I am writing a paper that tests multiple hypotheses under a Bayesian framework.

The hypotheses are ~15 previously established hypotheses which I am retesting under a phylogenetic framework (to control for interdependencies between data). Effectively the procedure works like this, I have a hypotheses binary variable X co-evolves with binary variable Y. I then run two models, one where changes in X depends on changes in Y and vice-versa along the phylogeny, and a model where they can change independently and calculate a Bayes Factor to determine which model fits the data better.

I think that from a more philosophical standpoint this isn't an issue because I am testing previously established hypotheses, and this is effectively re-testing them under a different approach. Which is discussed here

However, is there also a statistical argument here that can be linked to this argument? Here, it seems to be suggested that Bayesian models (w/o flat priors which is the case for me) don't need to worry about multiple comparisons when making claims of confidence. While I can follow the simulations in the linked post, I don't really understand why this is the case, unless it is just that 95% credible intervals are just more conservative?

TLDR; Is there a statistical argument for not needing to correct for multiple comparisons when performing bayesian model comparisons tests?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.