Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Second Repost "Simple Cases of Multi-Level Models"


From   SamL <saml@demog.berkeley.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Second Repost "Simple Cases of Multi-Level Models"
Date   Mon, 10 Jun 2002 07:40:56 -0700 (PDT)

Thanks for your response.  However, the time-point issue seemed to confuse
matters.  It seemed that on that basis alone you suggested I use GLLAMM.  
But, with all due respect, this suggestion side-steps my question; can one
use the XT series for a two-level model if one wants to explain the
variation with additional context-level covariates?  To focus the question
on this aspect, I deleted the time references to simplify the question.  
Indeed, I can also do so in the analysis; I'll avoid a lengthy substantive
exposition here of why this is not a bad idea for my research question,
but if you wanna know just send me e-mail.

To avoid pestering Rich Goldstein I sent the message to the list, in an
effort to overturn the "No good deed goes unpunished" rule. The
simplification leaves only two-levels.  Can I use the XT-series to do
multi-level models as the presentation seemed to suggest?

Thanks a bunch.
Sam

On Mon, 10 Jun 2002, Richard Goldstein wrote:

> Since the quote is from a talk I gave, I note to the list that in
> private
> email, prior to Sam writing to Statalist, I told him that I thought he
> had a 3-level model (there are 2 time points).
> 
> Rich Goldstein
> 
> SamL wrote:
> > 
> > My question concerns the XT series of models in STATA.  My question is
> > whether the XT series allows one to treat second-level variation as
> > something to study, rather than simply as a nuisance to be controlled.
> > Given my interest in the former, I read with interest the
> > presentation/paper titled "Simple Cases of Multi-Level Models" by Richard
> > Goldstein at a STATA Users Group Conference, which seemed to suggest that
> > the XT series may meet my needs.  I presume attending would have clarified
> > my question, but given that I was not there I have a question.
> > 
> > My question probably flows from the many different meanings that are
> > floating around for the terms that are in use in different disciplines.
> > I am specifically interested in the statement on a slide from the talk,
> > that reads "I know that the following is true for at least some of the XT
> > models and I believe it is true for all of them: they can be used to
> > analyze any two-level random intercept model."
> > 
> > My aim is to estimate a model of the following form:
> > 
> > Pr(Y_ij=1) = b1_j White1_ij + b2_j Black1_ij + b3 X1_ij + b4 X2_ij + e_ij
> > b1_j = g01 + g11 Z1_j + g21 Z2_j + d1
> > b2_j = g02 + g12 Z1_j + g22 Z2_j + d2
> > b3_j = g03
> > b4_j = g04
> > 
> > In this model the Zs are measured at the higher level; there are a few
> > dozen higher-level entities and lots and lots of lower-level units.  The
> > aim is to estimate the "effect" of Zs on the black and white intercepts.
> > 
> > That's the set-up.  The questions: can this be estimated using the "XT"
> > models in STATA?  And, if so, how?
> > 
> > Thanks for any references to GLLAMM, but I am familiar with GLLAMM.  Any
> > insights on this in terms of the XT series of models are greatly
> > appreciated.
> > 
> > Thanks a bunch.
> > Sam
> > 
> > *
> > *   For searches and help try:
> > *   http://www.stata.com/support/faqs/res/findit.html
> > *   http://www.stata.com/support/statalist/faq
> > *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
> 

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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