Bayesian RM ANOVA sphericity and post hoc options

64bit

Two questions about the Bayesian RM ANOVA...

1. Any update on when testing for sphericity will be implemented?
2. For post hoc tests the mean differences (raw units) and credible intervals aren't displayed, nor are the Cohen's d and credible intervals. In the frequentist version these options are available, so can these be added for the Bayesian version please?

Cheers

Grant.

andersony3k

My understanding is that "Bayesian Repeated-Measures ANOVA" is a misnomer. What JASP calls "Bayesian RM ANOVA" is really a Bayesian linear mixed-effects model (with a tweak or two). And I'm fairly sure that while "sphericity" is an assumption underlying RM ANOVA, linear mixed effects models don't assume "sphericity."

64bit

Thanks for the info 😊. I was aware that the Bayes RM ANOVA in JASP is a mixed model under the hood, but wasn't aware that sphericity wasn't an assumption of mixed models 👍️.

andersony3k

@harlowhenry

Is this forum an official JASP support channel?😉

EJ

Yes this is the official channel :-) Sphericity is still on the to-do list. Equality of standard deviations is implicitly assumed by LML also. We have concrete ideas on how to implement this, but have not gotten round to it. See also https://www.frontiersin.org/articles/10.3389/fpsyg.2017.01841/full

EJ

EJ

Oh, and about the post-hoc tests: it works best if you push this feature request on our GitHub page (for details see https://jasp-stats.org/2018/03/29/request-feature-report-bug-jasp/) because this is what most of the team sees.

EJ

andersony3k

But there does not appear to be a consensus that linear mixed models rely on a sphericity assumption:

"Another advantage of mixed models is that we don't have to be consistent about time. For example, and it does not apply in this particular example, if one subject had a follow-up test at 4 months while another had their follow-up test at 6 months, we simply enter 4 (or 6) as the time of follow-up. We don't have to worry that they couldn't be tested at the same intervals. A third advantage of these models is that we do not have to assume sphericity or compound symmetry in the model. We can do so if we want, but we can also allow the model to select its own set of covariances or use covariance patterns that we supply."

https://www.uvm.edu/~statdhtx/StatPages/More_Stuff/Mixed-Models-Repeated/Mixed-Models-for-Repeated-Measures1.html