This article investigates whether the commonly used quantitativemeasure of crash severity, its velocity change in the crash (delta v), is an accurate measure to use to forecast fatality risk in car collisions. Data taken from the National Accident Sampling System (NASS) for accidents occurring from 1980 to the end of 1986 were used. These data were: driver survival or death; the total velocity changein the crash; and the speed limit of the highway upon which the crash occurred. A plot of the probability that a driver is killed versus delta v indicated that a good relationship exists between the two.However delta v is not known for a large number of cases. This was imputed by two different methods. In the first, it is assumed that, for a given speed limit, the delta v for the cases where it is unknown has the same distribution as for those where it is known. The second approach is based on a closer examination of the relations between risk and speed limit. In both cases, the relationship between delta v and risk is found to be the same as for the cases with known delta v, but shifted downwards. The results indicate that the exponent4 may reasonably reflect the relation between the fatality risk anddelta v
Abstract