Model with category-specific effects AND separate random effects for each group (brms error)

I’m fitting a sequential ordinal model (family = sratio) and would like to set up a random-effects variance-covariance matrix for each level of a grouping factor AND to include category-specific effects as random slopes.

This works:

bf(Response ~ 1 + cs(Group * Condition) +
             (1 + Condition | gr(Participant_ID, by = Group))

This also works:

bf(Response ~ 1 + cs(Group * Condition) +
             (1 + cs(Condition) | Participant)

But this doesn’t:

bf(Response ~ 1 + cs(Group * Condition) +
             (1 + cs(Condition) | gr(Participant_ID, by = Group))

I get the following error message:

Error in frame_re(x, data = data, old_levels = old_levels) : 
  rsub$type[i] %in% c("", "mmc") is not TRUE

Is such a model possible in brms?

  • Operating System: Linux
  • brms Version: 2.22