As Maarten noted -lincom- is designed to handle the situation of dependent parameters for you.
It is rare outside designed experiments, where one goes to a great deal of trouble to guarantee independence, to have orthogonal parameter estimates.
-----Original Message-----
From: "Bell, Jacqueline S." <[email protected]>
To: "[email protected]" <[email protected]>
Sent: 10/15/2008 4:25 PM
Subject: RE: st: lincom - how are standard errors calculated?
Thanks Maarten. Is it valid to use lincom in this situation, with non-zero covariance?
Jacqui
-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Maarten buis
Sent: 15 October 2008 11:38
To: [email protected]
Subject: Re: st: lincom - how are standard errors calculated?
--- "Bell, Jacqueline S." <[email protected]> wrote:
> I'm using lincom in conjunction with svy:mean and getting standard
> errors for the difference between 2 estimates that are smaller than
> the individual SEs. This doesn't seem right given that var(A-B) =
> varA+varB. Can someone tell me what is going on?
That is because var(A-B) = var(A) + var(B) - 2cov(A,B) .
Hope this helps,
Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting address:
Buitenveldertselaan 3 (Metropolitan), room N515
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
The University of Aberdeen is a charity registered in Scotland, No SC013683
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
