RE: st: RE: variance in glm

["Wooldridge, Jeffrey" <wooldri1@msu.edu>]

I guess this is beyond my knowledge of how the glm command works in
Stata. My understanding is that if you use the Gaussian distribution
then the nominal variance assumption is constant -- regardless of the
link.

If you model the variance as, say, an exponential function of x, then
there are some pretty simple two-step methods that can be used to
estimate a conditional variance.

Jeffrey M. Wooldridge
University Distinguished Professor
Department of Economics
Michigan State University
110 Marshall-Adams Hall
East Lansing, MI 48824-1038
Phone: 517-353-5972
Fax: 517-432-1068
http://www.msu.edu/~ec/faculty/wooldridge/wooldridge.html
 

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Argyn
Kuketayev
Sent: Tuesday, March 01, 2011 4:03 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: RE: variance in glm

On Tue, Mar 1, 2011 at 3:52 PM, Wooldridge, Jeffrey <wooldri1@msu.edu>
wrote:
> Do you mean for each x in your data set or graphed as a function of x?

either graphed or data set.

>
> In either case, you first need to determine which distribution and
which
> link function you are using. What is the nature of y?

the distribution is gaussian, identity

>
> By the way, it is equivalent to say that you would like the estimates
of
> Var(y|x) without introducing an additive error term.

right, but the error term is additive in this case.

cheers
Argyn Kuketayev
*
*   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/

*
*   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/