I have two possible exposure variables (A and B) for use in a statistical model predicting a binary health outcome. I have fitted models with each variable separately and now know that one variable is a better predictor of my outcome based on a comparison of R-squared and BIC values. However, the ORs for the two variables are quite similar with overlapping confidence intervals. The sample composition and size is the same for each model.

I would like to comment on whether the odds ratios are significantly different from one another. How can I do this?

Best Regards.


It looks like you are trying to compare two odds ratios. Try looking at

Statistical test for difference between two odds ratios?

  • $\begingroup$ In this case the 2 ORs are not independent, I'd say. $\endgroup$ – boscovich Oct 7 '12 at 20:03
  • $\begingroup$ Hi - yes I did see the earlier post, but as a novice I need something with bit more explanation! $\endgroup$ – N26 Oct 7 '12 at 20:51
  • $\begingroup$ As pointed out above, can you assume independence? $\endgroup$ – Ken Dunn Oct 7 '12 at 20:58
  • $\begingroup$ I'm not sure. How would I evaluate that? Exposure A and Exposure B are correlated to some extent, but rho is about 0.5. $\endgroup$ – N26 Oct 8 '12 at 9:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.