Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: booststraping in two stage models


From   "Danny Cohen-Zada" <danoran@bgu.ac.il>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: st: booststraping in two stage models
Date   Tue, 11 Mar 2008 21:17:35 +0200

I will give you an example why it matters if the two procedures are equivalent. Suppose that i have an ivprobit model with clustering. The ivprobit, two step does not allow to cluster. In this case i have to do it by hand. I did it as you suggest. Then, i took the regular case (without clastering) and found that the two procedures give different results. What do you think?

Best,

Danny


----- Original Message ----- From: "Leonor Saravia" <lmisaravia@gmail.com>
To: <statalist@hsphsun2.harvard.edu>
Sent: Tuesday, March 11, 2008 8:26 PM
Subject: Re: st: booststraping in two stage models



Danny,

Stata has the ivprobit command, which allow you to calculate the
bootstrapped standard errors. So you can estimate your model using it:
ivprobit, vce(boot)

Also, you can do it by hand:

1) estimate the first stage.
2) use predict, pp
3) estimate the second stage by probit (in which we substiture pp instead
of the endogenous regressor). Here you can use the - vce(boot) - option

I hope this help,

Leonor


2008/3/11, Danny Cohen-Zada <danoran@bgu.ac.il>:
Leonor,

Thank you very much for the quick reply

I am asking about the standard errors. How should i obtain the correct
standard errors of the second stage? Don't my procedure produce the correct
standard errors?

Thanks
----- Original Message -----
From: "Leonor Saravia" <lmisaravia@gmail.com>
To: <statalist@hsphsun2.harvard.edu>
Sent: Tuesday, March 11, 2008 7:17 PM
Subject: Re: st: booststraping in two stage models


> Danny,
>
> The procedure you give is correct, you only need to correct the
> standard errors of the second stage.
>
> Leonor
>
> 2008/3/11, Danny Cohen-Zada <danoran@bgu.ac.il>:
>> I have a question that is relevant to all two stage models. For >> example,
>> take the ivprobit model. If for any reason i want to obtian >> bootstraped
>> standard errors. Can i do the following procedure.
>>
>> 1) estimate the first stage.
>> 2) use predict, pp
>> 3) estimate the second stage by probit (in which we substiture pp >> instead
>> of
>> the endogenous regressor) using the vce(boot) option.
>>
>> Will this be correct or should i bootstrapct the two stages together. >> If
>> i
>> must bootstrapt the two stages together, how should i do it?
>>
>> thanks,
>>
>> Danny
>>
>>
>>
>> Dr. Danny Cohen-Zada
>> Department of Economics
>> Ben-Gurion University of the Negev
>> P.O.Box 653
>> Beer-Sheva 84105
>> Israel
>>
>> Tel: +972-8-6472301
>> Fax: +972-8-6472941
>>
>> http://www.econ.bgu.ac.il/facultym/danoran/main.htm
>>
>> *
>> * 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/
>

*
* 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/

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index