blank space

South Dakota Department of Transportation
Project Synopsis
SD1998-12


Title: Identification of Abnormal Accident Patterns at Intersections
Project Researcher: Mary Storsteen, SDDOT
Project Manager: Hal Rumpca
Research Period: 5/10/1999 - 8/19/1999
Status:
Cost: $12,000.00

Problem Statement:An expected value analysis table, which is a scientifically based method for identifying abnormal accident patterns, can be used to determine if an intersection has a higher than normal number of a particular type of accident. The method requires the use of a statistical test to determine whether an accident type is more frequent or severe at a study location than at other similar locations. Presently South Dakota traffic engineers use accident rates which are calculated from known accident numbers and traffic volumes. Since site specific traffic volumes are not always readily available for all intersections, accident rates cannot be easily determined. In addition, there is often a need to compare accidents within a given range of traffic volumes and intersection types to determine if abnormal accident patterns are occurring.

South Dakota needs to develop expected value analysis tables from accident data collected at random intersections located throughout the state. The values provide a standard for identifying abnormal accident characteristics in consideration of local roadways, drivers, and environmental conditions. The development and use of this statistical method could improve identification of serious safety problems at any intersection in South Dakota and could be used to determine if an intersection should be considered hazardous or not hazardous with relation to numbers of a particular type of accident.

South Dakota is a rural state with the majority of its rural highways having average daily traffic volumes of less than 3,000 vehicles per day. Eleven Class 1 Cities have populations ranging from 5,000 to 25,000. Two Class 1 Cities have populations of 55,000 and 110,000 respectively. A statistically valid sampling method needs to be developed to obtain intersection locations, characteristics, and other information necessary to create the expected value analysis tables for public roadways in South Dakota.



Findings: The expected value analysis table is a scientifically based statistical method for determining if an intersection has a higher than normal number of a particular type of accident. Fourteen (14) tables were created for the types of intersections requested by the project’s Technical Panel. A sampling method was devised to find information on thirty (30) intersections per type or fifteen (15) intersections per type depending on the total number of intersections in each category. The rural, four lane, one way stop, three leg intersections; three leg, signalized intersections in Rapid City; four leg, signalized intersections with a volume below 15,000 in Rapid City; and rural, four way stop, four leg intersections did not require a sampling method since the total number of intersections was nine (9), twenty-two (22), seven (7), and seven (7) respectively. One example of an intersection that did need a sample taken was urban, four leg, signalized intersections with a volume below 15,000 for the state of South Dakota. There were a total of 137 intersections for this category; and a sample of thirty (30) intersections was taken. The mean and 90th and 95th percentile were calculated and five (5) more intersections were then added and the calculations were computed again. The difference in the numbers was very insignificant. This shows that the sampling technique used to create the expected value analysis table appears to be a valid and reliable method. The expected value analysis tables will be very useful in determining if an intersection has an abnormally high number of severe or fatal accidents. The tables will be used in assisting the South Dakota Department of Transportation (SDDOT) in identifying serious safety problems at intersections.

Research Objectives:
1  To develop an efficient means of evaluating intersections which may require safety improvements.
2  Develop expected value analysis tables based on South Dakota information.

Research Tasks:
1  Review and summarize literature and contact a representative number of State DOT’s and local agencies to identify what data, procedures, and sampling methods are being used to develop expected value analysis tables.
2  Meet with the project’s Technical Panel to review the project scope and work plan.
3  Obtain accident, traffic volume, and other data from the SDDOT, local units of government, and sampling methods, which are necessary to create the expected value analysis tables for the most current five year period.
4  Prepare the expected value analysis tables for each of the accident and location types approved by the Technical Panel.
5  Compare the expected value analysis tables developed for South Dakota to a representative sample of similar tables developed for other states and determine the applicability of using this method on intersections in South Dakota.
6  Prepare a final report and executive summary of the literature review, findings and conclusions, including expected value analysis tables for the State of South Dakota.
7  Make an executive presentation to the Research Review Board at the conclusion of the project.

Documents Available:
SD1998_12_final_report.pdf
blank space