# st: setting cross-group constraints using gologit2

 From William Magee To statalist@hsphsun2.harvard.edu Subject st: setting cross-group constraints using gologit2 Date Thu, 17 May 2007 13:03:49 -0400

Dear Statalist Members,

I am trying to set constraints on cutpoints in gologit2
(whether that is a good idea or not is also a question)

I did the following, and got the following error.

. constraint 101 [0]_b[_const] = -.4716279

. constraint 102 [1]_b[_const] = -1.378691

. constraint 103 [2]_b[_const]= -2.806735

.
.
end of do-file

. do "/tmp/SD00244.000000"

. gologit2 homestranx workstranx age0 age0sq if female ==1, link (cloglog) cons(101-103) auto(.01)

------------------------------------------------------------------------ ------
Testing parallel lines assumption using the .01 level of significance...

Step 1: Constraints for parallel lines imposed for workstranx (P Value = 0.9847)
name found where constraint number expected
constraint numbers and matrices may not be combined
r(7);
-------------
I did this because I assumed that the equations are numbered as follows
(from the above model, both men and women included) :

------------------------------------------------------------------------ ------
homestranx | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------------- +----------------------------------------------------------------
0 |
workstranx | .1889934 .0476276 3.97 0.000 . 0956449 .2823418
age0 | .0269906 .0133104 2.03 0.043 . 0009027 .0530785
age0sq | -.0005567 .0002772 -2.01 0.045 -.0011001 -.0000133
_cons | -.4716279 .1688549 -2.79 0.005 -.8025775 -.1406784
------------- +----------------------------------------------------------------
1 |
workstranx | .1889934 .0476276 3.97 0.000 . 0956449 .2823418
age0 | .0269906 .0133104 2.03 0.043 . 0009027 .0530785
age0sq | -.0005567 .0002772 -2.01 0.045 -.0011001 -.0000133
_cons | -1.378691 .1744944 -7.90 0.000 -1.720694 -1.036688
------------- +----------------------------------------------------------------
2 |
workstranx | .1889934 .0476276 3.97 0.000 . 0956449 .2823418
age0 | .0269906 .0133104 2.03 0.043 . 0009027 .0530785
age0sq | -.0005567 .0002772 -2.01 0.045 -.0011001 -.0000133
_cons | -2.806735 .1964953 -14.28 0.000 -3.191858 -2.421611
------------------------------------------------------------------------ ------

I would welcome comments on whether it is ok to try to do this as well as how to do this (conatrain cutpoints across equations). BTW -- I did check for gender-related heterogeneity using oglm...

thanks,
bill magee

*
* 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/