Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: postestimation after asclogit


From   "Murali Kuchibhotla" <muralik@iastate.edu>
To   statalist@hsphsun2.harvard.edu
Subject   st: postestimation after asclogit
Date   Mon, 27 Oct 2008 15:54:06 -0500 (CDT)

Hello,
      I have a few questions regarding the use of certain post-estimation
commands after asclogit. I am estimating a model for training choice, with 3
alternatives for training and where individuals are allowed to choose more than
one training category(off-the-job, on-th-job or none).

My questions are- (1) I am interested in obtaining marginal effects, where the
change in probabilities associated with any given variable are conditioned on
the actual no. of alternatives chosen. Specifying the following does not seem to
work though, as I get the following error message-

estat  mfx, at(mean) k(observed)
option k() incorrectly specified

How can I get over this problem?

(2)I am also interested in testing the hypothesis that the coefficient estimates
are the same across the different training categories. For example, with years
of education entering as a case specific independent variable, I want to test
the following hypothesis:

     education coefficient[alternative 1]=education coefficient[alternative 2]

How do I accomplish this?

Also, is there a way to test the above hypotheses jointly for all the case
specific independent variables together? 

Suggestions would be greatly appreciated. Many thanks.

Murali Kuchibhotla



 



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index