topic Logistic Regression and Interactions in Statistical Procedures

Logistic Regression and Interactions

lefty2108 — Thu, 22 Sep 2011 20:39:38 GMT

So I'm fairly new to both logistic regression and SAS, so I was a bit stumped when ran a proc logistic and found a significant 2-way interation. The variables in question are "age" (continuous) x "treatment" (dichotomous). Anyone have any suggestions as to what my next step would be in SAS in order to evaluate this interaction?

Logistic Regression and Interactions

DLing — Thu, 22 Sep 2011 20:47:26 GMT

Not sure what your question is:

- Were you not expecting any interaction but there turned out to be? (Why surprised?)

- Are you wanting to understand the strength of the interaction in relation to other factors? (Start with looking at standardized estimates)

In analytics, finding the unexpected sometimes is a good thing. It's a discovery of new knowledge.

Re: Logistic Regression and Interactions

lefty2108 — Thu, 22 Sep 2011 20:59:33 GMT

Interactions in this case were entirely possible. My question was more about what post-hoc analyses I might employ using SAS in order to understand the nature of this particular interaction. Thanks for the replies btw.

Re: Logistic Regression and Interactions

Reeza — Thu, 22 Sep 2011 22:12:12 GMT

Have you looked at the odds ratio with the oddsratio statement?