1
$\begingroup$

How can I get the standard error for odds ratios. I know how to obtain ORs, either by exp the estimate or by using LSMEANS or SLICE statements.

Unfortunately wasnt able to find the options to have SE for ORs in the output datasets. Only the estimate SE is in there.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

You can use the delta method for that. Lets call the raw coefficient or log odds ratio, $\beta$, and its standard error $se_\beta$, the odds ratio $OR$, and its standard error $se_{OR}$.

You have $\beta$ and $se_\beta$ and you know that $OR=\exp(\beta)$.

Per the delta method, $se_{OR} \approx \frac{\partial \exp(\beta)}{\partial \beta}se_\beta = \exp(\beta)se_\beta$

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thank you so much! This did it. Same approach for rel risk out of GENMOD with poisson and bin $\endgroup$
    – joe
    Commented Sep 12, 2013 at 16:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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