I frequently use the your survival package and I have recently updated it and have seen you have had changed something in the way you take into account NA coeff in the cox.zph function (since update 3.1-12).
I think, there could be issues with following lines :
186 else temp[j, i] <- fit$coefficients[j]
and
212 if (length(j) ==1) sresid[,i] <- sresid[,i] + fit$coefficients[j]
because j is an asgn value which was modified at line 71 if there are NA in coef in fit object.
And so the new asgn indicies don't correspond anymore with the original ones in fit$coefficients
Hello,
Firstly, I thank you for your great work.
I frequently use the your survival package and I have recently updated it and have seen you have had changed something in the way you take into account NA coeff in the
cox.zphfunction (since update 3.1-12).I think, there could be issues with following lines :
and
because
jis anasgnvalue which was modified at line 71 if there are NA in coef infitobject. And so the newasgnindicies don't correspond anymore with the original ones infit$coefficientsSorry, I don't have any example to illustrate it.
Am I right ?
Best,