Hello,
I am modelling a multiple linear regression with varying-slopes in brms.
From the vignettes and function descriptions, I understood I can feed a prior for the standard deviations, e.g. set_prior(“gamma(2,5)”, class = “sd”, coef …).
Reading the model code after running brm(), trying to understand what brms actually does, I find that the group-level effect is, for example, r_1_1 = (sd_1[1] * (z_1[1]));
I could give a prior for sd, but not for z_1[1]
where z_1[i] is a standard normal, i.e. normal(0,1) . Have I understood correctly?
So I can only change the sd_ prior ?
In that case, all group-level effects have a prior centered at 0 ? what if I have a strong prior that the group-level effect should be something between 2 and 9, I want a prior centered at 5 and a sd around 3 ?
Is that even possible within brms?
Thank you very much for your attention and support!