Re: st: Interacting the main regressor with age in cnreg

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

--- On Sat, 18/12/10, Jen Zhen wrote:
> I have run the following estimation, where "age" is
> centered around the sample mean of "Age", and "Age" is what
> it suggests it is:
> cnreg outcome X age age2 age3 year_*, censored(cens)
> 
> Now I wanted to see whether the marginal effect of "X"
> varies by "Age" or, for that matter, by "age". Hence, on top
> of controlling for age, age^2 and age^3, I decided to
> include also interactions of the latter three variables with
> "X":
> cnreg outcome X X_age X_age2 X_age3 age age2 age3 year_*,
> censored(cens)
> 
> I would have expected the coefficient on "X" to be the same
> as in the first equation, seeing that at the sample mean "age",
> "age2" and "age3" and hence also the three interactions of "X"
> with these should be zero anyway. Nonetheless the coefficient
> which I now get for X is way larger, and hence the predicted
> marginal effect of X is very different from what I had obtained
> previously.

The effect of X means something different in both regression, in
the first model is the effect of X assuming it is constant over 
age, while for the latter model it is the effect X for someone of
average age, assuming the effect X can change non-linearly over 
age. So there is no reason why the two should be the same.

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/