Howdy, Stranger!

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

Supported by

Effect of covariates on post-hoc tests in Bayesian ANCOVA

Hi,

I am trying to run a Bayesian ANCOVA in JASP with one fixed factor with three levels and one (continuous) covariate (added to the null model).

I have noticed that when conducting pairwise post-hoc tests, the inclusion of the covariate does not seem to have any effect on the results (i.e. Bayes Factors etc. for the post-hoc comparisons are the same whether the covariate is included in the model or not).

Is there any way of conducting post-hoc tests that take into account covariates as it is possible in the case of the frequentist ANCOVA?

Many thanks for your help!

Best wishes,

Julia

Comments

  • Hi Julia,

    Good question -- I'll forward this to our expert.

    Sorry about the delay.

    Cheers,

    E.J.

  • Hi Julia,


    The pairwise post-hoc tests are simply independent t-tests with a correction for multiple testing. So including or excluding a covariate does indeed not impact the results. As far as I know, we currently do not offer a way to include covariates in Bayesian post-hoc tests.

    I hope that answers your questions.

    Best,

    Don

  • Addendum: but it would be a good idea to develop this functionality in the future!

    EJ

Sign In or Register to comment.