Dear Xiaoyi,
You can run a competing risks model using STCOX and the data augmentation
technique of Lunn and McNeil (1995). For each risk, you replicate the data
so for your 3 risks you would have 3 repeated records. You use a dummy
variable to assign each set of records to one of the three risks and
identify failure only for the appropriate risk in the appropriate risk set.
After checking for proportionality in the dummy variable for risk type you
have the option of adding risk type as a covariate (if proportional) or
stratifying the model by risk type (if non-proportional) and interacting
risk type with your covariates to observe differences in covariate effects
across risk types. The competing risks formulation with Cox provides
efficient estimates of the failure rate for each type of risk as well as
robust coefficient effects. Because you have repeated observations in your
sample you use the cluster option to obtain robust standard errors.
Feel free to contact me directly ([email protected]) if you have any
questions.
Jacqui
Date: Wed, 20 Apr 2005 08:43:21 -0700 (PDT)
From: xiao yi <[email protected]>
Subject: st: Competing risks in Stata
Hi,
Does anyone know how to run a competing risks model in
Stata? My data is about firms' investment behavior.
Once a firm enters my sample, it has three options:
invest, disinvest, or do nothing. In the language of
survival analysis, doing nothing would be survivor,
invest and disinvest would be two competing risks. I
expect the same independent variables would have
different impact on the probability of investment and
disinvestment. Thanks.
Xiaoyi
********************************************
Jacqueline L. Frair
Ph.D. Candidate
Department of Biological Sciences
University of Alberta
Edmonton, Alberta T6G 2E9
PHONE: 780-492-0083
FAX: 780-492-9234
EMAIL: [email protected]
********************************************
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
