Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Testing of differences in R-square


From   "Lachenbruch, Peter" <Peter.Lachenbruch@oregonstate.edu>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   RE: st: Testing of differences in R-square
Date   Wed, 31 Mar 2010 08:09:13 -0700

To amplify a bit - this is specifically demonstrated in Draper and Smith as "extra SS" - and most likely in others.  My initial reaction was the same as Steve's, but by the time I read it, the answer was already on the list.  The use of e(sample) was an idea that hadn't occurred to me (of course, all data in regression are complete!  (-:) )


Tony

Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Steve Samuels
Sent: Tuesday, March 30, 2010 3:24 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Testing of differences in R-square

"R-squares is equivalent to assessing a difference between the mean
square errors."

I should clarify: This statement is true because the R-squares are
being computed on the same data and for the same response variable. To
ensure this condition, restrict the sample to observations with
non-missing values for all predictors in the two regressions.

On Tue, Mar 30, 2010 at 10:24 AM, Steve Samuels <sjsamuels@gmail.com> wrote:
> Vuong's test is likelihood based and corrects for the differing number
> of parameters in two models. For your problem you can directly
> compared adjusted R-squares, which also correct for the number of
> parameters.. However assessing a difference between two adjusted
> R-squares is equivalent to assessing a difference between the mean
> square errors. So I suggest you bootstrap a difference in log mean
> square errors. Because you want to bootstrap two regressions, you'll
> need to write your own program. See:
> http://www.ats.ucla.edu/stat/Stata/faq/ownboot.htm
>
> Steve
>

Steven Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax: 206-202-4783
*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index