Means are not different but significant


New Member

I have a query, generated from PrediXcan analysis but more related to statistics.

I have the two means of predicted expression values for geneA: 0.724868 (controls) and 0.700837 (cases).

But the P-values reported in association file form PrediXcan is 1.56132574034295e-22 (controls) and 7.41715987764052e-16 (cases), and if we combine p-values of both cases and controls, the result is 9.964981e-36

My question is if we have two means very close to each other, could still they are significant??

My apologies if you feel this as a naive question....


Looking forward for your responses.



Ambassador to the humans
What tests are those p-values associated with? I think you need to take a step back and identify what question is being answered by each p-value in your problem.