[R-meta] metafor: Categorizing a continuous moderator in random part of rma.mv

[Jack Solomon]

Dear Luke,

It's possible that what you are looking for is available in struct="GEN"
(not yet documented). I'm not sure it's going to be relevant to what you're
doing, but I just offered some thoughts on it here:
https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2021-August/003061.html .

Kind regards,
Jack

On Sun, Aug 15, 2021 at 8:07 AM Michael Dewey <lists using dewey.myzen.co.uk>
wrote:

> It would seem a bizarre thing to do to me unless it corresponds to some
> scientific question.
>
> Michael
>
> On 14/08/2021 16:59, Luke Martinez wrote:
> > Thanks, Michael. But my question is wouldn't it be ***incongruent*** if
> > I use the same variable (`count_of_exposure`) as continuous in the fixed
> > part but as categorical in the random part (at least that congruence is
> > needed in conventional multilevel models)?
> >
> > Kind regards,
> > Luke
> >
> > On Sat, Aug 14, 2021 at 6:35 AM Michael Dewey <lists using dewey.myzen.co.uk
> > <mailto:lists using dewey.myzen.co.uk>> wrote:
> >
> >     Comments in-line
> >
> >     On 14/08/2021 05:53, Luke Martinez wrote:
> >      > Dear Experts,
> >      >
> >      > In metafor's multivariate specification (which gives control over
> the
> >      > `struct` argument), apparently there is a requirement for the
> >     variable
> >      > appearing on the left-side of `|` to only be a categorical factor.
> >      >
> >      > First, why is this required conceptually?
> >      >
> >
> >
> >     Looking at the different options for the struct parameter it is hard
> to
> >     see what sense they make for a continuous variable. What dimensions
> >     would the variance-covariance matrix have?
> >
> >      > Second, suppose I want to use, say, `count_of_exposure` as a
> >      > **continuous variable** in the fixed part, then would it be
> >     appropriate if
> >      > I use `factor( count_of_exposure) | study` in the random part?
> >      >
> >
> >     If converting it into a factor is going to lead to a model which fits
> >     then it must have relatively few levels otherwise the VCV matrix is
> >     going to expand uncontrollably. If it has few levels is it useful to
> >     fit
> >     it as continuous rather than as an ordered factor?
> >
> >     Michael
> >
> >      > Thank you for your clarification,
> >      > Luke
> >      >
> >      >       [[alternative HTML version deleted]]
> >      >
> >      > _______________________________________________
> >      > R-sig-meta-analysis mailing list
> >      > R-sig-meta-analysis using r-project.org
> >     <mailto:R-sig-meta-analysis using r-project.org>
> >      > https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
> >     <https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis>
> >      >
> >
> >     --
> >     Michael
> >     http://www.dewey.myzen.co.uk/home.html
> >     <http://www.dewey.myzen.co.uk/home.html>
> >
> >
> > <
> http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient>
>
> >       Virus-free. www.avg.com
> > <
> http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient>
>
> >
> >
> > <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>
> --
> Michael
> http://www.dewey.myzen.co.uk/home.html
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>

	[[alternative HTML version deleted]]