Bayesian Hierarchical Models for Hunting Success Rates;Doctoral thesis

摘要

This dissertation considers a problem of small area estimation that occurs in the Missouri Turkey Hunting Survey (MTHS). This problem occurs in the estimation of a binomial success rate under post survey stratification. This dissertation proposes model based solutions to this problem through the use of Bayesian hierarchical models which incorporate spatial correlation. Models including covariates are proposed and evaluated. For the MTHS the observed sample size may be a function of the success rate. Additional models which make use of this relationship are formulated and fit to the data. This fitting is achieved through the use of a 2-fold Conditional Auto-Regressive (CAR) model. This 2-fold CAR model allows modeling of two spatially correlated variables which are themselves correlated.