--- On Mon, 16/2/09, Michael Stobernack wrote:
> Another possibility could be to incorporate dummy variables
> and interaction terms for the different groups of interest.
> Since I have got four different groups I need three dummies
> and many interactions terms, so I loose (to) many degrees of
> freedom (n=50).
By estimating separate models for each group you actually
estimate 4 extra coefficients, i.e. loose 4 degrees of
freedom, compared to a model with all the interaction
effects. The reason is that in the interaction model you
constrain the residual variance to be equal across groups.
-- Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting address:
Buitenveldertselaan 3 (Metropolitan), room N515
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
*
* 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/
