Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: IV Probit


From   Robert Davidson <rhd773@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: IV Probit
Date   Thu, 15 Mar 2012 18:38:30 -0400

Thank you for the link; it is quite helpful.  To keep things simple,
it appears I should use a linear iv model, such as ivreg2, or a
bivariate probit model, depending on whether I am more interested in
consistent estimates of ATT or ATE, or depending on which assumptions
I am more comfortable (most likely) violating.  If I use the biprobit
function, can I interpret the significance (or lack thereof) of the
likelihood ratio test of independent equations as evidence for the
need to use such a model (when rho is significant).

Thank you,
Rob

On Thu, Mar 15, 2012 at 2:22 PM, Austin Nichols <austinnichols@gmail.com> wrote:
> Robert Davidson <rhd773@gmail.com>:
> I think you should still report the -ivreg2- estimates and diagnostics.
> In particular, *you should not use -ivprobit- for this case*
> --look at this from the -ivprobit- help file:
> "Both estimators assume that the endogenous regressors are continuous
> and are not appropriate for use with discrete endogenous regressors."
> See also http://www.stata.com/meeting/chicago11/materials/chi11_nichols.pdf
>
> On Thu, Mar 15, 2012 at 12:46 PM, Robert Davidson <rhd773@gmail.com> wrote:
>> Thank you for your response.  My endogenous regressor is also binary,
>> so even the first stage model is binary in nature.  Does this
>> basically mean that I cannot produce much in the way of diagnostics
>> other than to simply show that the instrument is highly correlated
>> with the endogenous regressor and weakly correlated with the error
>> term in the original model for which the instrument is to be used?
>>
>>
>> On Thu, Mar 15, 2012 at 12:18 PM, Austin Nichols
>> <austinnichols@gmail.com> wrote:
>>> Robert Davidson <rhd773@gmail.com>:
>>> The theory of the diagnostics is not developed for -ivprobit- or any
>>> other nonlinear model,
>>> but you can certainly report the diagnostics for the linear probability model
>>> that corresponds using -ivreg2-. Since weak instruments are a property
>>> of the first stage, this is as much as any reviewer is likely to want.
>>> Overidentification tests are also possible via -gmm-.
>>>
>>> On Thu, Mar 15, 2012 at 11:19 AM, Robert Davidson <rhd773@gmail.com> wrote:
>>>> Hello,
>>>> I am trying to estimate an ivprobit model in Stata 11.2.
>>>> I would like to estimate an F-statistic to assess weak instruments and
>>>> perform other tests to assess the potential endogeneity of the
>>>> instrument. None of the commands available with ivreg seem to work. I
>>>> have looked at the help for ivprobit and the post estimation commands
>>>> and cannot find these things. I imagine these calculations are basic,
>>>> but I have not used ivprobit before and cannot find guidance online.
>>>> I am estimating a standard ivprobit though the standard errors are
>>>> clustered by groups.
>>>> Thank you,
>>>> Rob
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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   |   Site index