10-21-2013 11:59 AM
Hello!! I have a SAS/STATA question if anyone can help!! SAS considers you at risk for the event just before you have the event, but STATA considers you at risk at the time you have the event - WHY the difference?!?
10-22-2013 11:04 PM
It's arbitrary. The Cox Proportional Hazards Model is based on the assumption that time is a continuous variable and the hazard is a limit as delta-t approaches 0 ( SAS/STAT(R) 12.3 User's Guide ). However, our ability to measure time is much more granular, so the software developers had to make a choice for the ATRISK table and the Stata and SAS folk made different choices.
If your data are more granular (e.g. time in days for disease survival), the estimates are nearly the same whichever way the developers chose. If your data are less granular, you have lots of ties and the choice may have some impact on the estimates (an educated guess, I've not done a formal comparison of SAS and Stata).