Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: collinearity diagnostics and competing condition numbers


From   "Tomislav Kovandzic" <tkovandzic@bham.rr.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: collinearity diagnostics and competing condition numbers
Date   Mon, 10 Jul 2006 20:44:30 -0500

Dear List,

I’m trying to identify collinearity problems with the user-written
program (COLLDIAG) which computes Belsley et al.’s collinearity
diagnostics.  I was hoping someone could clarify something in the
colldiag help file for diagnosing potential dependencies in the data.
The help file states:    

	  "Those variates whose aggregate variance-decomposition
proportion
        exceed the threshold value p* are involved in at least one of
        the dependencies. The aggregate is formed over the competing
condition
        numbers (condition numbers of the same order of magnitude that
exceed
        the threshold value n*)."

What does it mean to say “condition numbers of the same order of
magnitude”?  Is this the number of powers of 10 contained within the
number?  How close do the condition numbers have to be to each other
before one considers them of “the same order of magnitude"?  I
understand CI”s above 30 point to potential collinearity problems, but
what I don’t understand is which variance-decomposition proportions
should be aggregated (to see if they’re over the rule of thumb of .50). 
Any help would be appreciated.  Thanks in advance.
 
Best,
 
Tom Kovandzic    
 
      



*
*   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