Estimating Hunting Success Rates via Bayesian Generalized Linear Models

Post-season harvest surveys provide data used in the management of Missouri wildlife. These surveys provide information on the number of animals harvested, hunting pressure and hunter success rate. These estimates provide unbiased results at the statewide level due to the large sample size. However, if this survey information is used to make county estimates, poor results often occur due to small sample sizes. To estimate hunter success at the county level for the 1996 Missouri Turkey Hunting Survey, we developed a hierarchical Bayesian model. Speci cally, we evaluate a generalized linear model that incorporates linear covariate terms in addition to a conditional auto-regressive structure for spatial correlation. Calculation of the posterior distribution is achieved through Gibbs sampling and adaptive rejection sampling. The inclusion of covariate terms is then evaluated using Bayes factors.