Credible intervals for group varying intercepts

[fkgruber]

Hi if I run something like: fit = brm(count ~ log_Age_c + log_Base4_c * Trt_c + (1 | patient) + (1 | visit) + (1 | obs), data=epilepsy, family="poisson")

the summary function only shows the sd estimates and CI of the group intercepts (say patient level). How can I get the CI of the actual intercept estimates? Like for example why you get when you run the same model in stan_glmer

thanks F

[paul-buerkner]

When you just want the estimates of the group intercepts, you can use the ranef method. In order to extract the CIs, you have to do a bit of manual work, for instance:

ps <- posterior_samples(fit, pars = "^r_patient")
t(apply(ps, 2, quantile, probs = c(0.025, 0.975)))

[fkgruber]

thanks!