Post by Jian Yang on Aug 3, 2016 4:04:26 GMT
GREML p-value = 0?
This is a precision issue. This means that the p-value is extremely small. You can calculate a more precise p-value in R.
1) p-value = 0.5 * pchisq(LRT, df=1, lower.tail=FALSE) # one-tailed test, e.g. h2g is constrained to be positive in a GREML analysis.
2) p-value = pchisq(LRT, df=1, lower.tail=FALSE) # two-tailed test (recommended to test whether rg = 0 in a bivariate GREML analysis or to test if h2g = 0 in a unconstrained GREML analysis).
No LRT reported in *.hsq output file?
LRT ~= (estimate / SE)2
This is a precision issue. This means that the p-value is extremely small. You can calculate a more precise p-value in R.
1) p-value = 0.5 * pchisq(LRT, df=1, lower.tail=FALSE) # one-tailed test, e.g. h2g is constrained to be positive in a GREML analysis.
2) p-value = pchisq(LRT, df=1, lower.tail=FALSE) # two-tailed test (recommended to test whether rg = 0 in a bivariate GREML analysis or to test if h2g = 0 in a unconstrained GREML analysis).
No LRT reported in *.hsq output file?
LRT ~= (estimate / SE)2