This post is targeted directly @paul.buerkner but anyone who knows the answers please feel free to jump in as well on the discussion.
We are trying to fit a hierarchical bayesian survival model with a site-specific intercept and a site-specific coefficient for the covariate trt (i.e. treatment) using the brms and rstanarm packages.
In rstanarm the model formulation for this is: formula = Surv(eventtime, status) ~ trt + (trt | site).
We have tried several formulations in brms and are able to match the estimates however our credible intervals are completely different. This is whilst using the same priors, seed, chains, cores, iterations and adapt delta specifications. Is there a way to model the above rstanarm specification in brms? Many thanks in advance.