[R-meta] Standardized mean differences with rma.mv?

[Viechtbauer, Wolfgang (NP)]

Will, *please* read up on the extensive documentation provided with the metafor package, including:

https://wviechtb.github.io/metafor/reference/rma.mv.html
https://wviechtb.github.io/metafor/reference/vcalc.html
https://wviechtb.github.io/metafor/reference/misc-recs.html#general-workflow-for-meta-analyses-involving-complex-dependency-structures

and the various examples illustrating the application of multivariate/multilevel models for modeling dependent estimates with rma.mv(), such as:

https://wviechtb.github.io/metadat/reference/dat.assink2016.html
https://wviechtb.github.io/metadat/reference/dat.berkey1998.html
https://wviechtb.github.io/metadat/reference/dat.ishak2007.html
https://wviechtb.github.io/metadat/reference/dat.kalaian1996.html
https://wviechtb.github.io/metadat/reference/dat.knapp2017.html
https://wviechtb.github.io/metadat/reference/dat.tannersmith2016.html

and the various analysis examples provided on the metafor website, such as:

https://www.metafor-project.org/doku.php/analyses:gleser2009
https://www.metafor-project.org/doku.php/analyses:berkey1998

and the corresponding references.

Best,
Wolfgang

> -----Original Message-----
> From: R-sig-meta-analysis <r-sig-meta-analysis-bounces using r-project.org> On Behalf
> Of Will Hopkins via R-sig-meta-analysis
> Sent: Thursday, April 18, 2024 23:15
> To: 'R Special Interest Group for Meta-Analysis' <r-sig-meta-analysis using r-
> project.org>
> Cc: Will Hopkins <willthekiwi using gmail.com>
> Subject: Re: [R-meta] Standardized mean differences with rma.mv?
>
> Sorry, guys, yes, I got confused. I am guilty of not reading the
> documentation thoroughly, and I am still an R and metafor newbie. So if I
> understand correctly now, escalc() can turn mean effects into a dataset of
> standardized mean effect sizes, with their variances. That gets fired into
> rma.uni(). If I have multiple effect sizes for each study, escalc() treats
> them as independent effects. I can use aggregate() on the output of escalc()
> to get one effect per study and then analyze with rma.uni(), but from what
> Reza has stated, I can't fire the output of escalc() into rma.mv() to get
> within-study effects and within-study heterogeneity. Really? But I can
> always do my own standardizing of the multiple effects and fire that into
> rma.mv()?
>
> -----Original Message-----
> From: R-sig-meta-analysis <r-sig-meta-analysis-bounces using r-project.org> On
> Behalf Of Reza Norouzian via R-sig-meta-analysis
> Sent: Thursday, April 18, 2024 11:04 PM
> To: R Special Interest Group for Meta-Analysis
> <r-sig-meta-analysis using r-project.org>
> Cc: Reza Norouzian <rnorouzian using gmail.com>
> Subject: Re: [R-meta] Standardized mean differences with rma.mv?
>
> Will,
>
> It seems like you're possibly extending the "measure=" argument in
> rma.uni() to rma.mv() which the latter doesn't support.
>
> Reza
>
>
> On Thu, Apr 18, 2024, 5:19 AM Viechtbauer, Wolfgang (NP) via
> R-sig-meta-analysis <r-sig-meta-analysis using r-project.org> wrote:
>
> > Dear Will,
> >
> > I am afraid I don't understand your question. What do you mean by
> > "calculation of standardized effects" and what does escalc() have to
> > do with this? rma.mv() fits models, escalc() computes various types of
> > effect sizes and corresponding sampling variances, so I cannot quite
> > make sense of how you are contrasting these two functions.
> >
> > Best,
> > Wolfgang
> >
> > > -----Original Message-----
> > > From: R-sig-meta-analysis
> > > <r-sig-meta-analysis-bounces using r-project.org>
> > On Behalf
> > > Of Will Hopkins via R-sig-meta-analysis
> > > Sent: Tuesday, April 16, 2024 23:39
> > > To: 'R Special Interest Group for Meta-Analysis'
> > > <r-sig-meta-analysis using r- project.org>
> > > Cc: Will Hopkins <willthekiwi using gmail.com>
> > > Subject: [R-meta] Standardized mean differences with rma.mv?
> > >
> > > I searched the archive, and if I missed it, I apologize, but is it
> > > true
> > to
> > > say that currently multivariate analyses of the kind offered by
> > > rma.mv (when, for example, studies provide two or more estimates for
> > > the meta)
> > do
> > > not allow for calculation of standardized effects of the kind
> > > offered by escalc (e.g., using measure="SMD1")? I realize one could
> > > do the analyses oneself with rma.mv by pre-standardizing the mean
> > > effects, but you would have to be quite savvy with estimating the
> > > standard errors of the standardized effects.
> > >
> > > Will