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]

Re: st: xtmelogit syntax


From   Maarten buis <[email protected]>
To   [email protected]
Subject   Re: st: xtmelogit syntax
Date   Mon, 12 Jul 2010 08:18:28 +0000 (GMT)

--- On Sun, 11/7/10, Robin Jeffries wrote:
> I'm also a little concerned because I've had to run it
> several different times due to other problems, and the
> same model seems to take a different number of iterations
> to achieve convergence each time, and it converges to a
> different log-likelihood value each time. I was not able
> to monitor the parameter estimates to see how they
> differed at all. Is this a concern?

If these are the same model than particularly the different
log-likelihoods is a reason of concern. There are a couple
of things that can make these models unstable, common ones
are some of the estimated variances are very small (close 
to zero). In that case you can best leave that level out
of your model. A variable in your model has the value 0, 
which lies way outside the range of your data variable. A
common example is year of birth? In that case you can best
subtract some meaningful value from this variable to ensure
that the constant in your model refers to some meaningful
point in your data: it is this constant that is made 
random in your models, so it tends to get hard to estimate
if that constant refers to a point that is way outside the
data.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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