[R-meta] Questions about tradeoffs w/ model averaging rma.mv models w/ other statistical nuances

[Slein, Margaret]

Hello,

I am currently trying to assess the tradeoffs in paths for a meta-analysis that I have been working for a while (too long) now.

Essentially, we did model selection on a slew of rma.mv models with different fixed effects and ended up with 3 models within 4 AIC of each other. We decided to model average using the model.avg() function from the MuMIn package.

Wondering if anyone has advice for scaling predictor estimates based on partial standard deviations to assess statistical significance and methods you can cite for this?

Also wondering about how to run omnibus tests on these models if we are scaling based on partial standard deviations?

Any assistance would be greatly appreciated.

Cheers,
Maggie




	[[alternative HTML version deleted]]