Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: How to reduce correlation between variables


From   "Justina Fischer" <JAVFischer@gmx.de>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: How to reduce correlation between variables
Date   Tue, 07 Aug 2012 11:59:42 +0200

I completely agree with Nick. The only feasible technical solution I could think of is to assume different functional forms (based on some theory) - that is anlinear function in one variable but a log-function in the other.

Economists would often assume a log-relation for the independent 'income', based on the theory of decreasing marginal returns to income.

Justina 


-------- Original-Nachricht --------
> Datum: Tue, 7 Aug 2012 10:50:33 +0100
> Von: Nick Cox <njcoxstata@gmail.com>
> An: statalist@hsphsun2.harvard.edu
> Betreff: Re: st: How to reduce correlation between variables

> I agree with Maarten and Joseph. I note also that Ebru is referring to
> correlation between variables as posing a problem but doesn't say what
> that problem is. Personally I am more wary of a model being
> unnecessarily complicated than of the bugaboo called
> multicollinearity. If two or more predictors are highly correlated you
> can use scientific judgement to choose which makes more sense or
> choose according to which performs better as a predictor. Neither
> strategy requires an extra command, but rather a willingness to think
> about what you are doing rather than treating analysis as purely
> mechanical.
> 
> Nick
> 
> On Mon, Aug 6, 2012 at 7:38 PM, Maarten Buis <maartenlbuis@gmail.com>
> wrote:
> > On Mon, Aug 6, 2012 at 7:24 PM, Ebru Ozturk wrote:
> >> They are still correlated. Sample size is 285 firms and I cannot
> increase it. Is there other way to sole this problem?
> >
> > The new variables created by -orthog- shouldn't be correlated. Can you
> > show us the exact command you used (*) and the output for -corr- among
> > the new variables created by -orthog-?
> >
> > -- Maarten
> >
> > (*) the command you gave us some time ago would have produced an
> > error, so that cannot be the one.
> *
> *   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/

-- 
Justina AV Fischer, PhD
COFIT Fellow
World Trade Institute
University of Bern

homepage: http://www.justinaavfischer.de/
e-mail: javfischer@gmx.de. justina.fischer@wti.org
papers: http://ideas.repec.org/e/pfi55.html


*
*   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   |   Site index