I Squared Values
Could anyone inform how JASP calculates the I^2 values in meta-analyses? For example, the Q value in the attached meta-analysis is 5.9 and N is 6. I^2 would be:
=𝑄−(𝑛−1)/𝑄
= (5.9 - 5)/5.9
=15.3
How does JASP Calculate 25.6%?
Thanks!
Ravi
Comments
Hi Ravi,
The relevant blog post mentioned in the help file (https://jasp-stats.org/2017/11/15/meta-analysis-jasp/) says:
"(For instance, Ι2 expresses the excess variance as a percentage of the total variance observed in the effect size estimates across studies, and hence is akin an intraclass correlation; for the other measures, please consult the meta-analysis literature.)"
Google tells me that the definition is I² = 100% x (Q-df)/Q. So yeah, I don't immediately see how JASP gets the I^2. I'll ask our experts! (and we should really get the help file done)
Cheers,
E.J.
Thanks!
Ravi