[R-meta] How to get an I^2 and tests for Residual Heterogeneity and Moderators for a model averaged output

[Viechtbauer, Wolfgang (NP)]

Dear Maggie,

I am not aware of any 'canned solutions' for this. But I don't see why one cannot just compute the weighted average of the I^2 values like one would do with other quantities (where the weights are the model probabilities).

As for directly combining the results from tests (e.g., the test for residual heterogeneity or an omnibus test of multiple moderators): No idea. I don't recall seeing anything about this in Anderson (2008) or Burnham and Anderson (2002) (they don't consider meta-analysis, but the issue must come up in other situations like regular regression models) and a quick online search also didn't yield any useful results. Maybe some method for combining p-values that allows for weights could be used? But this is just a thought, nothing further than that.

Best,
Wolfgang

> -----Original Message-----
> From: R-sig-meta-analysis <r-sig-meta-analysis-bounces using r-project.org> On Behalf
> Of Slein, Margaret via R-sig-meta-analysis
> Sent: Saturday, May 11, 2024 05:36
> To: r-sig-meta-analysis using r-project.org
> Cc: Slein, Margaret <maslein using zoology.ubc.ca>
> Subject: [R-meta] How to get an I^2 and tests for Residual Heterogeneity and
> Moderators for a model averaged output
>
> Hello,
>
> I have several multilevel models (rma.mv) that I have model averaged using the
> MuMIn package.
>
> I’m wondering how calculate an I^2 as well as obtain the test for residual
> heterogeneity and moderators (are typically output in the individual model
> outputs but are not output in the model averaged output). Any assistance with
> this would be greatly appreciated.
>
> Many thanks,
> Maggie
>
> Margaret A. Slein
> PhD student, O’Connor Lab
> University of British Columbia (UBC)
> UBCV occupies the unceded and ancestral territory of the xʷməθkʷəy̓əm (Musqueam)
> People and has been stewarded by them since time immemorial