Predicting older drivers' accident involvement--Smeed's law revisited.

The present study used Swedish archival data on police-reported accidents and license statistics and Swedish questionnaire data on car use in order to test an underinvolvement hypothesis of older drivers' (65-84 years) accident involvement, postulating that the increase in older drivers' accident involvement is smaller than the increase in older drivers' presence in traffic. The results gave support to the hypothesis. Compared to the accident involvement trend in the general driver population, there was a relative decrease over time in older drivers' accident involvement per driver license and per active driver from 1983 to 1999. It is concluded that linear predictions of older drivers' accident involvement directly based on their increasing presence in traffic are likely to be overly pessimistic.