to two similar experimental sites. The before data were collected at 127 experimental sites and 120 comparison sites.

Some sections were eliminated from the study because the planned speed limit alterations were not made, or road and/or utility construction was in progress. Several comparison sites were lost when speed limit changes and/or construction was encountered.


The speeds and headways of more than 1.6 million vehicles were collected during this study. Volume, speed, and headway data were collected for a 24-h period simultaneously at both the experimental and comparison sites prior to the speed limit change on the experimental section, and again after the change was made.

While all vehicle speeds were collected, only the speeds of free-flow vehicles, defined as vehicles having a headway of 4 s or more, were used in the analysis. At headways of more than 4 s, drivers can select their operating speed based on geometry, environmental conditions, and their own vehicle and driving ability instead of being influenced by other vehicles. Accordingly, if speed limits affect driving behavior, free-flow drivers are most likely to be influenced by speed limit changes. With very few exceptions, the roadway sections included in this study were not located in large urban areas with reoccurring congestion. An examination of the all-vehicle speeds vs. free-flow speeds indicated that free-flow speeds at the study sites were normally less than 2 mi/h (3 km/h) higher than all-vehicle speeds.

In most cases, the after data were collected in the same season and on the same day of the week as the before data. At selected sites, multiple measurements were taken to examine seasonal and other effects.

The volume, speed, and headway data were collected with Sarasota VC1900 automated roadside units, which classifies vehicle speeds in 1-mi/h (2-km/h) intervals from 1 to 128 mi/h (2 to 206 km/h). Headway data were collected In 1-s intervals from less than 2 s to more than 11 s. The data were collected for both directions of travel at a point representative of typical conditions on the section. Shown in figure 4 is a typical data collection setup showing the inductive loop mats in the roadway and the roadside units chained to a utility pole. Six inductive loop mats and two roadside units were deployed in the center of each lane at each site to collect speed and headway data. For speed data collection, the 3- by 6-ft (0.9- by 1.8-m) inductive loops were placed at a distance of 10 ft (3 m) from leading edge to leading edge.

Before and after data were collected at the same point to permit comparisons and eliminate locational differences. Date were not collected near major intersections, driveways, or other features that would affect normal driving speeds. Temporary inductive loop mats were used as sensors in order to differentiate speeds by vehicle length. Vehicles with a length less than 20 ft (5.1 m) were classified as short vehicles. Vehicles greater than 20 ft (6.1 m) in length were classified as short vehicles. Every attempt was made to conceal the roadside units and make the sensors as inconspicuous as possible.

The date were extracted with the use of a laptop computer and stored on diskettes. A computer program, written especially for the study, was used to reduce the speed and headway data for analysis. The two-person data collection team also recorded geometric and roadway features data such as number of lanes, lane width, number of public streets, and number of commercial and residential driveways.

The accident data base for this study contains over 20,000 reported accidents. For most sections, accident data were collected for 3 yr before the speed limit was changed and for a 2-yr after period. The participating highway agencies furnished the accident data in a variety of formats, including copies of accident reports, computer summaries, and individual accident listings. The data were encoded using dBASE IV, then recorded in a compatible format for analysis. The accident variables coded include severity, number of persons injured and killed, crash type, day, date, hour, number of vehicles, lights and surface conditions, and estimated vehicle speeds (when speeds were recorded on the accident report)

Next: Site Characteristics | Prev: Objective and Scope | Contents