Double arcsine transform not appropriate for meta-analysis

The double arcsine transform is a variance stabilizing transformation, i.e., the variance of transformed deviates is (at least approximately) independent of the underlying binomial probability.2 Differing conventions are common regarding the overall scaling; above, a factor 1 2 was applied, but other definitions are also used.1,2,3 It is worth also pointing out the closely related arcsine transform here, which is defined as