Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE:multicollinearity in mlogit


From   "Yatawara, Ravi" <yatawarr@lerner.udel.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE:multicollinearity in mlogit
Date   Mon, 30 May 2005 16:51:49 -0400

Folks,
I have a dependent variable which is 3 possible trade policy stances.
(status quo, liberalize or increase protection) for 40 countries for 30
years each.

The independent variables are alternate forms of crisis and a political
variable for government change. The govt change variable is
statistically significant- the crisis variables are not.

The issue with a referee is that maybe crisis is causing the government
change.

I have dropped govt change , and the crisis variables do not  become
significant.
I estimated a logit model of govt change using the crisis variables as
independent  variables  and carried out an F test to determine
collinaerity
Fk-2,n-k+1= [R squared/(k-2)]/ [1-R squared/(n-k+1)

Where k= numberof explanatory variables including intercept.

My questions are 1) is this sufficient under a multinomial framework and
2) are there better altenatives  I should investigate.

Any feedback will be much  appreciated!
Thanks
Ravi


*
*   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/



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