Congrats on the publication of the package, I'm looking forward to testing it out.
While working with the Python version, I found that on line 275 of lmm.py you are doubling the p-value twice, resulting in p-values > 1:
pval = 2 * t_dist.sf(np.abs(tval), df=n - p) * 2
It seems to be correct in the R version:
pval <- 2 * pt(-abs(tval), df = n-p)
Congrats on the publication of the package, I'm looking forward to testing it out.
While working with the Python version, I found that on line 275 of lmm.py you are doubling the p-value twice, resulting in p-values > 1:
It seems to be correct in the R version: