A review of the spatial accuracy of FARS coordinate locations in California

Background:In California, local law enforcement officials submit motor vehicle collision reports to the California Highway Patrol for entry into a central database known as the Statewide Integrated Traffic Records System (SWITRS). The fatal collision records in SWITRS are subsequently submitted to the National Highway Traffic Safety Administration for use in a national database known as the Fatality Analysis Reporting System (FARS). Since 2005, FARS has begun providing latitude and longitude coordinates for collision records. However, usually only the collisions' primary street location is included in the records, so the spatial accuracy of the geocoding can rarely be determined. The purpose of this paper is to evaluate the geocoding accuracy of FARS data.

Methods:FARS records were matched to a random selection of 100 fatal collisions from SWITRS using dates, times, street intersections and postmile information to provide necessary locational information for reviewing. Collision records were exported to Google Earth to measure the positional accuracy within 300 feet.

Results:81 collisions were correctly geocoded, 12 were incorrect and seven were unknown. Locations on non-state highways were correctly matched 86% of the time, while crashes on state highways matched only 76% correctly. Rural state routes had lower match success than more urbanized areas.

Conclusions:This research shows that there are discrepancies that should be considered when using the FARS coordinate locations and provides support for the incorporation of GPS technology to complement written location descriptions in the future.