Maria has two different samples with the same model, not different
models on the same sample. If the two samples are truly independently
taken, then using the sum of variances in denominator is fine, as it
boils down to something like a z-test, but if those data sets are two
parts of the same sample, then doubts might arise.
On 8/8/07, Maarten Buis <[email protected]> wrote:
> --- [email protected] [mailto:[email protected]] wrote:
> > Do you reckon that now I can assess the significance of the difference between
> > two R2's: R2_a and R2_b (obtained from the same model run on two different
> > samples, a and b) using the boostrap SEs in the following way?
> >
> > (R2_a-R2_b)/sqrt[(SE_a)^2+(SE_b)^2]
>
> The way to compare R squares is to use F-tests (as long as you are using the
> same dependent variable). Though, I don't like the interpretation of a test
> on "improvement" in R2. A much more sensible interpretation is as a test of
> a constraint on the parameters, since it is the parameters that are of
> substantive interest not the R2.
>
> -- 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 Z434
>
> +31 20 5986715
>
> http://home.fsw.vu.nl/m.buis/
> -----------------------------------------
>
>
> *
> * 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/
>
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: Please do not reply to my Gmail address as I don't check
it regularly.
*
* 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/
