Propagating Uncertainty in Dependent Models

Interfaces brms
1

Hello,

In this comment: https://github.com/paul-buerkner/brms/issues/303#issuecomment-353170246
how is the uncertainty in the point estimates of y1 (i.e. fy1) propagated in the subsequent call to fitted (for the second model). The second call to fitted (seemingly) only takes into account point estimates of the fitted y1.

Thanks in advance

2

fy1 does not contain the point estimates but the posterior distribution (in the form of posterior draws) of all predictions. I suggest you run the provided code line by line and closely look at the results of each step. Maybe then the procedure becomes clearer.

3

aah, you are right. thanks much.