3 general questions about JASP

Kindred

Hello,

1 - I change the prior settings “r scale fixed effects", r scale random effects" "r scale covariates”, yet, my P(M) column in my output model comparison table does not change values. I have 5 models so each one is given a P(M) of 0.200, no matter what I change in the settings.

2 - Can you have a strong inclusion Bayes factor (BFincl > 30) for a specific predictor AND strong+ evidence for the Null (BF10 < 0.04) in the model comparison output? Because if the BFincl isn’t high, then we have to exclude it and if we do that, then the BF10 can’t be interpreted, right?

3 - Where are the repeated contrasts in JASP? I've seen people talking about it but I can't find it in JASP

Thanks!

Kindred

Bump :)

EJ

Hello Kindred,

Sorry about the tardy response.

1. The prior settings affect the prior distribution for the model parameters inside the models; they do not affect the prior model probabilities themselves.
2. When the data strongly support H0 over all other models, I do not see how you could at the same time get an inclusion BF that supports the addition of a parameter. Maybe it is not mathematically impossible, because the inclusion BF depends on multiple models, but it should be a curiosity.
3. I am not sure what you mean with "repeated contrasts". We do have contrasts in the frequentist ANOVA, but not yet in the Bayesian ANOVA.

Cheers, E.J.