[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: help
> -----Original Message-----
> From: firstname.lastname@example.org
> [mailto:email@example.com] On Behalf Of
> Maximo Rossi
> Sent: 30 November 2006 19:01
> To: firstname.lastname@example.org
> Subject: st: help
> Dear Statalisters,
> We are estimating an -ivprobit- model and have some questions on the
> - Is the model specified only for continuous endogenous
> regressors ?
> In the case of a binary specification, it would be more
> correct to get the predicted values after a probit
> specification and use it as instrument ? Or, in the case of
> having a continuous specification of the endogenous
> regressors, is it more convenient to take it and use
> ivprobit straightforward ?
> - What is the difference between the predicted probability of
> the -mfx- command after -ivprobit- and the mean of the
> predicted values of -predict (p)- after estimation ?
> - We find contradictory results between the overid test
> after -ivprobit- and the same test specifying a Linear
> Probability Model -ivreg2-. Is this correct ?
Could be. They are different models. The LPM is heteroskedastic as
well. But the easiest thing to do is to install the latest version of
-overid-, because it supports an overidentification test for -ivprobit-
with the -twostep- option.
> - Finally, we find important differences in terms of endogeneity
> testing when using the ML and the Two-Step procedeures and
> can only wonder what is the reason for that. Which estimation
> should we "trust" for endogneity testing ?
> - In the case of choosing ML are there any test of relevance and
> validity of instruments available ? because overid only works
> after the two-step option.
> Thanks in advance,
> Maximo Rossi
> Department of Economics
> Faculty of Social Sciences
> University of Uruguay
> * 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: