Logistic Regression using proc genmod

[ajb]

Hello All, 

I'm using SAS, Version 9.2

Variables in my model:

• Event: 0, 1

• height: continuous

• group: 0, 1 (ref=group 0)

• height*group: interaction term

I currently need to estimate the OR for height by different groups

Do I do the following:

Exponentiate height to get OR for group = 0

Exponentiate height height*group to get OR for group =1

proc genmod data=temp;

class group;

model event(ref="0")=height group height*group/ link=logit dist=binomial;

estimate "height group 0" height 1 / exp;

estimate "height group 1" height 1 height*group 1/exp;

run;

I also have a model with 3 groups instead of 2.

• Event: 0, 1

• height: continuous

• group: 0, 1, 2 (ref=group 0)

• height*group1 

• height*group2

would the code be 

proc genmod data=temp;

class group;

model event(ref="0")=height group height*group/ link=logit dist=binomial;

estimate "height group 0" height 1 / exp;

estimate "height group 1" height 1 height*group 1 0 0 /exp;

estimate "height group 1" height 1 height*group  0 1 0 /exp;

run;

[tarheel13]

http://support.sas.com/kb/42/728.html