U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-RD-02-089
Date: July 2002
Safety Effectiveness of Intersection Left- and Right-Turn Lanes
PDF Version (1.48 MB)
PDF files can be viewed with the Acrobat® Reader®
4. DATA COLLECTION
This section of the report documents the data collection performed for the intersection sites selected for the safety evaluation of left- and right-turn lanes. The types of data collection addressed includes geometric design and traffic control data, traffic volume data, and traffic accident data. Each type of data is addressed below.
Data were collected on the geometric design and traffic control features of each improved, comparison, and reference site. Nearly all of the study sites were visited in the field by a research team member to obtain geometric design and traffic control data. In addition, geometric design and traffic control data were obtained from the following sources, whenever available:
These sources were also useful in documenting what specific geometric changes were made as part of a project.
The field visits provided a key opportunity to observe the characteristics of each site and record data of interest. Some intersections were visited twice, once during the selection of improved sites and once during the data collection activities. Time spent in the field in each state was also used to identify or review candidate comparison and reference sites.
The field activities involved visits to both highway agency offices and field sites, and had multiple purposes including:
The vast majority of field visits to improved sites were made after completion of project construction. This provided an opportunity to verify in the field that the project had, in fact, been constructed and that its geometrics in the period after construction matched the data provided in the office. The geometrics before construction were often evident in the field, due to differences in pavement surfaces, but were also documented from office records. Relying on both office and field data, a record was made of the geometric design and traffic control changes made as part of the improvement project (see appendix D).
Geometric Design and Traffic Control Variables
Geometric design and traffic control data were collected for each study intersection. For each individual intersection approach, the geometric design and traffic control variables obtained were:
For the intersection as a whole, variables obtained were:
The set of geometric design and traffic control variables obtained was purposely broader than needed for the planned analyses so that issues beyond those planned could be addressed, as needed. It was never envisioned that all of these variables could, or should, be related to traffic accidents, but they were obtained to assure that the documentation of each study intersection was very complete. Appendix D provides definitions of the measurement methods and codes used for each of these geometric design and traffic control variables.
Traffic volume data were obtained for each study intersection. The desirable traffic volume data set for any study intersection included:
It was found, as a practical matter, that the participating states nearly always had ADT data on file for the major-road in the vicinity of each intersection. Minor-road ADT data were often, but not always, available for the improved sites; ADT data were likely to be available for the improved sites because there had often been a traffic count made at the intersection as part of the design of the project. Minor-road ADT data for comparison and reference sites were available for virtually every intersection of potential interest in some states and only for a very limited number of intersections in other states.
Intersection turning movement counts were of direct interest to the study. In evaluating the safety effectiveness of intersection left- and right-turn lanes, it would be valuable to know the volume of vehicles turning left or right and using the turn lanes of interest. However, turning movement volumes were not available for most of the intersections. In particular, turning movement volumes were only available for less than 10 percent of the improved sites, and an even smaller percentage of the comparison and reference sites. Therefore, as a practical matter, it was not feasible to use intersection turning volumes in the safety evaluation because the sample size for any given type of project would have been substantially reduced.
It was decided that, for a intersection to be used in the evaluation, ADT data should be available for both the major- and minor-road legs of the intersection for at least one year during the study period. If this minimal traffic volume data set was not available, any improved, comparison, or reference site was dropped from the study. For most intersections, major-road ADT data for several years and minor-road ADT data for at least one year were available. These ADT data came from many sources in the participating highway agencies, including state ADT maps and logbooks, county and city ADT maps, traffic volume data bases and manual files, and, in some cases, traffic counts made specifically for this evaluation. However, as stated above, no intersection was used unless major- and minor-road ADT data were available for at least one year.
