Directional hypothesis in Bayesian Anova

danilinares

Hello,

I wonder whether it is possible to test a directional hypothesis using JASP. I have a 2 x 2 anova without interaction and I would like to test a directional hypothesis for a main effect. I posted a question in crossvalidated:

https://stats.stackexchange.com/questions/532531/how-to-perform-a-directional-hypothesis-using-bayes-factors

Thanks.

EJ

Hi Danilinares,

Yes that is a really good question, and one that is important too. In fact we currently have this under investigation. It is not something that JASP does right now (but we will in the future).

Cheers,

E.J.

andersony3k

My thought is, that would only make sense in a narrow range of circumstances. One circumstance is when there are just two levels of a factor being compared, in which case a directional t test could be used instead of an ANOVA. Another is when the design is a two by two between-subject or mixed factorial. However, ANOVA in the more general case assesses whether a partitioned, across-conditions variance is significantly greater than zero, which is not a direction hypothesis: A variance across three means, for example, does not have a direction.

EJ

Yes, but I believe that substantive hypotheses usually translate to ordinal expectations. We have developed something like this for the multinomial, see

https://psyarxiv.com/bux7p/

Cheers,

E.J.