Skip to content
Prev 4680 / 5636 Next

[R-meta] Meta-analysis of R^2 Values

Dear Paul,

Correct - this is for standard R-square values, not adjusted ones. The number of predictors (mi) is needed for computing the sampling variances of the R^2 values (or for converting F-statistics or the corresponding p-values to R^2).

However, this approach to meta-analyzing R^2 values assumes that the sample sizes of the studies are sufficiently large for the difference between the standard and adjusted R-square values to be negligible anyway.

Would be great to have a real dataset, maybe as a contribution to the 'metadat' package (https://wviechtb.github.io/metadat/).

Best,
Wolfgang