Hi all, thanks for reading! I'm running discrete-time survival analysis and trying to plot the step-shaped cumulative survival functions (s) after a logit model.
xtlogit event duration, i(ID)
I basically followed what online resources told me to do and run the following commands:
predict hazard, pu0
bysort ID duration: gen hazard_cumulative = sum(1-hazard)
gen s = 1 - hazard_cumulative
twoway line s duration, sort ylab(0(0.1)1) xlab(1(1)5) connect(J)

Somehow there occurs 2 problems:
(1) At each point of connecting the values, the data points of individual observations occur on the x axis, no matter if I specify line/connect. Array


(2) Even if the values of cumulative survival functions are descending across duration(which I checked with tabulate and browse), somehow the shape is not always descending.

Would anyone know the solutions to these problems? Many, many thanks indeed!