Page 1 of 1
Correlation Coefficients for a Meta-analysis
Posted: Wed Jan 04, 2023 6:30 am
by maymay
Hi,
I use the option of Correlation Coefficients for a Meta-analysis.
Unfortunately, I noticed that the z confidence intervals in the forest plot are a little different from what I get by manual calculation.
I use Model measures: Fisher's r-to-z option to get the z.
I was wondering is the formula that the software use different from the one I know.
The formula I use is:
[1/(n-3)]^0.5= SE
minZ = r-1.96SE
maxZ = r+1.96SE
Thank you
Re: Correlation Coefficients for a Meta-analysis
Posted: Wed Jan 04, 2023 6:59 pm
by MAgojam
Hey @maymay,
yes the formula is slightly different than yours, but I think the attached screenshot is self-explanatory and satisfies your question.
- MAJOR_correlation_coefficient.png (267.85 KiB) Viewed 1522 times
Cheers,
Maurizio
Re: Correlation Coefficients for a Meta-analysis
Posted: Wed Jan 04, 2023 7:35 pm
by maymay
Hey Maurizio,
Thank you very much!!
Can you please let me know how can I see this code for next time?
Re: Correlation Coefficients for a Meta-analysis
Posted: Thu Jan 05, 2023 12:43 am
by MAgojam
maymay wrote: ↑Wed Jan 04, 2023 7:35 pm
Can you please let me know how can I see this code for next time?
Hey @maymay,
the R code at the bottom right is not possible in MAJOR, but I put it (RStudio console) as a step of the calculation to get the confidence intervals provided by the module.
But I am attaching the jamovi file, where with a few lines of R code in Rj editor+, 4 new variables related to
rtoz,
se,
cil and
ciu can be saved in the data sheet of the file for a comparison with the forest plot values.
If you want, you can try on your data, simply replacing the name
rho in
data$rho and
n in
data$n with that of your variables for correlations and sample size.
Cheers,
Maurizio
Re: Correlation Coefficients for a Meta-analysis
Posted: Tue Jan 10, 2023 8:28 am
by maymay
Thank you very much