RE: st: Stratifying after regression (with 2 interactions)

["Raoul C Reulen" <r.c.reulen@bham.ac.uk>]

Oke, let me clarify this:
This is the model that I am using:

.xi: regress BP i.age*group sexe


Now I want to know how the regression coefficient of the variable group
varies with age.
So I could use .lincom to calculate the different combinations. For
example,
.lincom _age_1Xgroup + group, would give me the regression coefficient
of "group" in age group 1. But if I were to use .xi: regress BP group
sexe if age==1, the regression coefficient of group is different from
above.