Dear Dr. Therneau,
Thank you very much for your exhaustive answer!
I now see the issue. Perhaps even more important was your confirmation that my
approach with karno:ns(time, df=4) is theoretically correct. (I knew that
plot.cox.zph is sound, so I was afraid that the difference can be attribute
This is an excellent question.
The answer, in this particular case, mostly has to do with the outlier time
values. (I've
never been convinced that the death at time 999 isn't really a misplaced code
for
"missing", actually). If you change the knots used by the spline you can get
quite
dif
Dear All,
I was thinking of two possible ways to plot a time-varying coefficient
in a Cox model.
One is simply to use survival::plot.cox.zph which directly produces a
beta(t) vs t diagram.
The other is to transform the dataset to counting process format and
manually include an interaction with t
3 matches
Mail list logo
