Article: Bayesian Hodges-Lehmann tests for statistical equivalence in the two-sample setting

This was just published in August. I have not read the paper in its entirety, but I thought this would be of interest.

Addendum: After reading, I thought the paper was a good example of using frequentist criteria to judge the output of a Bayesian method.

I would have liked to have seen the “second generation p-value” approach mentioned. I had thought that using a bootstrap CI of the HL estimate in the context of 2 pre-defined frequentist intervals would be adequate, but never pushed it too far.

But it pulls together a lot of theory and simulation results in one place.

Here is an earlier paper by the same author