Hi everyone, Performed meta-analysis using METAL. I tried both "inverse variance weighted approach" and "sample weighted" approach. The effect directions for both the outputs match, but i see huge difference in the meta-p value between the two approaches. Its a Binary phenotype, total sample size >5000. In the attached image, blue highlighted is for inverse weighted, and yellow for sample weighted. Have been trying to find a reasonable explanation but no success so far. My expectation was that the p-values from the two approaches should be similar, but here i see huge difference though the effect directions are identical. Any explanation would be highly appreciated.

Thank you![enter image description here][1]

[1]