[R-meta] meta-analysis for regression commonality analysis

[Viechtbauer, Wolfgang (NP)]

Dear Catia,

I am not aware of anybody ever doing this. There are several practical issues:

1) Are the sampling distribution of communality coefficients normal? If I am not mistaken, the communality coefficients for the 'main effects' are >= 0 and will have a very skewed distribution for variables that are unrelated to the outcome. There won't be an easy normalizing transformation that one can use here. This also raises the issue that values of such a coefficient across multiple studies cannot 'cancel each other out' (as opposed to effect sizes that are directional).

2) What is the sampling variance of communality coefficients? I am not aware of any derivations thereof.

3) Will all studies provide the communality coefficients for the same set of predictors? If not, I am not sure how comparable they would be.

4) Are you interested in just one particular coefficient or all 2^p-1 of them (p = number of predictors)? In the latter case, you are dealing with dependent values, which leads to the usual complications.

Best,
Wolfgang

>-----Original Message-----
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org] On
>Behalf Of Catia Oliveira via R-sig-meta-analysis
>Sent: Friday, 31 March, 2023 21:50
>To: R meta; Philip Dale
>Cc: Catia Oliveira
>Subject: [R-meta] meta-analysis for regression commonality analysis
>
>Hello,
>
>I was wondering if anyone has experience doing meta-analysis for regression
>commonality analysis. Is it possible? Are there any publications reporting
>this?
>
>Thank you.
>
>Catia
>
>--
>Cátia Margarida Ferreira de Oliveira
>Research Associate
>Department of Psychology, Room C222
>University of York, YO10 5DD
>Twitter: @CatiaMOliveira
>pronouns: she, her