In your example, intead of estimating the six parameters
[eq1]pnsdif
[eq2]pnselig98_hat
[eq1]earndif
[eq2]earn98_hat
[eq1]matchdif
[eq2]pnsmatches98_hat
just estimate 4 of them, then solve for the other two terms of those 4 and
use them in your expression for the likelihood. For example, put
[eq2]pnselig98_hat , [eq1]earndif, [eq2]earn98_hat and [eq1]matchdif
directly in your model statement. Then in the .ado program which calulates
the likelihood, everywhere you have , `pnsdif', substitute
`pnselig98_hat'*`earndif'/`earn98_hat', and so forth.
Hope this helps -
Al Feiveson
-----Original Message-----
From: Kobes, Deborah [mailto:[email protected]]
Sent: Thursday, October 03, 2002 10:52 AM
To: '[email protected]'
Subject: st: RE: RE: constraints
thanks -- how do i hard code this in the ml-program? i couldn't find
anything in the FAQs.
-----Original Message-----
From: FEIVESON, ALAN H. (AL) (JSC-SD) (NASA)
[mailto:[email protected]]
Sent: Thursday, October 03, 2002 11:07 AM
To: '[email protected]'
Subject: st: RE: constraints
This question has been asked repeatedly (is it a FAQ?). Unfortunately other
than hard coding the constraints in your ml-program, Stata's ml-routine does
not have a capability for incorporating them. I would like to add this as a
desirable capability for Stata 8 (at least for linear constaints, but better
yet for noinlinear ones also).
Al Feiveson
-----Original Message-----
From: Kobes, Deborah [mailto:[email protected]]
Sent: Thursday, October 03, 2002 9:28 AM
To: '[email protected]'
Subject: st: constraints
I would like to use the constraints:
constraint define 1 [eq1]pnsdif =
[eq2]pnselig98_hat*[eq1]earndif/[eq2]earn98_hat;
constraint define 2 [eq1]matchdif =
[eq2]pnsmatches98_hat*[eq1]earndif/[eq2]earn98_hat
for a maximum likelihood estimator.
When I try to do this, I get the error:
Constraints invalid:
not possible with test
r(131);
Is there any way I can set up constraints that define one coefficient in
terms of multiple other coefficients? It works when I try to define one in
terms of only one other.
I am doing this because the ml model has two equations, each with a
different error term. I'd like to constrain the coefficients for each part
to be equal to each other, but this has to take into consideration those
error terms.
Thanks!
Deborah.
*
* 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/
