An empirical Bayes model to assess deer-vehicle crash safety in urban areas in Iowa

Abstract

Deer-vehicle crashes are a growing problem in Iowa. In 2008, deer-vehicle crashes represented 12% of all crashes reported and included 9 fatalities and 442 injuries. These crashes are especially problematic in urban areas of Iowa, where the prevalence of deer-vehicle crashes is becoming a more visible issue. Quite a bit of research has been conducted on countermeasure action that could help mitigate deer-vehicle crashes. However, little previous work has attempted to model deer-vehicle crashes in urban areas using the two data sources available: deer carcass salvage reports and deer-vehicle crash reports. The objective of this paper is to assess the safety of roadway segments in three cities with long-running deer management programs using both deer-vehicle crash and deer carcass salvage data. The authors reconciled records to help eliminate double counting and estimated count data models to examine deer-vehicle crash frequency as a function of roadway and environmental factors. The count model estimates were used in an empirical Bayes model to predict deer-vehicle crashes in the select urban areas of Iowa. This model can be used to help allocate safety funds to implement appropriate deer-vehicle crash countermeasures in high-risk locations.