[R-meta] unsigned effect sizes

[Daniel Noble]

Hi James,

A great question! Yes, this is something quite common in ecology and evolution meta-analyses. There are some nice papers you may want to consult by Mike Morrissey. Here are a few references.

Morrissey, M. B. (2016). Meta-analysis of magnitudes, differences and variation in evolutionary parameters. J. Evol. Biol. 29, 1862-1904.

Morrissey, M. B. (2016).. Rejoinder: Further considerations for meta-analysis of transformed quantities such as absolute values. J. Evol. Biol. , 29: 1922–1931.

Basically it involves transforming ‘post-analysis’ to a folded normal. I’ve used it myself in a Bayesian context. It’s quite easy to do there. Hope this is useful and what you’re thinking about.

Cheers,
Dan
_______________

Dr. Daniel Noble 
Senior Lecturer

Division of Ecology and Evolution (office W317),
Research School of Biology,
College of Science,
Robertson Building, 
46 Sullivans Creek Road,
The Australian National University
Canberra, ACT 2600
Australia

 T +61 02 6125 0423
M +61 430 290 053
daniel.noble using anu.edu.au
Noble Lab Webpage: www.nobledan.com
RSB: https://biology.anu.edu.au

CRICOS Provider #00120C



> On 12 Feb 2022, at 12:24 pm, James Pustejovsky <jepusto using gmail.com> wrote:
> 
> Hi meta-analysis folks,
> 
> I have a kind of vague question about something I've run across a few
> times. There are some (perhaps rare) situations where investigators are
> interested in the absolute magnitude of an effect but where the sign or
> direction of the effect is arbitrary or not meaningful. Consequently,
> meta-analysts of such effects might like to work with _unsigned_ effect
> size estimates rather than the estimates that describe both magnitude and
> direction. However, taking the absolute value of an estimate changes its
> sampling distribution--potentially quite drastically!--in a way that would
> make conventional meta-analytic models
> (fixed/common/random/multi-level/multi-variate) perform rather poorly.
> 
> Does anyone know of work on methods for synthesis of unsigned effects, that
> actually account for the consequences of using absolute effect size
> estimates?
> 
> James
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://stat.ethz.ch/pipermail/r-sig-meta-analysis/attachments/20220212/4a2197af/attachment-0001.html>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 25069 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-meta-analysis/attachments/20220212/4a2197af/attachment-0001.png>