Hi
I am doing a survival analysis and have run into a couple of things that I was
hoping I could get some advice on.
The first thing is that when I run an ordinary Cox regression in R I get the
same results that I do using Stata (provided that I specify the Efron method for
handling ties) but I also get the following warning message:
"Warning message:
In coxph(Surv1 ~ Out + site + haem) :
X matrix deemed to be singular; "
Is this warning important and definitely applicable? I would have thought that
if there was a problem with the data I would also get some sort of warning in
Stata?
The second problem concerns the estimation of a weighted Cox regression i.e I am
using coxph with the weights option (I have calculated the weights in Stata).
Basically I am getting the following error message:
"Error in fitter(X, Y, strats, offset, init, control, weights = weights, :
NA/NaN/Inf in foreign function call (arg 10)
In addition: Warning message:
In fitter(X, Y, strats, offset, init, control, weights = weights, :
Ran out of iterations and did not converge"
Is there anything I can do to try and get around this? I tried reducing the
model from one that initially had 7 parameters to one with just one but kept on
getting the same message.
BTW I have run the same commands on a different outcome variable where I did not
get either of these problems so I am aware that they don't always appear and
are related to the data.
Thanks very much
Greg
[[alternative HTML version deleted]]
