RE: Output of SE in assoc.logit1

[Jacobs, Kevin (NIH/NCI) [C]]

You are correct. The confidence bounds on the OR are asymmetric and can be derived from the SE of the log(OR).

Best regards,
-Kevin

-----Original Message-----
From: Mark [mailto:m.bro...@gmail.com]
Sent: Monday, March 19, 2012 9:28 AM
To: glu-users
Subject: Output of SE in assoc.logit1

Hi

I am using the --se option in the assoc.logit1 module.

While the output gives the OR and the --ci option gives the intervals around this OR, the standard error given seems to be the standard error on the estimate, i.e on log(OR), rather than the standard error on the OR. Is this correct?

Mark

[Mark]