2017 Research Project

Completing the Picture of Traffic Injuries: Understanding Data Needs and Opportunities for Road Safety

Principal Investigator
Chris Cherry
University of Tennessee, Knoxville
View Bio

Eric Dumbaugh
Florida Atlantic University
View Bio
David R. Ragland
University of California, Berkeley
View Bio
Laura Sandt
University of North Carolina, Chapel Hill
View Bio


Police-recorded crash data has improved over time, but still fails to report all aspects of crashes that are important to developing a full understanding of crash mechanism, injury burden, pre-crash conditions, and ultimately total health and cost outcomes. Traditionally, safety and injury analysis has occurred in siloed fields, with road safety researchers relying predominately on police-recorded crash reports, and public health researchers relying on hospitalization records. Depending on the context of the study and the database used, findings vary. This is the case for the micro-level (e.g., injury severity of an individual) to the macro-level (e.g., injury rate) scale.

This work will begin to map disparate data sets to inform questions surrounding crashes. The data-mapping process will aim to build linkages between police-crash datasets and other datasets (i.e., incident-oriented data, spatial data, emerging datasets) and scale it up to larger geographic areas. Efforts to augment crash data are not new. A notable health-oriented example which sought to link health and police records was the Crash Outcome Data Evaluation System (CODES). Although this federal program ended in 2013, some states, including California, North Carolina, and Tennessee, have continued this effort.

Added data and analytics will result in a more “complete picture” of crashes and injuries. This complete picture enables researchers to improve their modeling, assist policy makers, and contribute to visualization that helps tell compelling safety stories that guide safety improvements,

The key objectives of this study are:

  • To complete the picture of crashes and determine which elements of data that exist outside of conventional crash data can contribute to this picture. These elements likely include EMS, ED, DMV, health expenditure, census, and land use, among others. We will build on existing efforts (e.g., CODES, CMOD, SW8 etc.) and hope to understand how emerging datasets can be mapped to crash data.
  • To identify innovative statistical, probabilistic, and big data visualization tools to link crashes with other records, either by record-matching or by augmenting datasets based on spatial or temporal indicators to perform more-advanced safety analysis

Project outcomes of this study are:

  • Comprehensive data map: A safety-oriented data map that will inform methods to link datasets. This map will result in a more complete picture of crashes, where researchers can use the framework to improve safety analysis.
  • Improved modeling efforts: Supplementing crash data with complementary legacy and emerging datasets will help improve modeling efforts. These linkages will be tested through three proof-of-concept applications.
  • Journal publications: We intend to publish results of this study, focusing both on framework methodology and applications, and anticipate at least three journal articles from this effort.
  • Stakeholder engagement and tech transfer: We will engage stakeholders through our partnerships with CMOD and other NC and TN departments of health, transportation, and safety. The project will contribute to workforce development through the education, training, and professional development opportunities provided to students engaged in the project.

Project Details

Project Type: Research
Project Status: Active
Start Date: 3-1-2017
End Date: 4-30-2018
Contract Year: Year 1
Total Funding from CSCRS: $220,000
Co-sponsors:  MacArthur Foundation Endowment
Collaborating Organizations: Florida Atlantic University; University of California, Berkeley; University of Tennessee, Knoxville; University of North Carolina, Chapel Hill