Interpreting Accident Statistics

Abstract

Accident statistics have often been used to support the argument that an abnormally small proportion of drivers account for a large proportion of the accidents. This paper compares statistics developed from six-year data for 7, 800 California drivers with results predicted using compound Poisson models for driver accident involvement that assume specific variations in accident likelihood among drivers. The results indicate that the fraction of drivers accounting for various proportions of all accident involvements is too high to suggest that "chronic" accident repeaters are involved in most accidents.