[R-meta] sample size correlations for meta-analysis

[Michael Dewey]

Dear Catia

When I review papers for journals I often find authors have excluded 
small studi where small usually means n<10. When I query this on the 
grounds that the variance will take care of the weighting I never get an 
evidence-based reply so I suspect it is just an arbitrary limit. Of 
course it is hard to pprove a negative and someone else may have a 
source justifying the practice.

Michael

On 15/07/2023 17:00, Catia Oliveira via R-sig-meta-analysis wrote:
> Dear all,
> 
> I hope this email finds you well.
> I was wondering if anyone has ever published or seen a published paper that
> reflects on the minimum sample size for correlations to be used in
> meta-analysis? We may need to look at the raw data to see if participants
> fit the inclusion criteria, leaving us with very small sample sizes, so we
> want to establish what would be too small to reject. I know we need to
> ensure that the sample size needs to be big enough to allow for estimating
> the correlation and its confidence interval, but is that enough?
> 
> Thank you!
> 
> Best wishes,
> 
> Catia
> 

-- 
Michael
http://www.dewey.myzen.co.uk/home.html