The National Highway Traffic Safety Administration estimated in 1984 that manual 3-point safety belts reduce the fatality risk of front-seat occupants of passenger cars by 45 percent relative to the unrestrained occupant. The agency still relies on that estimate. Shortly after 1985, the prime analysis technique for Fatality Analysis Reporting System (FARS) data, double-pair comparison, began producing inflated, unreliable results. This report develops an empirical tool to adjust double-pair comparison analyses of 1986-99 FARS data. It validates the adjustments by comparing the belt use of fatally injured people in certain types of crashes to belt use observed on the road in State and national surveys. These methods reconfirm the agency's earlier estimates of fatality reduction by manual 3-point belts: 45 percent in passenger cars and 60 percent in light trucks. Furthermore, they open the abundant 1986-99 FARS data to additional analyses, permitting point-estimation of belt effectiveness by crash type, occupant age and gender, belt type, vehicle type, etc. (A)
Abstract