Statalist The Stata Listserver


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

Re: st: comparing model with gllamm


From   Joseph Coveney <jcoveney@bigplanet.com>
To   Statalist <statalist@hsphsun2.harvard.edu>
Subject   Re: st: comparing model with gllamm
Date   Fri, 24 Nov 2006 10:27:31 +0900

Sami wrote:

I would like to compare between two  non nested models, I am using gllamm
with a continuous  outcome and a mixture of discrete and continuous
independent variables.

Model 1: includes a variable X1 as continuous
Model2: includes a decrete version of X1

--------------------------------------------------------------------------------

You might have a particular reason for discretizing the continuous variable,
X1, but in general this tactic is not recommended.  See

http://biostat.mc.vanderbilt.edu/twiki/bin/view/Main/CatContinuous?
CGISESSID=eff272349c15d71a3755a889a11b79fe

for a list of reasons not to, and for an alternative to it.  (I put a space
in the URLafter the question mark to allow wrapping.  Patch the URL back
together without the space in your browser.)

Comparison of nonnested models often will include use of something like
Akaike's Information Criterion or Schwarz's Information Criterion (-findit
akaike- or -findit schwartz-).  If you've just dichotomized X1, then you
have the same degrees of freedom, and you could just compare the log
likelihoods directly.  Other statistical and nonstatistical considerations
would probably outweigh the relative magnitude of two numbers, however.

Joseph Coveney

*
*   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/



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