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]

Re: st: How to tackle the r1400 error message


From   rgutierrez@stata.com (Roberto G. Gutierrez, StataCorp)
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: How to tackle the r1400 error message
Date   Thu, 26 Aug 2010 09:54:01 -0500

Nicola Man <nman1011@gmail.com> asks:

> I am running into problems with numerical overflow according to the r(1400)
> error message, but really wanted to fit the multi-level model I am fitting.
> It was working fine with the model fitting only random effects and
> intercept.

> When I added a few other fixed effects, it still managed to chuck out
> something (though not very sensible) when I ran xtmelogit with intpoints(1).
> I then tried subsequently to fit with higher number of integration points.
> This is when I start getting "initial values not feasible" with r(1400).

Maarten Buis <maartenbuis@yahoo.co.uk> offers excellent advice on rescaling
your variables and on ensuring that baseline is reasonable for your model.

I would add that we've also improved the random-effects estimation strategy
that -xtmelogit- uses as a step toward calculating the log-likelihood.  This
addition was made in the 11feb2010 ado update of Stata 11, and a side effect
of this refinement is a reduced incidence of the r(1400) error.

If your Stata 11 is fully up-to-date and the model persists, feel free to
email me with the data and I can take a closer look.

As Nicola suspects, moving from IC to SE or MP would not change anything
concerning the above.

--Bobby
rgutierrez@stata.com
*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index