Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Standard error of the estimate for svy: reg


From   Steven Samuels <sjhsamuels@earthlink.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Standard error of the estimate for svy: reg
Date   Wed, 22 Aug 2007 12:47:33 -0400

Needless line

Ah, yes--the "Standard Error of the Estimate" nomenclature threw me. The root MSE does not appear to me to be the standard error of any estimate, but the terminology is obviously different in political science.

The root MSE is an estimate of the assumed constant SD of single error terms in the OLS model. But -svy; reg- does not require the assumption of constant SD. This may have been one reason why the - svy- writers were 'queasy' about including the estimate: its presence could be misinterpreted to mean that the assumption is true.


Steve

On Aug 22, 2007, at 11:16 AM, Richard Forshee wrote:


Thank you for the pointers, Nick.

I share your concern about why root MSE was not included in svy:, and I hope
someone will shed some light on that.

--Rich


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: Wednesday, August 22, 2007 11:10 AM
To: statalist@hsphsun2.harvard.edu
Subject: RE: st: Standard error of the estimate for svy: reg

Well, the root mean square is best not
described as an "average". It is evidently
a kind of average, but a more precise description
is easy enough.

After a model you can use -predict- to get at residuals
and then run -summarize- (and anything else
needed) to get whatever summary measures you want.

Essentially the same issue for -glm- is
addressed by -glmcorr- on SSC.

Apart from doing it by hand, most of the code
in -glmcorr- might be used for your problem.

However, a much bigger question is this: Why did the
writers of -svy: reg- not provide this? Perhaps
they were queasy about what summary measure(s)
are justifiable in a -svy:- context. I don't
know, but that's what you need to worry about.

Nick
n.j.cox@durham.ac.uk

Richard Forshee


Steven, this measure is also called the root mean squared
error (which is
reported in the -reg- command but not in the -svy: reg-
command), and it
measures the average deviation of a predicted value from the
actual observed
value in the sample.
Steven Samuels

I don't have access to this journal.  My question: standard error of
WHAT estimate?
Richard Forshee

A recent article in my field encourages reporting both the R-
squared and the
Standard Error of the Estimate (or Root MSE).  (James S. Krueger
and Michael
S. Lewis-Beck. Goodness-of-Fit: R-Squared, SEE and 'Best
Practice'.  The
Political Methodologist, vol. 15, no. 1, pp 2-4.  2007)

When I run a svy: reg command, I do not see the SEE (or Root MSE)
reported.
I checked ereturn list, and I did not see anything that
appeared to
be the
SEE.  Is there a way calculate the SEE when running a svy: reg
command?
Maybe it could be derived using the predict , stdp postestimation
command?
Or is there a reason that SEE is not appropriate in the context of
a svy:
reg command?
*
*   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/
Steven  Samuels

sjhsamuels@earthlink.net
18 Cantine's Island
Saugerties, NY 12477
Phone: 845-246-0774
EFax: 208-498-7441




*
*   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