Re: st: Interacting dummies and multicollinearity issue

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

--- On Wed, 23/6/10, Özlen D. Luznar wrote:
> variable d - has 3 values - dummied out to d1 d2 d3
> variable e has 4 values dummied out to e1 e2 e3 e4
> 
> In addition to include d1 d2 e1 e2 e3 in my model I want to
> include interaction terms between d dummies and e dummies:
> 
> 12 possible variables interaction between d dummies and e
> dummies.  I am confused as to which of these 12 I can/should
> include in the model so as to avoide multicollinearity.

Martin correctly pointed out -help fvvarlist-. If you have
Stata 11, you should definately use the factor variable 
notation. 

You can also do it manually. All you need to do is leave
for each variable one dummie out, and create product terms
for the remaining dummies. 

The real problem is the interpretation. The best solution 
is to use the factor variable notation in combination with 
-margins-. If you don't have Stata 11, then these 
interpretations usually get so messy that the best way 
forward is to find multiple big sheets of paper and do the 
math.

Hope this helps,
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/