Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Bayesian 'Power analysis' for calculating sample size with fixed N

Hello,

In the 'The JASP Guidelines for Conducting and Reporting a Bayesian Analysis' it says that 'there is no Bayesian need to pre- specify sample size at all' but if you do need to plan a study with a fixed sample size what is the best means of doing so?

I'd only ever used G*Power prior to learning bayesian methods and I was pleased to have found the Stefan, Gronau, Schönbrodt & Wagenmakers Bayes Factor Design Analysis paper/app, but I'm aware this currently only specifies N for t.tests. Is there a way to calculate a fixed N for Bayesian ANOVAS with default priors? Or is G*power reliable as long as you correct specify power and effect size estimates?

Many Thanks,

Gabriel

Comments

  • Hello Gabriel,

    In general, BFDA is the way to go. We have not implemented this for ANOVAs, however. For fixed sample sizes, Bayesian analyses are more conservative (less trigger-happy is the right way to put it), so what you could do is select an alpha-level of .005 (see the Benjamin et al. 2018 paper on "Redefine statistical significance" for reasons). Of course, when you then do your Bayesian analysis and your evidence is not compelling, you can always run more participants.

    Cheers,

    E.J.

  • Hi E.J

    Thanks again for the advice and help with this!

    All the best,

    Gabriel

Sign In or Register to comment.