The Restraint Systems Evaluation Program (RSEP) and the National Crash Severity Study (NCSS) utilized a "towaway" threshold for inclusion of accident vehicles in their samples. It is also going to be the major criterion for much of the National Accident Sampling System (NASS). In this report, the effect of using such a threshold rather than "all accidents" is investigated. In the first part of the report, the independent variables which are most closely associated with post-crash vehicle drivability are identified. Two different data sources, 1976 North Carolina accident data and 1975 New York accident data, were examined. The variables identified in single vehicle and multi-vehicle accidents in the two files were very comparable. A log-linear model was fitted to the North Carolina data and predicted towaway odds as a function of the identified independent variables presented. For example, the odds of being towed versus the vehicle being drivable in high speed, front impact single vehicle crashes are from four to ten-fold greater depending on the object struck. In the second part of the report, accident and injury characteristics along with seat belt usage and effectiveness estimates are compared as a function of the sampling criterion..
Abstract