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

st: multi-level modelling


From   "McElroy, Brendan" <BMcElroy@economics.ucc.ie>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: multi-level modelling
Date   Thu, 24 Oct 2002 16:02:08 +0100

Dear stata-list users,

I'd appreciate a piece of advice.  I've got a two-level hierarchical
dataset: patients at the lower level and GPs at the higher level.  Say I
want to assess the relationship between medical expenditures and patient age
controlling for GP years of training as follows:

Expenditures = f(PatientAge, GPtraining)

I can use cluster robust standard errors, to relax the assumption of
independence of observations on a GP patient list:

reg Expenditures PatientAge GPtraining, cluster(gpid)

If I do so, is it valid to comment on the coefficient on the GPtraining
variable in the normal way?

If not, I could use fixed effects specification:

xtreg Expenditures PatientAge GPtraining, fe i(gpid)

But when I do this, the GPtraining variable gets dropped, so instead I've
done this:
xi i.gpid

reg Expenditures PatientAge GPtraining I_gpid_2 - Igpid_455

This seems to work. However I can't use xthausman to compare a random
effects and a fixed effects specification.  Can I modify xtreg, fe in some
way so that it drops one GP dummy rather than all the GP-level variables, so
that I can apply xthausman?

Brendan McElroy
University College Cork
Western Road
Cork
Ireland
Tel: +353 21 490 3522

*
*   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