Barankay, Iwan T wrote:
> I am doing some joint tests on coefficients of highly
> correlated variables (like x x^2 x^3 x^4 etc.) with
> testparm. The output of testparm tells me that some
> constraints are dropped. I understand (right?) this
> is done to get a nonsingular matrix. Yet, what I would
> like to know is what the pattern or structure is with
> which Stata decides which constraints to drop?
Iwan:
The easy answer is that there is not really a meaningful
pattern. There shouldn't be since if Stata drops a
variable because of multicolinearity than that means it
is unable to distinguish between variables. If it is
unable to distinguish between variables than why should
Stata prefer to keep one variable over the other?
If the correlation between variables is due to the fact
that the variables are polynomials, than you might want
to look at the -orthpoly- command, which creates
orthogonal polynomials.
HTH,
Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z214
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
*
* 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/
