turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- odds ratio estimates for interaction terms with GL...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

12-22-2014 08:57 PM

Hello,

I was wondering how to obtain OR estimates and confidence intervals for a fixed effects model I am fitting with GLIMMIX (GLM with logit link)- so basically a logistic regression.

I am able to obtain predicted probabilities at each level of the interaction with the lsmeans statement:

lsmeans var_a*var_b/ilink cl, and realize that I can probably exponentiate the parameter estimates for each interaction term to obtain odds ratios (provided I've set my referent group for both var a and var b), but the confidence intervals are tricky. Can I use a command or option in the lsmeans statement to automate this and make it easier?

Thanks!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

12-30-2014 09:41 AM

The LSMEANS options DIFF, OR and CL will give you the odds ratios with confidence limits. It is harder for interactions, but I suppose that by using the SLICEDIFF option you can narrow this to the odds ratios of interest.

Steve Denham