Effective use of finite roadway improvement budgets to accommodate an increasing number of older drivers requires that we be able to identify those locations where older drivers appear to have a heightened accident risk. Ideally, the accident records from a location (such as a particular intersection) should provide the information needed to assess the risk experienced there by a given group of drivers, but the lack of location and age-specific measures of exposure coupled with the relatively small accident samples available for particular locations makes the standard methods of high-hazard identification inapplicable. In this paper it is first shown how, by using an induced exposure approach, it is possible to test for the equality of group-specific accident rates at a given site by testing for the equality of two binomial probabilities which arise from a particular type of contingency table. It is next shown how an Empirical Bayesian approach to computing point and interval estimates for binomial probabilities, which has appeared in the statistical literature, can be adapted to this problem. The resulting computational procedures are relatively straightforward and can be implemented on a microcomputer. The method is tested using accident data from two Minnesota highways, with the following results. First, the method's underlying probability assumptions appear to be supported by the data sets. Second, a Normal distribution approximation used to compute confidence intervals showed acceptable accuracy. Third, the EB estimation procedure showed a tendency to reduce the scatter characteristic of Maximum Likelihood estimates while still identifying locations showing elevated accident risk for a given driver age group. It is concluded that the method could be a useful tool for safety engineers, and some additional research and development necessary for practical implementation of the method is recommended.
Davis, Gary A..
Statistical Method for Identifying Areas of High Crash Risk to Older Drivers.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.