[R-meta] Questions about interpreting the intercept in a rma.mv model

[Reza Norouzian]

Hi Maggie,

Without reproducible code, it's difficult to pinpoint what has gone
wrong on your machine (e.g., an installation problem, a data/model
issue, or something else).

As an alternative, you can try the code provided in the article linked
below. There, you'll need to first install the R programs by running
code snippet #1 on p. 22. Then, you can do:

post_rma(trial, ~region+impact)

It also seems that you're using log response ratios as your effect
size measure? if so, then in case, you prefer your marginal mean
estimates to be reported on the original scale (i.e., ratio of the
means), then you can do:

post_rma(trial, ~region+impact, type = "response")

Kind regards,
Reza

https://www.researchgate.net/publication/370656317_Meta-analysis_of_Second_Language_Research_with_Complex_Research_Designs


Reza Norouzian (he/him/his)
Senior Researcher for Multilingual Learners | Oregon Department of Education


On Mon, May 22, 2023 at 2:19 PM Slein, Margaret <maslein using zoology.ubc.ca> wrote:
>
> Hi Reza,
> Thank you for the speedy reply, I really appreciate it. I have tried running your code from the second link you sent, but I have no luck in running the emmprep function --> am I missing something?
>
> #install the dev version of metafor
> #install.packages("emmeans")
> #install.packages("remotes")
> #remotes::install_github("wviechtb/metafor")
> #remotes::install_github("rvlenth/emmeans", dependencies = TRUE, build_opts = "")
>
> library(metafor)
> library(emmeans)
>
> trial <- rma.mv(yi, vi, mods = ~region + impact, method="ML", random = ~1|study_id, data=heatdome_ROM, dfs = "contain")
> grd <- metafor::emmprep(trial) #this does not run
> grd <- emmeans::emmprep(trial) #this does not run
> grd <- emmprep(trial) #this does not run
> emmeans(grd, ~region+impact)
>
> Any assistance with this would be greatly appreciated.
>
> Cheers,
> Maggie
>
>
> On 5/19/23, 6:39 PM, "Reza Norouzian" <rnorouzian using gmail.com <mailto:rnorouzian using gmail.com>> wrote:
>
>
> [CAUTION: Non-UBC Email]
>
>
> Dear Maggie,
>
>
> This question has come up with some frequency on the list. For
> example, in the archives
> (https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-April/subject.html <https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-April/subject.html>),
> I found the following post:
> https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-April/004552.html <https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-April/004552.html>
> which, I believe, exactly answers your questions.
>
>
> Kind regards,
> Reza
>
>
>
>
>
>
> On Fri, May 19, 2023 at 6:49 PM Slein, Margaret via
> R-sig-meta-analysis <r-sig-meta-analysis using r-project.org <mailto:r-sig-meta-analysis using r-project.org>> wrote:
> >
> > Hello,
> >
> > Currently working on a meta-analysis and wondering about how to interpret the intercept in a rma.mv output with lnRR as the effect size and several binary categorical moderators? When I put in the -1 to remove the intercept, it doesn’t provide the full estimates for levels across all categorical moderators? How do I get the model output to provide estimates for all levels of all categorical moderators – is there an additional argument for this?
> >
> > Any help would be greatly appreciated.
> >
> > Cheers,
> > Maggie
> >
> > [[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>
> > To manage your subscription to this mailing list, go to:
> > https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis <https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis>
>
>
>