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

RE: st: Reading the output of xtmelogit

From   Maarten buis <>
Subject   RE: st: Reading the output of xtmelogit
Date   Wed, 21 Oct 2009 01:44:19 -0700 (PDT)

--- On Tue, 20/10/09, Jackson, Theron Keith wrote:
> My DV is binary.  Would this be
> the reason why most my variance is zero.  Is there a
> way to correct for this or is both the county level and zip
> code level neighborhood disadvantage variable just not
> significant?

-xtmelogit- is designed for binary dependent variables, so
that is not a problem. Generally, the way to debug these models
is to start with a simple model: one level (country) and only
a random intercept and no random effect of the other covariates.
Make sure that that model works, add a random effect, make sure
that that model works, ad a level, etc. etc.

Is the neighbourhood disadvantage variable constant over zip code?
If that is the case than it probably makes no sense to make the
effect of that variable random at the zip code level, as there
is no variation at that level. You can still make that effect 
random at the country level.

Hope this helps,

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

Send instant messages to your online friends 

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index