BSTA 521 Bayesian Methods for Data Analysis
This course is an elective course for students in the MS, MPH and Graduate Certificate program in Biostatistics and may also be used as an elective course for students in MPH and PhD in Epidemiology programs and other programs, if they have taken the appropriate prerequisites. The methods students learned in the biostatistics applied and theoretical sequences were based on the “frequentist” method of statistical reasoning, where probability is understood to be the longrun frequency of a ‘repeatable’ event, and statistics that are computed are based on a specific study only. Bayesian methods are based on a different philosophy – that probability of an event is based on ALL information known at the time. Bayesian methods for data analysis enable one to combine information from previous similar and independent studies (prior information), with information from a new study, yielding updated inference for model parameters. This course will cover the concept of Bayesian analysis, posterior distribution, Bayesian inference and prediction, prior determination, one parameter and two parameter models, Bayesian hierarchical models, Bayesian computation, model criticism and selection as well as basic comparison of Bayesian and Frequentist Inferences. Real life examples in medical and health science will be used to explain the concept and application of Bayesian models.