Re: Antwort: Re: st: Multicollinearity in panel data

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Wed, 9/2/11, Justina Fischer wrote:
> would an alternative way of dealing with a high correlation between
> x and x*x and the loss in precision be to conduct tests of joint 
> significance (for both coefficients) ?

If you can estimate the model then this might work. However, sometimes
the correlation becomes so strong that Stata will drop one of the 
variables. Recently there was such a case on Statalist where x was
year of birth and the poster wanted to add x, x^2 and x^3 and ended up
with this problem. Than, obviously, no amount of joint testing can save 
you (but centering your year of birth variable prior to creating the
squares and cubes can).

-- Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/