search

wwiecek / baggr

R package for Bayesian meta-analysis models, using Stan
GNU General Public License v3.0
46 stars 12 forks source link

individual-level data mu & tau model #96

Open wwiecek opened 3 years ago

wwiecek commented 3 years ago

In the next release (sslab branch) rubin_full and logit models use the same syntax that allows setting a prior on baselines -- but there is no option to do a joint prior on basekline and trt effect the same way we do it in mu & tau model; can we fix it?

Ideally 1) mu&tau would then have the same syntax as rubin_full 2) rubin_full could handle both a joint prior on bsl and trt effect, as well as 2 separate priors

wwiecek commented 3 years ago

If 2 independent priors, then specify via prior_hypermean and prior_control etc. If joint, then specify via prior_hyper... but set joint=TRUE?

wwiecek commented 3 years ago

Then we would have to re-write the help files and "Prior choice in baggr" in the main vignette

wwiecek commented 3 years ago

I had a stab at this but it seems to require a longer re-design. Should use the same Cholesky decomposition that I am planning for mu&tau model in #102