BayesFactor::oneWayAOV.Fstat - only for between subject designs?

MSB

Is the `BayesFactor::oneWayAOV.Fstat` function appropriate only for between-subject designs? Or can it also be used for repeated-measures designs?

(I suspect this is a question for @richarddmorey ?)

EJ

I'll attend him to your post

richarddmorey

It can actually be used for any F statistic that has the desired prior distribution over the effect size (see https://www.sciencedirect.com/science/article/abs/pii/S0022249611000666 for details of the model). It won't give you precisely the same answer as, say, anovaBF, but that's because using it for another model implicitly assumes that the other parameters have been integrated out with non-informative priors.

So the answer is: BayesFactor only uses it for between-subjects designs, but it would be possible for you to use it for other things if you were happy with the model. It's just a Bayes factor for the effect size, given a particular prior and an F statistic. You could thoretically use it anywhere you had an F statistic and you understood what the effect size meant.

MSB

Thanks Richard! This is very helpful!