Paramater Estimation ANOVA
This is my first time doing Bayesian analysis and first time using JASP. So there is a high probability this is a dumb question, but I am hoping someone can send me in the right direction.
I have data structured as a 2x2 ANOVA. I've done the Bayes ANOVA and found my Bayes exclusion factor. It shows a moderate evidence that the null is true.
Another paper in my field does a Bayesian ANOVA and follows it up with parameter estimation for the coefficient on the interaction term. I can't find anything at all on this so I'm not sure where to even start. Could anyone point me in the right direction for parameter estimation? Also, I know that with parameter estimation I need to set a ROPE. There is no prior work to rely on to set a ROPE...is there anyway to have a default ROPE? Or if you can't set a ROPE do you recommend not doing this analysis? Thanks!
Comments
Hi ElizaSells,
I'll answer your first question regarding parameter estimates but can't answer your question regarding ROPE.
Regarding parameter estimates, there is a great tutorial paper on Bayesian ANOVA that has a section titled 'Parameter estimates'. That should guide you on how to get parameter estimates. You can even download the raw .jasp files in the tutorial paper and see what options were checked/unchecked. I believe the 'Estimates' checkbox will give you what you want.
Cheers :)
P.S - Just saw your other question about looking at the effect of a certain Factor. You could do it via the way you mentioned there but I would probably be inclined to do it via the 'Analysis of Effects' option. This is also described nicely in the paper.
Thanks rohanp16. I will check that paper out. It sounds really helpful!
Dear Eliza,
As far as ROPE is concerned: I have some issues with that analysis, in the sense that it depends crucially on the bounds of an interval and that it does not quantify evidence. So I personally do not believe it answers a relevant question. However, it is popular, and if you know what you're doing there may be an argument in its favor, so we might include this approach in a later version of JASP.
Cheers,
E.J.
Thank you very much for the help EJ! I think we will forego this analysis.