Bayes vs frequentist 2 way interaction

Ajestudillo

Hi all,

I am new to Bayesian analysis and I am a bit confused, so hopefully someone will be able to help me here. My design is a 2x2 with one of the variables being between groups (group) and the variable type being within. In the frequentist analysis, I found a main effect of the variable group, F(1,50) = 8.69 , p < .01. A main effect of the variable type, F (1,50) = 181.78, p < .001. The interaction approached to be significant (p = .054).

I paste below what I found in the bayesian analsyis:

I am slightly confused here, as it seems that the data are better explained by a model including the interaction. Am I right? what does this mean in terms of the interaction? do I have support for the interaction here? Should I then explore the interaction?

Thanks for your help!

EJ

Hello Ajustudillo,

The posterior model probability is slightly higher for the model that includes the interaction (0.532) than it is for the model that only has the main effects (0.420). So a smidgen of evidence for the interaction, is what I conclude. You may get a more informative table if you select the best model on top.

Cheers,

E.J.

Ajestudillo

Thanks a lot EJ.

Cheers,

Alex