Set individual priors for random intercepts?


#1

I’m a new user to this is probably a basic question, but I’m trying to figure out how to set priors on specific random intercepts.

brms makes it ridiculously convenient to fit something like this:

brm(bf(response ~ instructions*diagnosis + (1+instructions | subject) + (1+instructions | stimulus)), data=df, family=cumulative("logit"))

In this case, I have prior information about each stimulus (mean and sd of ratings from a large population). Can I specify a gaussian prior on the intercept for each stimulus?

It’s again super convenient to set a prior on the stimulus group, but I’m not sure how to expand #17 to specify parameters for each individual prior. Am I missing something obvious?

15                            sd                              stimulus                      
16                            sd instructionsreappraise       stimulus                      
17                            sd              Intercept       stimulus 

Thanks!!

  • Operating System: MacOS 10.14.1
  • brms Version: 2.6.0

#2

You can’t set individual priors on specific random intercepts via the brms interface.

Of course, you may extract the stan code via make_stancode and amend it to your needs.


#3

good to know, thanks!