Bookmark and Share

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]

st: Re: xtmixed does not converge after 12 hours


From   "Clyde Schechter" <clyde.schechter@einstein.yu.edu>
To   statalist@hsphsun2.harvard.edu
Subject   st: Re: xtmixed does not converge after 12 hours
Date   Mon, 18 Jul 2011 10:21:48 -0700

It may be that the data do not identify the parameters in your model:
starting with a simpler model that has fewer effects and then adding them
in (until you can't any more) may get you what you want.

A specific possibility is that one or more of your random effects is zero
or very close to zero.  This can lead to non-convergence, too.  It is
fairly easy to pick this up by re-running the commands with the -trace-
option.  At each iteration you will get a display of the current values of
all the parameters being estimated.  If you see one of the lnsigma
parameters marching down towards negative infinity, then you know that the
corresponding random effect is too small to converge to.  That will
usually be obvious after a small number of iterations, so it won't take
you very long to find out.  Removing that (those) specific random effects
may be all you need to get a convergent model.

Hope this helps.


Clyde Schechter
Department of Family & Social Medicine
Albert Einstein College of Medicine

Please note new e-mail address: clyde.schechter@einstein.yu.edu

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index