Bayesian methods to account for observational uncertainty.

This dissertation is comprised of three chapters that cover Bayesian inference under the event of biases in epidemiology and observational studies. In the first chapter, we provide a brief overview of the topics in the remainder of the dissertation. In chapter two, We develop a Bayesian logistic regression model that simultaneously accounts for covariate misclassification and an unmeasured confounder. In the third chapter, we propose the unmconf R package, providing researchers with a user-friendly way to account for unmeasured confounders in Bayesian regression modeling. Chapter four details the proposal for an R package, sizemebayes, that performs a fully Bayesian sample size determination under the consideration of covariate measurement error and/or potential correlation among the predictor variables. In chapter five, we conclude with a brief summary and discussion.