P.M.J. Stromberg <[email protected]>:
In a linear regression, you can run the fully interacted model or use
-suest-, but see e.g.
http://www.stata.com/statalist/archive/2008-02/msg00898.html
http://www.stata.com/statalist/archive/2008-02/msg00911.html
on the -probit- equivalent. In short, logit and probit do not
identify both the coefs *and* the error variance, so differences can
be observed across subsamples because the coefs differ *or* because
the error variance differs (or both).
On 01 Apr 2008 10:10:51 +0100, P.M.J. Stromberg <[email protected]> wrote:
> Hi,
>
> I want to run probit estimations on three sub samples (representing one
> ethnic group each), with the same variables (Y and X). But first I need to
> know if this is justified: is there a way to test if the difference among
> the sub samples is statistically significant (such as log likelihood ratio
> test)?
>
> An alternative is to use the pooled sample, and include dummies: but I
> suspect that the different ethnic groups affect (1) basically all
> variables, and (2) the slope, not only the intercept.
>
> Many thanks in advance,
> Per
*
* 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/
