Re: st: RRR with CI from logit model

[Michael Ingre <Michael.Ingre@ipm.ki.se>]

On 2004-11-01, at 10.41, maartenbuis wrote:

> Second, you can confidense intervals of relative risk
> ratios with -nlcom-.
Thanks Maarten. Your suggestion seems to do the trick:

nlcom exp(81 * squared + 9 * linear + constant) /              ///
         (1 + exp(81 * squared + 9 * linear + constant)) /    ///
      exp(4 * squared + 2 * linear + constant) /            ///
	     (1 + exp(4 * squared + 2 * linear + constant))

The above code gives the expected RRR for 9 vs. 2 on the scale. 
However, I was a bit disappointed with the statistical power. A linear 
combination (-lincom-) for the odds ratio was significant with p=.007 
but the RRR was far from significant with p=.666.

I guess this is one reason why RRRs are seldom reported with this kind 
of models.

Michael

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/