|
Post by fernand0andrade on Dec 12, 2015 16:09:58 GMT
Hello,
I am doing some bivariate analyses on some simulated data to learn how to use the software and I came across one question. When I try to do bivariate analysis and testing aginst genetic correlation == 0 I get a p-value for this test. My question is, as I´m testing for a correlation == 0, shouldn´t the test be for the 2 tails? (the output file writes "one-tailed test" on the side of the p-value)
Thanks for the attention
Fernando
ps. I´m using GCTA 1.24
|
|
|
Post by Jian Yang on Dec 13, 2015 7:46:36 GMT
That's right. It should be a two-tailed test in this case. Two-tailed test p-value = 2 * one-tailed test p-value. The reason why GCTA reports one-tailed test p-value is because if it's a test against rg = 1 and rg estimate is constrained between -1 and 1, it will be an one-tailed test.
|
|
|
Post by fernand0andrade on Dec 14, 2015 15:02:15 GMT
Thanks for your help
|
|