Some of the analyses performed required separate estimates of intersection ADTs for each year of the study period. These estimates for each individual year were obtained by interpolation and extrapolation from the ADT data obtained from the participating states. All extrapolations were checked very carefully to assure that the rates of ADT growth or decline were reasonable for the site conditions and consistent with ADT growth or decline patterns at nearby sites. Where ADT data were available for only one year, extrapolations to earlier and later years were made using the following data sources for guidance:
Table 17 presents the distribution of ADTs for the improved and comparison/reference sites, including mean, minimum, and maximum values of the ADT for the year 1999, and annualized percentage growth rates in ADT over the period from 1988 to 1999, for major-road ADT, minor-road ADT, and total ADT entering the intersection. Where the ADTs at an intersection differ between the two major-road approaches or the two minor-road approaches, Table 17 is based on the larger of the two major- or minor-road ADT values; for this reason, the mean total entering ADT is not necessarily equal to the sum of the mean major- and minor-road ADTs. Table 18 presents comparable data for the matched improved and comparison sites. The 260 matched improved sites shown in Table 18 are a subset of the 280 total improved sites shown in table 17.
Traffic accident data for the study intersections were obtained from the computerized accidents records of the participating state highway agencies. In some cases, the computerized data were supplemented with collision diagrams prepared by manual or computer means.
Accident data were obtained for all study intersections in each state for a period of 9 to 13 years. Table 19 shows the specific time periods for which data were available in each state. In most states, the study period began with the calendar year 1988. However, because of limitations on data availability and changes in data formats, data for Minnesota and Virginia were obtained for a period beginning in 1990 and data for North Carolina for a period beginning in 1991. The final year of the study period was 1999 for all states except one; in Oregon, the study was extended to include data for the year 2000 because both accident and ADT data for that period were available.
Data were requested from each state for all accidents during the study period that occurred on any intersection leg within 300 meters (1,000 feet) of each intersection. The 300-meter (1,000-foot) distance was not selected because accidents that far from the intersection are necessarily related to the intersection, but simply to assure that all accidents of potential interest were available and that no request for supplementary data would need to be available.
After evaluation of the available data, a criterion for identifying intersection-related accidents of interest to the evaluation was established. Intersection-related accidents were selected from the available data including accidents assigned mileposts within 75 meters (250 feet) of the study intersection, and had were designated by the investigating officer or accident data coder that they were related to the operation of the intersection. Where closely spaced intersections were present, the 75-meter (250 foot) boundary was decreased to a point half the distance to the adjacent intersection. Accidents indicated as being non-intersection-related or driveway-related were excluded from the evaluation. Table 19 includes data only for accidents that meet this definition of being related to the intersection.
The one exception to this procedure described above was in accident data from Illinois. Illinois codes all intersection-related accidents to the milepost of the intersection. Therefore, the milepost cannot be used to distinguish the distance of a collision from the intersection in question. In Illinois data, all accidents assigned to the intersection milepost are presumed to be related to the operation of the intersection and were included in the analyses.
In some states, the accident location milepost or reference point assigned to an intersection may change from year to year. These changes were accounted for so that a consistent set of accident data from year to year were extracted from the available accident data.
The accident data elements obtained from each state varied; in most cases, the accident data provided by the state included more accident descriptors than were needed for the study. Both accident-level and vehicle-level accident descriptors were obtained. The variables that were actually used in preliminary investigations and in the safety evaluation itself were:
The dates for which accident data were obtained are shown in Table 19. The table shows that the periods for which accident data were available varied among the states. In each state, the accident data period extends back to 1988, whenever possible; where a later date is shown for the beginning of the accident data period, data before that date were unavailable. Study periods before and after improvement of each treated site were determined based on criteria described in Section 5 of this report.
Table 19 documents the magnitude of the available accident data base. The table shows that there were a total of 26,056 intersection-related accidents during the study period for all 580 intersections combined. Approximately 49 percent of the accidents occurred in Illinois which, as documented above, included approximately 45 percent of the study intersections. The table also shows the distribution of accident severity, by state and overall; accidents involving fatalities ranged from 0.3 to 0.9 percent of all accidents, and accidents involving non-fatal injuries ranged from 35.5 to 50.9 percent of all accidents.
Table 20 compares the total intersection accident experience, exposure, and accident rate per million entering vehicles, for periods before and after the improvement projects, for the 260 matched improved and comparison sites at rural intersections. Table 21 presents comparable data for urban intersections.
MEV = million entering vehicles
MEV = million entering vehicles