I'm not sure about the answer to your question, but cluster generally
INCREASES the standard errors, unless you have negative intracluster
correlation, which is very rare, but not unheard of.
Bryan Sayer
Statistician, SSS Inc.
[email protected]
-----Original Message-----
From: DANIEL CARO [mailto:[email protected]]
Sent: Thursday, February 27, 2003 4:40 PM
To: [email protected]
Subject: st: regress with cluster for data with nested structure
Dear all,
I'm working with data with nested or hierarchical structure (two levels).
I've been using HLM software to take into account this structure. However,
I'm not interesting in varying the slopes or intercept across the level 2
units, so I would like to know if using the STATA's regress command with
the "cluster" option (reduces standard errors) could be an alternative that
doesn't violates the assumption of independence of observations between
level 1 units. I've already tried gllamm.ado, but it seem not to be an
efficient algorithm for contextual models.
Any guidance apreciated.
Daniel
*
* 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/
