[R-meta] Using variance components with effect size transformation
Viechtbauer, Wolfgang (NP)
wo||g@ng@v|echtb@uer @end|ng |rom m@@@tr|chtun|ver@|ty@n|
Wed Dec 14 15:44:07 CET 2022
The 'transf' argument in confint() only applies to fixed effects ('model coefficients'). One cannot directly transform the variance components from one scale to another (there are in principle ways of doing this, but this rarely is appropriate).
>From: Wolfgang Viechtbauer [mailto:wviechtb using posteo.de]
>Sent: Wednesday, 14 December, 2022 15:40
>To: Viechtbauer, Wolfgang (NP)
>Subject: Fwd: [R-meta] Using variance components with effect size transformation
>Dear Meta Community,
>I have a 3-level meta-regression model with a categorical moderator (3
>levels) plus some covariates fit as:
>m6 = rma.mv(r2z_transformed ~ 0 + cat_mod + covariates, random = ~ 1 |
>I wonder to calculate probabilities from the distribution of effects as
>explained in this post (
>whether I should use the transformed variance components (A) or
>back-transformed variance components (B)?
>A: confint.rma.mv(m6) # transformed i.e., 'r2z'
>B: confint.rma.mv(m6, transf = transf.ztor) # back-transformed i.e.,
>Currently, A and B give the same result, and I wonder why?
More information about the R-sig-meta-analysis