PROBLEM IDENTIFICATION

The Traffic Records Division of the Indiana State Police currently utilizes a Pseudo Code system to locate crashes as part of the crash data entry process. The current system has significant problems. For example, the same physical road can have multiple pseudo codes, making it impossible to determine locations that have frequent crashes. There are crash sites for which pseudo codes have not been identified, making location of that particular crash impossible to determine for data evaluation purposes. The availability of future pseudo codes is limited and cannot meet the growth rate of Indiana roads.

GOALS AND OBJECTIVES

The goal of the assessment was to conduct a test using a suitable sub-sample that represented the typical characteristics of Indiana crashes in order to:

• determine and evaluate the accuracy of the existing Pseudo Code process versus the Coordinate-Based Geographic Information System (GIS);

• determine and evaluate data entry costs and efficiency of workflow for both systems; and

• identify potential issues and evaluate alternatives associated with converting from the Pseudo Code process to a GIS system.

STRATEGIES AND ACTIVITIES

Seven counties were selected for the pilot project based upon the availability of sufficiently detailed street maps and aerial photographs. Approximately 9000 crashes were located using both the GIS and Pseudo Codes in order to make comparisons between the two systems.

Purdue University – Center for the Advancement of Transportation Safety conducted the assessment that included the following activities.

• Developed a set of representative Indiana counties with roadway systems that had been mapped, photographed, and was available in the Esri-ArcView (GIS) format. Determined an estimated necessary sample size of two percent of annual number of Indiana crashes. (Note: When project was completed, the actual sample size represented four and one half (4 ½ percent).

• Purchased the necessary hardware and identified workstations for two temporary employees who would work on the project for approximately 90 days.

• Created a separate database for the GIS data, using the existing unique crash identifier to link the existing crash record data file with the GIS file.

• At conclusion of data entry process, linked the two files and completed analysis.

• Documented data entry time for both processes.

RESULTS

A sample of four and a half (4 ½) percent of the crashes from seven representative Indiana counties was evaluated using the existing pseudo code crash identification system against the Esri-ArcView coordinate system. On the basis of the crashes used in the test, the overall rate of error with the existing pseudo code system was estimated to be four to six (4-6) percent compared to the GIS location system. The assessment also demonstrated the following about the pseudo code system.

• It exposes Indiana to under-counting crashes occurring at the same site, thereby possibly leading to funds (capital and enforcement) being misdirected.

• The system reveals one to two (1- 2) percent missing data.

• It demonstrates that approximately two to three (2-3) percent of crashes that occurred at the same location had different pseudo codes assigned.

• The system requires significantly more actual data entry time and exposed the State to additional unnecessary data entry errors.

   WINTER 2002