[R-meta] Computing treatment effects with 95%CI from a meta-regression output

[James Pustejovsky]

That makes sense. RVE won't work and is not appropriate if you only have a
single study falling into a given category. For it to work, there need to
be multiple independent studies that contribute effects to a given
category.

On Sun, Sep 6, 2020 at 10:22 AM Tobias Saueressig <t.saueressig using gmx.de>
wrote:

> Hello James,
>
> thank you. It worked. In cases were there was only a single study it gave
> alot of NaNs. But I have the CIs from the analysis with the intercept in
> it.
>
> Regards
>
> Tobias
>
>
>
> *Gesendet:* Sonntag, 06. September 2020 um 16:53 Uhr
> *Von:* "James Pustejovsky" <jepusto using gmail.com>
> *An:* "Tobias Saueressig" <t.saueressig using gmx.de>
> *Cc:* "r-sig-meta-analysisr-project.org" <
> r-sig-meta-analysis using r-project.org>
> *Betreff:* Re: [R-meta] Computing treatment effects with 95%CI from a
> meta-regression output
> The easiest thing to do is probably to simply recalculate the model after
> omitting the intercept term:
>
> robu(formula = es ~ 0 + Time, data = SMD,studynum = study, var.eff.size =
> var,rho = .8, small = TRUE, modelweights = "CORR")
>
> Sent from my iPhone
>
> > On Sep 6, 2020, at 7:25 AM, Tobias Saueressig <t.saueressig using gmx.de>
> wrote:
> >
> > robu(formula = es ~ 1 + Time, data = SMD,studynum = study, var.eff.size
> = var,rho = .8, small = TRUE, modelweights = "CORR")
>

	[[alternative HTML version deleted]]