Speaker: Dr. Joseph Lucke, University at Buffalo
Research Institute on Addictions
Title: A foundational Introduction to Bayesian Statistics
Abstract: Bayesian statistical methods are often adopted for purely pragmatic reasons in order to solve statistical problems. Here I instead present Bayesian Statistical Theory from its foundations on probability. I first critique the frequency theory of probability as chasing ephemera. I next introduce Bayesian statistical theory from its three pillars: coherence, exchangeability, and Bayes's Theorem. I explicate each of the three pillars and conclude that subjective probability is objective. I briefly discuss the sources of prior information required for Bayesian analysis. Finally as a concrete example, I compare Frequentist and Bayesian approaches in simple hypothesis testing.