Undercounting of large trucks in federal and state crash databases: Extent of problem and how to improve accuracy of truck classifications

ABSTRACT

Objective: Prior research suggested that single-unit trucks are undercounted when using vehicle body codes in the Fatality Analysis Reporting System (FARS). This study explored the extent of the misclassification and undercounting problem for crashes in FARS and state crash databases.

Methods: Truck misclassifications for fatal crashes were explored by comparing the Trucks Involved in Fatal Accidents (TIFA) database with FARS. TIFA used vehicle identification numbers (VINs) and survey information to classify large trucks. This study used VINs to improve the accuracy of large truck classifications in state crash databases from 5 states (Delaware, Maryland, Minnesota, Nebraska, and Utah).

Results: The vehicle body type codes resulted in a 19% undercount of single-unit trucks in FARS and a 23% undercount of single-unit trucks in state databases. Tractor-trailers were misclassified less often. Misclassifications occurred most frequently among single-unit trucks in the weight classes of 10,001–14,000 pounds.

Conclusions: The amount of misclassification of large trucks is large enough to potentially affect federal and state decisions on traffic safety. Using information from VINs results in more complete and accurate counts of large trucks involved in crashes. The National Transportation Safety Board recommended actions to improve federal and state crash data.

KEYWORDS:

• large trucks
• motor vehicle crashes
• injury
• database linkage
• misclassification
• fatality

Acknowledgments

We thank Dr. Robert Dodd, James Ritter, Dr. Joseph Kolly, and Barbara Czech of the NTSB for reviewing and commenting on the research. We are grateful to the following people and organizations who made this study possible. John Kindelberger of NHTSA collaborated with the NTSB by requesting state CODES data, making contractual arrangements between NHTSA and the NTSB, supplying gross vehicle weight ratings as derived from vehicle identification numbers, and providing extensive technical advice and assistance. Each participating state's CODES staff imputed linkages between hospital discharge records and police accident reports, compiled their state databases according to the authors' specifications, and provided advice on their databases. The CODES staff included Laurie Lin of the Delaware Department of Health and Social Services, Shiu Ho and Cynthia Burch of the National Study Center for Trauma and Emergency Medical Systems at the University of Maryland, Anna Gaichas of the Minnesota Department of Health, Dr. Ying Zhang of the Nebraska Department of Health and Human Services, and Andrea Thomas and Dr. Lawrence Cook of the University of Utah CODES Project. In addition, Andrea Thomas and Dr. Cook of the University of Utah worked closely with the authors in order to produce imputed missing values for each state and provided considerable statistical consultation concerning the correct methods for analyzing imputed data. Shiu Ho and Cynthia Burch at the University of Maryland provided centralized injury severity coding using ICDMAP-90 software and helped coordinate the work done by the states.

Funding

This research was funded by the NTSB as part of a larger safety study on single-unit trucks.