priorsense 1.1.0 is now available on CRAN.
priorsense is a package for efficient prior (and likelihood) sensitivity checking implementing the method described in the article by me, @topipa @paul.buerkner and @avehtari.
The main feature of this release is the ability to select priors to power-scale for brms models. This is a major improvement in the sensitivity checking workflow, as it is now much easier to pinpoint which of the priors is potentially informative or in conflict with the data/likelihood. Before this, it was only possible to check the sensitivity to all the priors in a brms model at once.
This makes use of the new tag argument in brms::prior() which is available in brms version 2.22.11 or later (not currently on CRAN, so you’ll need to get the GitHub version). Big thanks to @paul.buerkner for help getting this feature merged.
For details on selecting priors for sensitivity checks using priorsense and brms, see the new vignette. In this same vignette there is also guidance on which posterior quantities to inspect in sensitivity checks.
This release also brings improvements to plots, JAGS support via R2jags (along with a short vignette), and various bug fixes.