Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Comparing models following Bayesian ANOVA

Hello! Hope someone can point me in the right direction. I know the concept but unsure how to process it via JASP.

I have 2 independent variables, A and B. I have done a Bayesian ANOVA to compare the following model:

  1. A vs Null
  2. B vs Null
  3. A + B vs Null
  4. A + B + A*B vs Null

So far so good! But I also need to compare the following:

  1. A + B + A*B vs A + B
  2. A + B vs B
  3. A + B vs A

I am assured this is easily achievable in JASP! I tried to add variables to null hypothesis under the 'model' tab, but I'm unsure this is the correct way. I can obtain Bayes Factors using this method for 6 and 7, but adding A, B, and A*B to null model for 5 produces an error under nuisance variables.

Any pointers greatly appreciated!

H.

Comments

  • Dear H.,

    Yes, this is easily achievable, in two ways. For concreteness, suppose you want A+B+AB vs A+B.
    Method 1. Add "A" and "B" to the null model (under the "model" tab).
    Method 2. Use transitivity: if you already have BFfull = A + B + A
    B vs Null and BFmain = A + B vs Null, then the BF in favor of the full model over the two main effects model is BFfull / BFmain.
    Both methods ought to give the same results. The other comparisons you are interested in can be obtained similarly.

    Cheers,
    E.J.

  • Thank you, EJ.

    I've since learned how to do these multiple model comparisons in a table by hand to obtain the ratios (as you say, full/main) but great to learn how to do this in JASP too. This is our first time using it.

    Cheers!
    H.

    Thanked by 1EJ
Sign In or Register to comment.