A Bayesian mindset: From everyday reasoning to modeling

Abstract: In this talk we introduce the field of Bayesian statistics from the conditional probability concept. Bayesian methods combine prior knowledge (prior information) along with data to model and continuously update beliefs (probability) about some event. This area of statistics is based on the work of Thomas Bayes (18th century) but it was not popular until 50 years ago. The development of computer based MCMC methods, combined with increases in computing power and the availability of software for Bayesian computation are major contributors to the significant growth. Bayesian analysis brings a flexible framework to incorporate prior information when available. When non-informative priors are considered, inferences based on Bayesian and classical methods also provide results that are often very similar. It is proving especially useful in approaching complex problems, including clinical research, design and analysis of experiments, and engineering problems.