Ralph Bl�thgen <[email protected]>:
Roger's advice is to estimate a model identical to -ivreg- in two
steps, which would produce identical point estimates but incorrect
standard errors. -ivreg- and -ivreg2- will both produce consistent
estimates, even when the endog var is binary, and -treatreg- offers
greater efficiency if the distributional assumptions are justified.
On 6/21/07, Roger Smeets <[email protected]> wrote:
You could always estimate a first-stage probit separately, then predict
the dependent variable from the estimated coefficients (predict y if
e(sample)) and use the predicted variable as an explanatory variable in
your second stage model.
Roger
-----Original Message-----
From: Ralph
Bl�thgen
Sent: Thursday, June 21, 2007 2:18 PM
Hi to everyone,
I want to estimate a two-stage model by IV. Unfortunately, my endogenous
variable is a dummy.
Does anybody know how to introduce something like a probit model for the
first-stage-regression into ivreg?
*
* 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/
