Re: st: Fwd: Multicollinearity

[David Hoaglin <dchoaglin@gmail.com>]

It seems excessive to talk about "multicollinearity" when
"collinearity" is already general enough.

Whether collinearity is a problem depends on how serious it is and on
the particular variables involved.  Pairwise correlations may not
provide enough information if the collinearity relation involves more
than two variables.  One can get at the details by using the
regression coefficient variance decomposition developed by David
Belsley and described in the book by Belsley, Kuh, and Welsch (1980).
The implementation in the user-written command -coldiag2- (from SSC,
as I recall) is useful.

David Hoaglin

Belsley D.A., Kuh E., Welsch R.E. (1980). Regression Diagnostics.
John Wiley & Sons.

On Mon, Jun 3, 2013 at 3:32 AM, Maarten Buis <maartenlbuis@gmail.com> wrote:
> --- Prakash Kashwan wrote:
>> I am unable to use the stata-listserv, and hence this private
>> email to you.
>
> This is probably because you did not sent your message as plain text.
> This is explained in the Statalist FAQ
> <http://www.stata.com/support/faqs/resources/statalist-faq/#toask>
>
>> I am following up on your response to an old thread about
>> multicollinearity (http://www.stata.com/statalist/archive/2010-07/msg00675.html).
>> I liked your response, which goes against the grain of the
>> received wisdom which would have us treat multicollinearity
>> as a problem. Have you or someone else published
>> something to this effect, which I can cite in a paper?
>
> What is recieved wisdom is very much dependent on which
> (sub-(sub-))discipline one belongs to. Here is a fun quote:
>
> "Econometrics texts devote many pages to the problem of
> multicollinearity in multiple regression, but they say little about
> the closely analogous problem of small sample size in estimation a
> univariate mean. Perhaps that imbalance is attributable to the lack of
> an exotic polysyllabic name for 'small sample size'. If so, we can
> remove that impediment by introducing the term micronumerosity."
>
> Chapter 23.3. of Goldberger, A. S. (1991). A Course in Econometrics.
> Harvard University Press, Cambridge MA.
>
> Hope this helps,
> Maarten
*
*   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/