[R-meta] Calculate individual SDs from ES's SDs
Michael Dewey
lists at dewey.myzen.co.uk
Tue Feb 27 11:00:52 CET 2018
Dear Cesar
I think you have to tell the journal that this is not possible and
provide the data-set with NA for the unavailable fields.
Michael
On 26/02/2018 15:04, Cesar Terrer Moreno wrote:
> I currently have a dataset of 300 observations structures as MeanControl, MeanTreated, SEControl, SETreated, Ncontrol, Ntreated, and I calculate ES and VAR through metafor:escalc.
>
>
>
> Now, I found a few observations that provide only ES and VAR, which I would like to add to the original dataset.
>
>
>
> Obviously I could include these new observations ad-hoc to the post-escalc dataset, but journal suggests providing the raw data with the above fields as MeanControl, MeanTreated, etc so all observations are in the same format. For this reason, I want to back-transform from ES-VAR to MEAN-SE.
>
>
>
> C
>
>
>
> De: James Pustejovsky [mailto:jepusto at gmail.com]
> Enviado el: 26 February 2018 15:56
> Para: Cesar Terrer Moreno <cesar.terrer at me.com>
> CC: r-sig-meta-analysis at r-project.org
> Asunto: Re: [R-meta] Calculate individual SDs from ES's SDs
>
>
>
> No, that's not possible. But why do you need to do this? Couldn't you just use the ES and VAR directly?
>
>
>
> On Mon, Feb 26, 2018 at 8:53 AM, Cesar Terrer Moreno <cesar.terrer at me.com <mailto:cesar.terrer at me.com> > wrote:
>
> Dear James,
>
>
>
> I believe you are right, 1.38 is the effect size, and 0.07 the error of the effect size.
>
>
>
> My question is whether I can compute the standard error of the control and treated groups having the effect size’s SE/SD.
>
> ES = 1.38
>
> VAR = 0.07^2
>
>
>
> MeanControl = 1
>
> MeanTreated = 1.38
>
> SEControl = ?
>
> SETreated = ?
>
>
>
> De: James Pustejovsky [mailto:jepusto at gmail.com <mailto:jepusto at gmail.com> ]
> Enviado el: 26 February 2018 15:45
> Para: Cesar Terrer Moreno <cesar.terrer at me.com <mailto:cesar.terrer at me.com> >
> CC: r-sig-meta-analysis at r-project.org <mailto:r-sig-meta-analysis at r-project.org>
> Asunto: Re: [R-meta] Calculate individual SDs from ES's SDs
>
>
>
> I would hazard a guess that the +/- 0.07, which is described as "SD", might actually be the standard error of the effect size estimate. So in the example, ES = 1.38 and sampling variance = 0.07^2 = 0.049. Perhaps you can verify this speculation using other information in the source document.
>
>
>
> James
>
>
>
> On Wed, Feb 21, 2018 at 6:41 AM, Cesar Terrer Moreno <cesar.terrer at me.com <mailto:cesar.terrer at me.com> > wrote:
>
> Dear all,
>
> I have a series of effects sizes, expressed as e.g.:
>
> 1.38 ± 0.07 (MEANexp/MEANcontrol, SD, n=3)
>
> I have to express these experiments as MEANexp, SDexp, MEANcontrol, SDcontrol in order to put all data in a correct formar that can be read by escalc in metafor.
>
> I guess I can assume MEANexp = 1.38 and MEANcontrol = 1, but how can I compute SDs so that escalc computes the effect size’s variance properly?
>
> Thanks
> César
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis at r-project.org <mailto:R-sig-meta-analysis at r-project.org>
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>
>
>
>
>
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>
--
Michael
http://www.dewey.myzen.co.uk/home.html
More information about the R-sig-meta-analysis
mailing list