Date:         Sat, 28 Mar 2009 11:56:26 -0500
Reply-To:     leah rubin <lhr1807@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         leah rubin <lhr1807@GMAIL.COM>
Subject:      PROC LOGISTIC help with obtaining Odds Ratios for main effects
              when interaction is significant
Content-Type: text/plain; charset=ISO-8859-1

Hi,

I have conducted a logistic regression with 2 predictor variables. The first
has 4 categories and the second has 2 categories (both predictors are
included on the class statement to determine the reference categories). In
addition to including the 2 predictor variables I have also included the
interaction. I used the following statement to ensure that the main effects
enter before the interaction term.

/HIERARCHY= singleclass selection = stepwise include = 2

The problem I am having is obtaining the odds ratios for the main effects
when the interaction is significant.
When the interaction is significant PROC LOGISTIC (version 9.2) is not
producing the odds ratios. I know I can examine the odds ratios for the
interaction using the ODDSRATIO code.

Can someone please help me with the necessary code to obtain all odds ratios
for main effects and interactions?
When the interaction is not significant the odds ratios are given for the
main effects.

Thanks in advance,
leah