Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Checking Proportional Hazards Assumption in Multiply Imputed Data
From
Richard Williams <[email protected]>
To
[email protected], [email protected]
Subject
Re: st: Checking Proportional Hazards Assumption in Multiply Imputed Data
Date
Wed, 21 Aug 2013 09:58:02 -0500
At 05:38 PM 8/20/2013, Steve Samuels wrote:
1. In your -mi estimate: stcox- command, add an interaction of a
predictor with time (tvc() and texp() options), and test the
significance of the coefficient.
2. -mi extract- a few imputated data sets; run (non-mi) -stcox-; then
run -estat phtest- with the "plot" option. The purpose here is informal
diagnosis, so just ignore the fact that the data were imputed.
I like the idea of informal diagnoses when there doesn't seem to be a
way to do a perfect test with mi. I mean, it is not like a diagnostic
test has to be accurate to 12 decimal places. If you are lucky, you
will either clearly see that there is a problem or there isn't. It
will just be in borderline cases or cases where the imputations give
conflicting results that you have to decide which way to go.
-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME: (574)289-5227
EMAIL: [email protected]
WWW: http://www.nd.edu/~rwilliam
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/