Bayesian repeated measures ANOVA, covariate in winning model has BFincl below 1?

Sofaj

Dear fellow JASP users,

I'm rather new to JASP and I am a bitt puzzled by the results I have recently got from a Bayesian rmANOVA. I have carried out a cognitive task, and I have three factors (2 levels in each) and six covariates. My winning model contains one of the covariates (TCI-R Persistence). When looking closer at analysis of effects, however, this same covariate have a BFinlc below 1. Another covariate, TCI-R Harm avoidance, has a BFincl above 1 but is not included in the winning model. I thought it had to be above 1 to be included in the winning model, but perhaps I have misunderstood this completely?

Thanks in advance for anything that may help me set this straight!

Best,

Sofie Nilsson

EJ

Hi Sofie,

The analysis of effects considers *all* models in the set; it sums the posterior probability for all models that include the effect (and then compares the posterior inclusion odds to the prior inclusion odds to obtain the inclusion BF). So apparently there are many decently-performing models that do not require Persistence, even though it is present in the winning model.

Cheers,

E.J.

Sofaj

Hi E.J.,

That makes sense, a lot of different models in our analysis seem to explain our data well.

Thank you very much for your reply, it was very helpful!

Best,

Sofie