Skip to content
Prev 5513 / 5636 Next

[R-meta] R-sig-meta-analysis Digest, Vol 89, Issue 13

Dear Wolfgang,

Many thanks for your advice and sorry about the delay. 
I added my answers below! Most things are sorted and clear! However, I did not follow your reply about the back-transformation of the directly estimated co/variance for fixed effects, i.e. from the rZ to the r the level, using the deltamethod. I can see how you derive the SE for the fixed effects at the back-transformed level with deltamethod, but not the variance and covariance. Could you please help here?

Many thanks,
Beate



Beate St Pourcain, PhD
Senior Investigator & Group Leader
Room A207
Max Planck Institute for Psycholinguistics | Wundtlaan 1 | 6525 XD Nijmegen | The Netherlands


@bstpourcain
Tel:?+31 24 3521964
Fax:?+31 24 3521213
ORCID: https://orcid.org/0000-0002-4680-3517
Web: https://www.mpi.nl/departments/language-and-genetics/projects/population-variation-and-human-communication/
Further affiliations with:
MRC Integrative Epidemiology Unit | University of Bristol | UK
Donders Institute for Brain, Cognition and Behaviour | Radboud University | The Netherlands

My working hours may not be your working hours. Please do not feel obligated to reply outside of your normal working schedule.

-----Original Message-----
From: Viechtbauer, Wolfgang (NP) <wolfgang.viechtbauer at maastrichtuniversity.nl> 
Sent: Monday, November 11, 2024 4:13 PM
To: St Pourcain, Beate <Beate.StPourcain at mpi.nl>; R Special Interest Group for Meta-Analysis <r-sig-meta-analysis at r-project.org>
Subject: RE: R-sig-meta-analysis Digest, Vol 89, Issue 13

Dear Beate,

Some brief comments from my side to your questions below.

Best,
Wolfgang
Unless your sample sizes are small, the difference between ML and REML should be negligible. For model selection including fixed effects, ML is the sensible choice.

R: Perfect. Thank you!
Maybe this is useful:

https://www.metafor-project.org/doku.php/tips:i2_multilevel_multivariate#multivariate_models

R: Very helpful, thank you!
I can't answer that because you would first have to define what you mean by 'residual heterogeneity'.

R: We aimed to estimate the heterogeneity in rZ-scores, analogous to the bivariate example approach (but now with two random effects)
https://www.metafor-project.org/doku.php/analyses:vanhouwelingen2002
The deltamethod() function from metafor can give you the full back-transformed var-cov matrix.


R: Sorry for being slow here, I do not know how to proceed. The variance covariance matrix is directly estimated and there is not transformation involved.  Could you please advise
I don't understand the question. What do you mean by 'more elegant'?
R: No worries, we will sort this out. I just meant a "shorter" way, but this is not a problem.
Message-ID: <753411b5e68046ea8d4fab25a0059440@mpi.nl>
In-Reply-To: <AS8PR08MB9193EE54D602951C42E8E4CF8B582@AS8PR08MB9193.eurprd08.prod.outlook.com>