This study seeks to merge two US DOT databases—one listing hazardous material release incident reports with the Motor Carriers Truck Crash reports. The goal is to identify crash attributes (including injuries and fatalities) with the type and amount of hazardous material released in order to estimate the economic cost of such crashes.
Researchers first estimate the likelihood scores for matches between records held in the two databases. These scores can then be used to estimate the probablility of a match, and run a Bayesian analysis to estimate the regression coefficients.
Findings will allow the DOT to achieve more realistic cost estimates of crashes on US roads and highways.