Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: st: Instrument Validity tests for Heckman 1979
From
"Kyrizi, Andri" <[email protected]>
To
"[email protected]" <[email protected]>
Subject
RE: st: Instrument Validity tests for Heckman 1979
Date
Thu, 13 Mar 2014 13:15:19 +0000
Dear all,
I am trying to understand how the Heckman MLE works. Is it still a two step procedure, i.e has a selection equation to calculate the mills ratio and include it in our main regression?
Prob(wi = 1|zi) = Φ(ziγ)
and
Prob(wi = 0|zi) = 1 − Φ(ziγ),
Thank you,
Andri
________________________________________
From: [email protected] [[email protected]] on behalf of Kyrizi, Andri [[email protected]]
Sent: 13 March 2014 06:01
To: [email protected]
Subject: RE: st: Instrument Validity tests for Heckman 1979
Dear Professor Schaffer,
Sorry for not giving more details, I didn't think they were relevant. Next time I will include more information.
In any case, I managed to do it. Thank you so much for your help.
All the best,
Andri
________________________________________
From: [email protected] [[email protected]] on behalf of Schaffer, Mark E [[email protected]]
Sent: 12 March 2014 17:52
To: [email protected]
Subject: RE: st: Instrument Validity tests for Heckman 1979
Andri,
You need to provide more information. "The first heckman regression" doesn't tell us much. Have a look at
http://www.stata.com/support/faqs/resources/statalist-faq/
for the kinds of things we would need to know, especially Section 3. Version of Stata, your code, sample output, etc. etc.
--Mark
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of Kyrizi, Andri
> Sent: 12 March 2014 16:58
> To: [email protected]
> Subject: RE: st: Instrument Validity tests for Heckman 1979
>
> Dear Professor Schaffer,
>
> Thank you for the information.
>
> I tried it but I couldn't get the first heckman regression.
>
> I kept getting
> Iteration 61: log likelihood = -2057.582
>
> with no estimations.
>
> Would it be because I use BHPS and has many observations?
>
>
> All the best,
> Andri
> ________________________________________
> From: [email protected] [owner-
> [email protected]] on behalf of Schaffer, Mark E
> [[email protected]]
> Sent: 12 March 2014 16:23
> To: [email protected]
> Subject: RE: st: Instrument Validity tests for Heckman 1979
>
> Andri,
>
> > -----Original Message-----
> > From: [email protected] [mailto:owner-
> > [email protected]] On Behalf Of Kyrizi, Andri
> > Sent: 12 March 2014 16:05
> > To: [email protected]
> > Subject: RE: st: Instrument Validity tests for Heckman 1979
> >
> > Professor Schaffer,
> >
> > I checked it but I wasn't sure if the selection bias is controlled for
> > if you do not specify the group.
>
> From the help file:
>
> Basic syntax
>
> heckman depvar [indepvars], select(varlist_s) [twostep]
>
> ...
>
> select(...) specifies the variables and options for the selection equation. It
> is an integral part of specifying a Heckman
> model and is required. The selection equation should contain at least
> one variable that is not in the outcome
> equation.
>
> If depvar_s is specified, it should be coded as 0 or 1, with 0 indicating an
> observation not selected and 1 indicating
> a selected observation. If depvar_s is not specified, observations for
> which depvar is not missing are assumed
> selected, and those for which depvar is missing are assumed not
> selected.
>
> --MS
>
> >
> > Thank you again,
> >
> > All the best,
> > Andri
> > ________________________________________
> > From: [email protected] [owner-
> > [email protected]] on behalf of Schaffer, Mark E
> > [[email protected]]
> > Sent: 12 March 2014 15:56
> > To: [email protected]
> > Subject: RE: st: Instrument Validity tests for Heckman 1979
> >
> > Andri,
> >
> > > -----Original Message-----
> > > From: [email protected] [mailto:owner-
> > > [email protected]] On Behalf Of Kyrizi, Andri
> > > Sent: 12 March 2014 15:51
> > > To: [email protected]
> > > Subject: RE: st: Instrument Validity tests for Heckman 1979
> > >
> > > Dear Professor Schaffer,
> > >
> > > Thank you so much for your reply and your examples.
> > > Everything is so much clearer now.
> > >
> > > I have one last question:
> > >
> > > After the select option shouldn't I specify the variable that I want
> > > to control for the selectivity bias (i.e a dummy variable indicating
> > > Unemployed and Employed individuals)?
> > > heckman lnEarnings Schooling Experience Exp2 Male married, select
> > > ( Unemp= Male married children )
> >
> > If you check the -heckman- help file you will see both syntaxes are allowed.
> >
> > --MS
> >
> > > All the best
> > > Andri
> > > ________________________________________
> > > From: [email protected] [owner-
> > > [email protected]] on behalf of Schaffer, Mark E
> > > [[email protected]]
> > > Sent: 11 March 2014 15:16
> > > To: [email protected]
> > > Subject: RE: st: Instrument Validity tests for Heckman 1979
> > >
> > > Andri,
> > >
> > > > -----Original Message-----
> > > > From: [email protected] [mailto:owner-
> > > > [email protected]] On Behalf Of Kyrizi, Andri
> > > > Sent: 11 March 2014 14:22
> > > > To: [email protected]
> > > > Subject: RE: st: Instrument Validity tests for Heckman 1979
> > > >
> > > > Dear all,
> > > >
> > > > In regards to yesterday's question and answer,
> > > >
> > > > Does anyone know if I could use a two step heckman approach
> > > > instead of the treatreg approach?
> > > > And test it using the -ltrest-?
