Next Bayes Club: Getting our hands dirty with some R and JAGS

Our next Bayes Club meeting (this Tuesday) will be a close analysis of Kruschke’s BEST model, which is a model for comparing two groups to see if they’re drawn from populations with different means (i.e. something you would do instead of a t-test). You can download all of his code for this model (and many others) from his website:

http://www.indiana.edu/~kruschke/BEST/BEST.zip

He has written an entire article specifically focusing on this model including details about the code, analysis, etc. Definitely skim it before Tuesday if you get a chance:

http://www.indiana.edu/~kruschke/articles/Kruschke2013JEPG.pdf

Or, if you’re sick of reading things all day long, watch the video instead: http://youtu.be/fhw1j1Ru2i0

 And if you want to play around with a simulation, check this out: http://sumsar.net/best_online/

We’ll spend almost the entire meeting on Tuesday going through the graphical model in detail and going through all the R and JAGS code line by line. We hope that by the end of the day, you’ll be able to manipulate the model yourself (e.g. change the number of groups to do a one-way ANOVA instead of a t-test, etc.). If you don’t already have R and JAGS ready to go on your computer, please try to get them installed before we meet. And, of course, email me or Jacob with any questions.

Extra tidbit: We won’t go through this on Tuesday, but as you move forward, you may want to have a copy of the JAGS manual, which includes a list of all the distributions JAGS understands, along with the code for calling them (see Chapter 6):

http://people.math.aau.dk/~kkb/Undervisning/Bayes13/sorenh/docs/jags_user_manual.pdf

P.S. Sorry none of these links are clickable. WordPress is refusing to play nicely with others. When it’s sorted out, I’ll go back and update this post with real links.