Most applied Bayesian inference is done approximately using samplingbased methods. In my experience, most students struggle to understand why sampling is necessary and also what sampling from a posterior distribution actually means. In this talk, I will provide a few pedagogical hooks that I have found useful for explaining what computational sampling means and why it is necessary. I will also discuss how the predominant sampling method used in applied Bayesian inference, Markov chain Monte Carlo, differs from more familiar independent sampling, which provides the student with an intuitive understanding of effective sample size.

This was recorded as part of the TALMO meeting on Teaching Bayesian Methods.
More information and slides, etc, can be found at http://talmo.uk/2024/teachingbayesian...