Is there any work-around for getting 95% point-wise confidence intervals to
show up on a survivor function graph that's been adjusted?
There are two ways I see of creating this plot:
Approach #1:
Take the variables you want to use in the hazard model and make new ones
such that when they equal 0, this corresponds to the actual value you want
them adjusted to. (eg. gen age50 = age - 50) Then perform -sts graph,
adjustfor(varlist)-. There's no current way to get confidence intervals on
this plot.
Approach #2:
Perform -stcox varlist, basesurv(newvar)-, then plot it using -stcurve,
survival at(varname=# varname2=#...)-. Again, the plot doesn't allow a CI
band.
Any ideas? Assume my only choice is to manually create these three
variables and then -twoway line- graph them? It's somewhat aggravating
because I know SAS does this with a straight-forward option, but I'd like
to do everything in Stata (ie. I dislike learning too many stat programs).
