How to interpret model with “main effect 1 + main effect 2 + main effect 1*main effect 2"?
Hello everyone,
I have got a model which has two independent variables (IV):
Group: Explorers vs. Time Controls (between-subjects)
Season: Summer 1, Autumn, Winter, Spring, Summer 2 (within-subjects)
In JASP's Bayesian RM ANOVA this shows me that this model has the most support:
Group + Season + Group*Season (BF10=870).
The closest single main effect model is Season (BF=190).
How do I interpret the Group + Season + Group*Season model correctly?
I've read this blogpost https://www.cogsci.nl/blog/interpreting-bayesian-repeated-measures-in-jasp
And I've read Part I and Part II of the JASP papers for psychology but I'm not sure how to make sense of this model.
Best regards
eniseg2
Comments
Hi Eniseg2,
Well, the model with the interaction is preferred over the other models. Compared to the model with only Season, the BF = 870/190 = 4.58. Maybe it also helps to select "Compare to best model" and set the Bayes factor display to BF_01 (instead of BF_10): this will tell you how much more support the data provide for the best model compared to each of the other models.
Cheers,
E.J.
Hi E.J.
Yes, I have understood that. But what I'm wondering how to explain what the model means. If you have a single main effect model, you can say 'Season affects Mood' for example. If it's an interaction you can say 'Season affects Mood depending on Group'. But what can I say with this model which has both main effects and the interaction in it?
Thank you very much,
Eniseg2
I'd say you interpret the interaction, as you propose. JASP respect the principle of marginality, so models with the interaction automatically also include the constituent main effects.
E.J.