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: oheckman w/twostep - can I bootstrap for a Wald?
From
Stas Kolenikov <[email protected]>
To
"[email protected]" <[email protected]>
Subject
Re: st: oheckman w/twostep - can I bootstrap for a Wald?
Date
Tue, 9 Jul 2013 10:17:20 -0500
My gut feeling is that you can do this whole thing using David
Roodman's award winning -cmp-, at least as an alternative FIML
procedure. It has some learning curve, but once you figure it out, you
won't run any non-standard multi-equation estimation problems other
than through -cmp- :).
-- Stas Kolenikov, PhD, PStat (SSC)
-- Senior Survey Statistician, Abt SRBI
-- Opinions stated in this email are mine only, and do not reflect the
position of my employer
-- http://stas.kolenikov.name
On Tue, Jul 9, 2013 at 9:25 AM, Trent Spaulding
<[email protected]> wrote:
> As a reference, I built the bootstrap test this way following page 176
> of Chiburis and Lokshin's article in the Stata Journal:
> http://www.stata-journal.com/sjpdf.html?articlenum=st0123
>
> Does this appear valid or should I find a different approach?
>
> On Mon, Jul 8, 2013 at 4:38 PM, Trent Spaulding
> <[email protected]> wrote:
>> Can anyone check me on this?
>>
>> What I am trying to get: Estimates of the Wald or LR tests.
>>
>> Details:
>> - Ordered categorical variable has five levels.
>> - Running a similar model on several outcome variables
>> - I have not successfully got the FIML estimation to converge, so I am
>> using twostep
>>
>> Question: Is the following providing me a Wald test (or reasonable substitute)?
>>
>> -------CODE--------
>>
>> capture program drop aepost
>> program aepost, eclass
>> tempname bb
>> oheckman **model specifications*** twostep
>> matrix `bb'= e(rho)
>> ereturn post `bb'
>> end
>> bootstrap _b, reps(100) nowarn: aepost
>> test rho0=rho1=rho2=rho3=rho4=0
>>
>> -------OUTPUT--------
>>
>> Bootstrap results Number of obs = 2275
>> Replications = 100
>>
>> ------------------------------------------------------------------------------
>> | Observed Bootstrap Normal-based
>> | Coef. Std. Err. z P>|z| [95% Conf. Interval]
>> -------------+----------------------------------------------------------------
>> rho0 | -.2564272 .2127177 -1.21 0.228 -.6733463 .1604919
>> rho1 | -.2837067 .1870317 -1.52 0.129 -.650282 .0828687
>> rho2 | -.043747 .3578368 -0.12 0.903 -.7450943 .6576004
>> rho3 | -.471063 .262215 -1.80 0.072 -.984995 .042869
>> rho4 | -.6076915 .1845782 -3.29 0.001 -.9694582 -.2459249
>> ------------------------------------------------------------------------------
>>
>> . test rho0=rho1=0
>>
>> ( 1) rho0 - rho1 = 0
>> ( 2) rho0 = 0
>>
>> chi2( 2) = 4.14
>> Prob > chi2 = 0.1264
>>
>> . test rho0=rho1=rho2=rho3=rho4=0
>>
>> ( 1) rho0 - rho1 = 0
>> ( 2) rho0 - rho2 = 0
>> ( 3) rho0 - rho3 = 0
>> ( 4) rho0 - rho4 = 0
>> ( 5) rho0 = 0
>>
>> chi2( 5) = 17.02
>> Prob > chi2 = 0.0045
>> *
>> * 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/