I am running a meta analysis with metal https://genome.sph.umich.edu/wiki/METAL_Documentation and I am trying to figure out the exact differences and implications of the different analysis schemes, especially Scheme STDERR and Scheme SAMPLESIZE. The first one uses effect size estimates and standard errors, the latter one uses p-values and direction of effect, weighted according to sample-size. I am getting quite different results for both. Does anybody know when one should expect similar results for both analyses, when what should be used and what the exact differences are?
Thanks a lot!