I'm not quite sure what the question is. The Huber-White sandwich estimator
is calculated using a Taylor series expansion. It's actually the clustering
and the stratification that drives the problem, not the weights.
Bryan Sayer
Statistician, SSS Inc.
-----Original Message-----
From: Safir, Adam
To: '[email protected]'
Sent: 9/10/02 8:39 AM
Subject: st: Sandwich estimator, Huber-White
I have a question about using svymeans and pweight in stata for complex
survey data. The corrected standard error reported comes close to the
standard error reported when using replicate weights, and is usually
somewhat smaller, although in some cases it is larger.
I understand that using the Taylor Series will sometimes generate a
corrected standard error somewhat larger than that estimated using
replicate
weights, and that this is because the Taylor Series cannot account for
the
raking, which can increase the precision of the estimate.
Is there something similar happening for the corrected standard error
when
the Huber-White adjustment is used in the svymeans pweight procedure?
Or
can anyone suggest a good text or article on this issue?
thanks in advance.
Adam
Adam Safir
The Urban Institute
[email protected]
202-261-5247 (voice)
202-293-1918 (fax)
*
* 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/
