Thanks, Joseph. I have the text to which you refer, but I couldn't find a way to formally test it in -xtreg-, just -bootcor-. I'll give your approach a try.
-p
-----Original Message-----
From: [email protected] on behalf of Joseph Coveney
Sent: Tue 3/7/2006 2:07 AM
To: Statalist
Subject: Re: st: rho in xtreg
Paul Visintainer wrote:
-xtreg- produces "rho" which can be interpreted as a reliability
coefficient (e.g., the intraclass correlation). If one adds another
covariate (e.g., gender), rho is "adjusted" for gender. Is there a way
in -xtreg- to test whether rho differs by level of the covariate? That
is, whether the rho for males differs from the rho for females?
--------------------------------------------------------------------------------
If I understand correctly, you can model separate residual variances and
then use a likelihood ratio test to see whether this adds anything to a
model that doesn't allow for residual variance to differ between sexes.
This can be done using -gllamm- (see S. Rabe-Hesketh and A. Skrondal,
_Multilevel and Longitudinal Modeling Using Stata_ (College Station, Texas:
Stata Press, 2005), pp. 92-94), but as far as I know not yet with -xtmixed-.
Joseph Coveney
sysuse bplong
tabulate sex, generate(sex_)
eq het: sex_1 sex_2
xi: gllamm bp i.when sex, i(patient) s(het) adapt
estimates store Het
xi: gllamm bp i.when sex, i(patient) adapt
lrtest Het .
exit
*
* 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/
<<winmail.dat>>
