Date:         Wed, 10 Nov 1999 08:32:44 +0100
Reply-To:     "Dr. Robert Nemeth" <Dr._Robert_Nemeth%FOCUS@NOTES.INTERLIANT.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Dr. Robert Nemeth" <Dr._Robert_Nemeth%FOCUS@NOTES.INTERLIANT.COM>
Subject:      Re: Freq/Rel.Risk Stats Question
Comments: cc: lnonn@SOC.DUKE.EDU
Content-type: text/plain; charset=us-ascii

Lynn,

the displayed odds ratio is generally supposed to be log-normally
distributed (i.e. the log is normally distributed).
the estimate is

x=ad/bc

(where a, b, c, d denote the cell frequencies, as usual)

the variance of log(x) is given by

1/a + 1/b + 1/c + 1/d

and the 95% CI is computed with this value

Regards

Robert