dear jennifer,
i'm currently working on a project where i'm doing exactly that. the way
i decided to solve the problem was by bootstrapping, which i think many
people would prefer to the "correct" standard errors, given that it's a
non-standard procedure.
Stas Kolenikov has a really helpful tutorial on how to use the - bstrap -
command for such purposes. he links to it from his webpage:
http://www.komkon.org/~tacik/stata/
i haven't gone & worked out any algebra for the "correct" standard errors,
but if that's what you want to do, then i'd recommend looking into the -
ml - procedure & investing in the Gould & Sribney book that scott merriman
just mentioned: http://www.stata.com/bookstore/mle.html
chris
On Sun, 30 Mar 2003, Jennifer Alix wrote:
> Hello,
>
> Does anyone have a program that allows one to interact an instrumented
> variable with other variables in a regression equation AND calculates the
> correct standard errors for this? Example:
>
> y = b1x1+b2x2+b3x1*x2
> I want to instrument x2 with some z's to get:
> y = b1x1+b2x2hat+b3x1*x2hat
>
> It seems to me that this should be out there somewhere...
>
> Thanks,
> Jennifer
>
>
> *
> * 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/
>
*
* 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/