> > >
> > > You can't use -lrtest- after the Heckman 2-step estimator because
> > > -heckman- with the twostep option doesn't save a log likelihood.
> > >
> > > But there's probably no particular reason to use the two-step
> > > estimator anyway when the ML estimator is available.
> > >
> > > So estimate your model using -heckman- with the default ML
> > > estimator, and then do an overid test by comparing it to a
> > > just-identified version, where just-identification comes from the
> functional form.
> > > And if you want to test the validity of some of the exclusion
> > > restrictions, do an LR test with/without them.
> > >
> > > Here are three examples using Stata's sample womenwk dataset. The
> > > first compares the overidentified version with 2 exclusion
> > > restrictions (and the functional form restriction) to one which is
> > > just-
> > identified via functional form.
> > > The second and third compare the overidentified version with 2
> > > exclusion restrictions to versions with just one exclusion
> > > restriction. The output of the estimations is suppressed just to
> > > make things
> > a bit more readable. The "m"
> > > and "c" at the end of the names of the saved estimation results
> > > refer to "married" and "children".
> > >
> > > HTH,
> > > Mark
> > >
> > > . webuse womenwk
> > >
> > > .
> > > . qui heckman wage educ age, select(married children educ age)
> > >
> > > . est store overidmc
> > >
> > > . qui heckman wage educ age, select(educ age)
> > >
> > > . est store justid
> > >
> > > .
> > > . lrtest overid justid
> > >
> > > Likelihood-ratio test LR chi2(2) = 358.82
> > > (Assumption: justid nested in overid) Prob > chi2 = 0.0000
> > >
> > > .
> > > . qui heckman wage educ age, select(married educ age)
> > >
> > > . est store overidm
> > >
> > > .
> > > . lrtest overidmc overidm
> > >
> > > Likelihood-ratio test LR chi2(1) = 350.01
> > > (Assumption: overidm nested in overidmc) Prob > chi2 = 0.0000
> > >
> > > .
> > > . qui heckman wage educ age, select(children educ age)
> > >
> > > . est store overidc
> > >
> > > .
> > > . lrtest overidmc overidc
> > >
> > > Likelihood-ratio test LR chi2(1) = 43.43
> > > (Assumption: overidc nested in overidmc) Prob > chi2 = 0.0000
> > >
> > >
> > > >
> > > > Would that still be correct?
> > > >
> > > > All the best,
> > > > Andri
> > > >
> > > > ________________________________________
> > > > From: [email protected] [owner-
> > > > [email protected]] on behalf of Kyrizi, Andri
> > > > [[email protected]]
> > > > Sent: 10 March 2014 18:09
> > > > To: [email protected]
> > > > Subject: RE: st: Instrument Validity tests for Heckman 1979
> > > >
> > > > Hello Prof Antonakis,
> > > >
> > > > Thank you so much for your reply and help!
> > > >
> > > > All the best,
> > > > Andri
> > > > ________________________________________
> > > > From: [email protected] [owner-
> > > > [email protected]] on behalf of John Antonakis
> > > > [[email protected]]
> > > > Sent: 10 March 2014 17:35
> > > > To: [email protected]
> > > > Subject: Re: st: Instrument Validity tests for Heckman 1979
> > > >
> > > > Hi:
> > > >
> > > > See the following (if you are talking about an overidentification test):
> > > >
> > > > http://www.stata.com/statalist/archive/2010-10/msg00915.html
> > > >
> > > > A -heckman- and -treatreg- (now -etregress-) basically do the same
> > > > thing, the difference being that the former has a truncated sample on
> y.
> > > >
> > > > So, it seems to me you can use the -ltrest- as explained by Mark
> Schaffer.
> > > >
> > > > Best,
> > > > J.
> > > >
> > > > __________________________________________
> > > >
> > > > John Antonakis
> > > > Professor of Organizational Behavior Director, Ph.D. Program in
> > > > Management
> > > >
> > > > Faculty of Business and Economics
> > > > University of Lausanne
> > > > Internef #618
> > > > CH-1015 Lausanne-Dorigny
> > > > Switzerland
> > > > Tel ++41 (0)21 692-3438
> > > > Fax ++41 (0)21 692-3305
> > > > http://www.hec.unil.ch/people/jantonakis
> > > >
> > > > Associate Editor:
> > > > The Leadership Quarterly
> > > > Organizational Research Methods
> > > > __________________________________________
> > > >
> > > > On 10.03.2014 18:06, Kyrizi, Andri wrote:
> > > > > Dear Statalisters,
> > > > >
> > > > > I would like to test the validity of my instruments after using
> > > > > the Heckman's
> > > > (1979) two-step sample selection model. In particular I want to
> > > > correct for the selectivity bias for using only employed individuals.
> > > > >
> > > > > I've searched everywhere to find a way to test them but could't
> > > > > find
> > > > anything. I am aware that a topic on 'diagnostics for the treatreg
> > procedure'
> > > > exists but I don't think it relates to my question.
> > > > >
> > > > > I would really appreciate any guidance you could give me!
> > > > >
> > > > > Many thanks,
> > > > > Andri
> > > > >
> > > > >
> > > > > *
> > > > > * For searches and help try:
> > > > > * http://www.stata.com/help.cgi?search
> > > > > * http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
> > > > * http://www.ats.ucla.edu/stat/stata/
> > >
> > >
> > > -----
-----
Sunday Times Scottish University of the Year 2011-2013
Top in the UK for student experience
Fourth university in the UK and top in Scotland (National Student Survey 2012)
We invite research leaders and ambitious early career researchers to
join us in leading and driving research in key inter-disciplinary themes.
Please see www.hw.ac.uk/researchleaders for further information and how
to apply.
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/