Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Cox Regression - flat region resulting in a missing likelihood


From   graupel75@gmail.com
To   Statalist <statalist@hsphsun2.harvard.edu>
Subject   st: Cox Regression - flat region resulting in a missing likelihood
Date   Tue, 30 Apr 2013 08:04:44 +0800

Theresa writes:

> I set my data stset and then I tried to run a stcox and got the following
> error message " flat region resulting in a missing likelihood r(430)";

I would first try to change the settings for ties (Efron, exactp,
exactm) if you have many tied failure times.

Beyond that, it likely due to so-called 'mono-tone' likelihood;
there's a solution called the Firth method. Unfortunately, it is not
implemented in Stata for Cox models (but is in Some Alternative
Software and in R).

You could also try to randomize your times slightly (which likely
won't work); use flexible parametric models which will require some
startup (-ssc install stpm2-); or do a Poisson regression to get
(essentially) the same coefficients as Incident Rate Ratios instead of
HRs. Best source of how to do this is Dupont's 'Statistical Modeling
for Biomedical Researchers'.

cheers-
Andrew Lover
______________________________________________
Epidemiologist
Centre for Infectious Disease Epidemiology Research (CIDER)
Saw Swee Hock School of Public Health
National University of Singapore
*
*   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/


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