Statalist


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

st: RE: Re: Predicting Random Effects from a Crossed-Level Model using xtmelogit


From   "Lee, Albert" <Albert.Lee@hud.gov>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Re: Predicting Random Effects from a Crossed-Level Model using xtmelogit
Date   Tue, 29 Dec 2009 08:15:32 -0500

Joseph,

Thanks.  Yes, and no.  Yes in that xtmelogit for crossed-level model uses Laplace approximation.  However, when I tried specifying the intpoints option, STATA generates an error message as below:

. xtmelogit fail refinance || _all:R.year || _all:R.hub || _all:R.originator_n, intpoints(5 5 5)
wrong number of intpoints() specifications
r(198);

Could you tell me what went wrong?

Thanks,

Albert.

________________________________________
From: owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu] On Behalf Of Joseph Coveney [jcoveney@bigplanet.com]
Sent: Tuesday, December 29, 2009 1:13 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: Re: Predicting Random Effects from a Crossed-Level Model using xtmelogit

Albert Lee wrote:

This may be obvious.  If so, I apologize in advance.

I'm estimating a logistic model with crossed-level random effects.  The outcome
binary variable is fail with one fixed-effect independent binary variable
refinance.  The three crossed-level random effects are VAR1, VAR2 and VAR3.  The
STATA command I used is as below:

xtmelogit fail refinance || _all:R.VAR1 || _all:R.VAR2 || _all:R.VAR3

After estimating this model, I tried to recover the random effects of VAR3
using:

predict u, reffects level(VAR3)

However, u only contains missing values.

I would very much appreciate if someone can help me recovering these predicted
random effects.

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

Doesn't -xtmelogit- use Laplace approximation by default when fitting a
cross-classified model?  I believe that you would need to specify at least
three, or so, integration points (abscissa weights) in order to get empirical
Bayes predictions of individual random effects.

Joseph Coveney



*
*   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/

*
*   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   |   What's new   |   Site index