06-28-2013 02:27 PM
I seem to remember that there is a way to test all possible combinations of a model all in one statement.
so lets say
proc logistic data=x;
model y= a b c d e f;
it would give you the results for:
Am I imagining things?
07-03-2013 09:29 AM
You can build a model with all interaction terms by using the "bar notation" such as
model y = a | b | c | d;
or restrict to only second order interactions with
model y = a | b | c | d @2;
As Data Null says, you can use variable selection to select a model that incorporates a subset of the terms listed on the model statement.