RE: st: cmp

["David Roodman (droodman@cgdev.org)" <DRoodman@cgdev.org>]

1. cmp doesn't do mlogit. It only does models based on the (multivariate) Gaussian distriubiton.
2. Your interpretation is correct. As explained in the help file, the equation for the lowest value of the dependent variable becomes the base alternative. You cannot change this unless you switch to the other mprobit syntax, in which you list a separate equation for each outcome and can control the order.
3. Can't answer it without specifics.
--David

-----Original Message-----
Subject: statalist-digest V4 #4695
Date: Wed, 7 Nov 2012 14:12:03 -0500
From: Anna Manzoni <anna.manzoni@gmail.com>
Subject: st: cmp

Dear STATA lister,
I am using cmp to estimate a two stage model which has a 4-category
variable as dependent variable in the first stage and another
4-category variable as dependent variable in the second stage.
I read Roodman's article in the Stata Journal , which was very
helpful, but I still do not have clear how to correctly interpret the
output (among other things). So, any references about that would be
appreciated.

First of all, I am currently using mprobit because I cannot find a way
to get it work with mlogit.
So my first question is : 1) Any suggestions about using mlogit in cmp?

My syntax reads something like
cmp(outcomest2= x1 x2 x3 outcomest1) (outcomest1= x4 x1 x2 x3),
ind($cmp_mprobit  $cmp_mprobit)

I am not sure how to interpret the output. In total, I get 8 tables
and I think the first 4 tables refer to the second stage (one for each
category of the dep var), and the following tables to the first stage.
However, I am not sure what's the reference category (the biggest
category?) and how I could specify which category I would like to use
as base (question 2)

3) The third question refers to  the many warnings I get. Does that
have anything to do with the choice of my instruments or is it
unrelated?

4) is there a way to test the validity of the instrument(s) with cmp?
Thank you. Anna Manzoni
*
*   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/