Random effects prior - comparison to JAGS

Interfaces brms
1

Hello,
I am making a very simple model with random effects (or which ever name is best used here), say
y ~ 1 + x1 + x2 + (x1)|Group
that is, the estimates for x1 and x2 would be slightly different for each Group, but under a partial-pooling, a shrinkage.
Maybe I am just new to it. Coming from JAGS, I would set (ignoring other settings for now)
y ~ beta0 + beta1[Group]x1 + beta2x2
beta1[Group] ~ normal( beta1mu, beta1sigma)

However, when I do it in brms, I can only set prior for the “sd” of x1.
Therefore, I understand, with brms I can set the prior for beta1sigma, but not for beta1mu.
Do I understand it correctly?

Thank you very much

  • Operating System: Windows 10
  • brms Version: 2.12.0
2

You should be able to set a prior on the mean of x1. That’s the x1 term that sorta comes about regardless of the group.

Not sure it’s exactly the same as setting the group mean but seems similar? Maybe make some fake data and see what happens when you fit the models (just to make sure things are identifiable and stuff).