Alfred - With regard to Question 2, you can try averaging your 0-1 response
variable over groups of values of a particular independent variable ignoring
the values of other independent variables, then do a scatter plot. This will
give a clue as to whether that independent variable should enter the
equation in linear, reciprocal, or whatever form. Of course, this tells you
nothing about how various independent variables woud interact in a joint
model.
Al Feiveson
-----Original Message-----
From: [email protected] [mailto:[email protected]]
Sent: Thursday, January 09, 2003 9:25 PM
To: [email protected]
Subject: st: Building a Multinomial Logit Model
Dear all,
I would like to build a multinomail logti model over
a large number of variables. I will be glad if
anyone could help me to solve problems as follows:
1. whether the multicollinearity between factors will
affect the psedo R-square and the log likelihood for
goodness of fit.
2. Is there any method that can tell me the possible
"functional form" in which the factors should be
transformed before putting into the model, and
3. Do the stepwise procedures work well for MNL and
is there any alternative methods?
Thanks a lot,
Regards,
Alfred
*
* 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/
