Michael Escobar (University of Toronto)

In 1990, there was a breakthrough in Bayesian computational methods. Previously, most Bayesian analyses were restricted to simple, limited applications. With the development of Markov Chain Monte Carlo (MCMC) methods, Bayesian inference has become an important applied technique and has been able to handle complex problems. In fact, some problems are now easier to compute with Bayesian methods than with frequentist methods.

The purpose of this course is to introduce applied Bayesian methods to a wide audience. The basic Bayesian philosophy will be discussed and the underlying principles of the MCMC algorithm will be explained. From there, this course will show how to compute and make inferences on complex data problems using these methods. This course does not assume or use any advanced mathematical statistics or calculus. This is not to “dumb down” the material, but instead the goal is to strip away mathematical jargon that may be needed to prove theorems but is not needed to analyse data nor is it needed to explain results to scientific collaborators. Therefore, the mathematical level of this course is at the level of an applied statistics course such as Weisberg’s Applied Linear Regression or Hosmer and Lemeshow’s Applied Logistic Regression. As such, this course should appeal to a wide audience including students in statistics as well as applied statisticians who wish to learn how to use this methods in their practice. Also, faculty members might be interested in this course so that they can present these methods to a general audience of students in applied fields such as epidemiology or psychology.