Wilsons confidence intervals query

Dotty

New Member
#1
Hi Everyone

Wilsons Confidence Interval Query

I have a proportion which needs to be x 1000 rather than x 100. Normally I would use Wilsons confidence intervals. What will be the consequence/error of using a Wilsons in this situation. I.e 10 people have an illness out of a total possible of 700. Instead of using the value 1.42% (10/700 * 100), I want to use the value 14.2 per 1000. I've asked my colleagues and I am getting differing answers. Some have suggested Byars but I don't feel that's correct - as it's not technically a rate (i.e. no person time in the denominator). Could anyone advise please?
 

Dason

Ambassador to the humans
#2
I'm not really sure what your question is? I mean the stuff you're talking about is how to present the numbers after the fact. You can create the confidence interval and then (as long as you're interpreting it correctly) translate the endpoints into whatever form you want. It's not like there is anything magic that would prevent somebody that got a confidence interval of (.25, 0.5) from saying (1/4, 1/2) - that's literally just a matter of how you're presenting the numbers. So how you actually make the interval is unrelated to the thing you're asking about in my mind.