I’d like to use brms to fit a finite discrete mixture model, with a bit of a twist. The scenario is one where I have a single binary outcome, and some of those outcomes (group A) have a fixed, known set of predictors (say, A, B, and C), and some others are drawn from a mixture of group A and another group, group B, which has partially-shared predictors (B, D, E). My intuitions say this should be doable in brms; I know it’s doable in raw stancode, but the downside to that is that I’d like to be able to make it hierarchical to deal with subject-specific and item-specific differences and I don’t know how to do that in raw stan. Would anyone be willing to offer a toy example that I could base my model on (if this is indeed possible in brms)?

Thank you!