Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Stcurve and time interactions


From   "Michael Greig" <j.michael.greig@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Stcurve and time interactions
Date   Wed, 13 Aug 2008 13:22:02 -0500

I have estimated a Cox model and now want to use stcurve to produce
the survival function at various values of interest for the parameters
in my model so that I can report the probability of survival at a
particular point in time given the parameters that I assign.  The
problem I am facing is that, due to violations of the proportional
hazards assumption, I have interacted one of the variables in my model
(a variable that I want to simulate values of in stcurve) with
ln(time).  For example, I have a variable that describes the level of
fatalities (fatal) in conflict as a categorical variable that I
include in the model along with an interaction with time
(fatal*ln(time)).  If I want to use stcurve to generate the values of
the survival function when fatal=4, what is the most reasonable value
to assign for the fatal*ln(time) variable (ie; what value should I set
for time)?

I am grateful for any advice that anyone might have.

Best,
Michael Greig
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index