29/03/2008

Resources on Bayesian Reasoning

I recently argued that everybody should learn the basics of Bayesian reasoning in school. But rather than just moan about the state of things, why not be helpful and provide a few online resources? So:

1. Elizer Yudkowsky: An Intuitive Explanation of Bayesian Reasoning. You can skip the first eight paragraphs (everything above the first horizontal line).

2. A short and entertaining talk by Peter Donelly on Bayesian reasoning (a term he doesn't use), with applications to medical testing and criminal justice. Warning: Contains statisticians humour.

3. Wikipedia: Bayes' Theorem. Starts out with lots of daunting formulas, but has useful examples and many further links. (Addendum: The previous Wikipedia link was all wrong; now fixed.)

4. Russel Roberts interviews Arnold Kling on the American health care system. Contains a discussion of Bayes' Theorem in the context of medical testing and how doctors are ignorant of it (starting at 16:50).

5. Gerd Gigerenzer and Ulrich Hoffrage, 1995: "How to Improve Bayesian Reasoning without Instruction: Frequency Formats", Psychological Review 102: 684-704. Shows that most people are better at Bayesian reasoning when the problem is presented in natural frequencies rather than probabilities. This is important because most people have a hard time memorizing the actual formula.

6. Overcoming Bias posts on all things Bayesian (advanced; it's probably best to start at the bottom).

If you don't know about Bayesian reasoning and you've recently been tested positive for some medical condition, you want to make sure to check out some of the links (start at the top) - chances are your doctor doesn't know what the result really means (see links 1, 4 and 5).

1 comment:

jsalvati said...

Richard Jeffrey has (he's dead now) has a free pdf book on his site: http://www.princeton.edu/%7Ebayesway/

(it's #1)

There's also a bayesian YouTube series on probability theory. I haven't watched this the whole way through.
http://www.youtube.com/watch?v=GXaKEWrNFww