|
Suppose I have two nominal variables A and B and one continuous variable x. I
have a contiuous response y. I fit the following model:
proc glm;
class A B;
model y = x | A | B;
run;
In effect, I am regressing y on x for each combination of A and B. I would
like to do pairwise comparisons of beta_{ab}, the expected effect of x on y
when A=a and B=b. As I understand it, means and lsmeans only work on nominal
variables. Does glm do this type of test?
Thank you,
Edward Malthouse
Assistant Professor
IMC Department
Medill School of Journalism
Northwestern University
Evanston, Illinois
ecm@nwu.edu
| 
