--------------- David Schoenfeld 2020-08-15
Freedman suggests two priors, a skeptical prior for efficacy and enthusiastic prior for futility.  I used this in a talk I gave several years ago which I think I showed you before.  At the time of the talk, Mike and Art were actually the names of two of the ARDS network investigator. Art Wheeler you probably knew as he was at Vanderbilt and was the PI of the ARDS network.  We were doing a trial that was suggested by Michael Matthay so both the skeptical and enthusiastic priors where probably realistic.  The idea, however, came from a paper by Larry Freedman which I haven’t been able to find, he is an Israeli statistician.  Speigelhalter also describes the idea in his book.  The talk was well received including the Bayesian section.