Publication Details


An investigation of safety belt usage and effectiveness

Type: Paper

Author(s): Hochberg, Yosef; Reinfurt, Donald W.

Pages: 68

Publisher: UNC Highway Safety Research Center

Url: https://trid.trb.org/view.aspx?id=23833

Publication Date: Mar-1975

Number: DOT-HS-801-427

Address: Chapel Hill, NC

Abstract: Seat belt effectiveness in accidents is initially discussed from a population parameter point of view. The report also includes some exploratory studies of the degree of misclassification errors involved in the police reports on belt usage and degree of injury along with some tools to partially resolve such problems. Methodology is developed for investigating the bivariate injury distribution for belted and unbelted drivers to provide insight into the mechanism of the injury-reducing potential of belts in accidents. This interim report contains primarily a theoretical discussion of problems of inference in studies on seat belt utilization and effectiveness. Seat belt effectiveness in accidents is initially discussed from a population paramenter point of view. Three measures are introduced and their features explored. Several inferential problems are discussed and certain simplifications obtained. This report includes some exploratory studies of the degree of misclassification errors involved in the police reports on belt usage and degree of injury along with some tools to partially resolve such problems. Methodology is developed for investigating the bivariate injury distribution for be1ted and unbelted drivers to provide insight into the mechanism of the injury-reducing potential of belts in accidents. The final report will include the analysis of a stratified random sample of 21,000 passenger car observations (taken in N.C. in October 1974) where belt utilization was recorded along with age(approximate}, sex, and race of driver and vehicle license plate number. Regarding effectiveness, the final report will have the analysis of frequency tables of the form: belt usage x injury, adjusted to match desired margins based on external information that will be collected.