Statalist


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

Re: st: xtmelogit - a very strange treatment of different levels


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: xtmelogit - a very strange treatment of different levels
Date   Fri, 12 Feb 2010 16:05:20 -0600

Make sure you understand the scaling issues with multilevel logit models;
see http://www.citeulike.org/user/ctacmo/article/3057661. Not all
comparisons will be meaningful when you introduce an additional level.

On Fri, Feb 12, 2010 at 3:19 PM, Dimitrije Tišma
<dimitrijetisma@gmail.com>wrote:

> The command was a quite typical -xtmelogit-. Persons are nested within
> households, i.e. they do not change the HH number over time. As the
> first step I wanted only random intercepts for person and household
> level, so I wrote the following:
>
> xi: xtmelogit depvar indepvars || personID: || householdID:
>
> I did not save the results, I am afraid, but I remember well what was
> puzzling. The group statistics was strange to me, as it showed the
> same number of units on both levels. Obviously, it considered the
> number of person-household combinations as number of households and I
> don't know why. Furthermore, while when using -gllamm- the results
> change considerably by introducing another level (household), here
> they were practically identical to those when using logit. Thanks in
> advance for the answer.
>
>
>
>
>
>
>
>
> 2010/2/12 Airey, David C <david.airey@vanderbilt.edu>:
> > .
> >
> > Show your commands and results.
> >
> >> Dear all,
> >>
> >> I am working with some household panel dataset and in order to check
> >> for random effects I did both gllamm and xtmelogit regression. While
> >> in gllamm everything worked fine, xtmelogit did not deal with the two
> >> levels (individuals and households) the way I expected. Namely, it
> >> reports the same number of units on both levels (!) and practically no
> >> random effects either level. Also, the results are practically
> >> identical to the results when "normal" logit regression is used. Does
> >> anyone know what this could be about? Thanks a lot.
> >>
> >> Dimitrije
> >
> > *
> > *   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/
>



-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   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