The Skeptic's Gambit

Case

Vale wants the Board to borrow strength from earlier trials, and Kastner wants to know what the priors are doing.

What this scene will ask

Bayesian NMA posterior draws under three priors, all supplied.

Planned flag: the only treatment whose probability of being best exceeds 0.9 under every prior; that probability under the skeptical prior, 3 dp; its posterior median effect versus reference under that prior, 3 dp; posterior odds* that it is best, 1 dp.

Status

[#] Sealed. This scene opens when it is written and passes review. Nothing on CTFEpi expires, so there is no rush.