[R-meta] Escalc function: Possible to compute single standardized mean as outcomes?

[Sera Wiechert]

Dear Michael,

Thanks for your thoughts on this. Most of them do report their standard deviation/error or CIs, so that should be no problem.

But for your main question; this is indeed something I thought about, too. But I am afraid that that will not be possible in most of the cases as the “word-pool” and thus, the denominator they use, is often not reported properly and many researchers seem to use different group totals to calculate their proportions.

Would there be, in this situation, a way to standardize the two in order to not be dependent on the denominator information?

Many thanks.
Best,
Sera

Von: Michael Dewey <lists using dewey.myzen.co.uk>
Datum: Mittwoch, 24. November 2021 um 17:27
An: Sera Wiechert <s.wiechert using uva.nl>, r-sig-meta-analysis using r-project.org <r-sig-meta-analysis using r-project.org>
Betreff: Re: [R-meta] Escalc function: Possible to compute single standardized mean as outcomes?
Dear Sera

Can I just clarify this

In the first group of studies each person recalls n words and the
authors calculated a proportion of possible words for each person and
then report the mean (and hopefully the standard deviation/error) of
those proportions.

In the second group the authors just report the mean of the number of
words again hopefully with a standard deviation/error.

Is it possible for the second group to impute the denominator which they
would have used if they had reported in the same way as the first group?
If they have then you can convert to mean proportions. Apologies if that
has already occurred to you but you have discarded it as impossible.

Michael

On 24/11/2021 10:03, Sera Wiechert wrote:
> Dear all,
>
> I would like to use the metafor package [escalc function] to calculate my effect size.
>
> I saw that the metafor escalc function measure = MN is for raw means. However, I am looking to calculate a "standardized mean". In my studies, only a single outcome is reported (�recalled number of words�). The problem is that sometimes this is reported as a proportion (e.g., percentage recalled words of all possible study words, percentage of recalled words of possible condition-specific words) and sometimes as raw numbers. Thus, I only have one quantitative outcome variable, which is sometimes operationalized in different scales across studies.
>
>
>
> This is why I had the idea to use some sort of �standardized mean� instead of the raw mean. Is this possible? And if so, is there an escalc function measure for this?
>
>
>
> Many thanks in advance.
>
> Best wishes,
>
> Sera
>
>
>
>
>        [[alternative HTML version deleted]]
>
>
>
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://eur04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstat.ethz.ch%2Fmailman%2Flistinfo%2Fr-sig-meta-analysis&data=04%7C01%7Cs.wiechert%40uva.nl%7Cd4a75b2fbbd34b78ce3b08d9af67387e%7Ca0f1cacd618c4403b94576fb3d6874e5%7C0%7C0%7C637733680423807743%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=LZz5D2aLew697HwCTeqotWRkDRCCqsqLDkA2t%2F1%2F4pM%3D&reserved=0
>

--
Michael
https://eur04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.dewey.myzen.co.uk%2Fhome.html&data=04%7C01%7Cs.wiechert%40uva.nl%7Cd4a75b2fbbd34b78ce3b08d9af67387e%7Ca0f1cacd618c4403b94576fb3d6874e5%7C0%7C0%7C637733680423807743%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=YH7VNeUCH6mhALPp%2Fo23zsZ2%2F7CF%2BC2BU%2FAdnzj8j6I%3D&reserved=0

	[[alternative HTML version deleted]]