Re: Fisher exact test with large N and rare events
- From: Jack Tomsky <jtomsky@xxxxxxxxxxxxx>
- Date: Thu, 27 Jul 2006 20:55:08 EDT
John Uebersax :
Under *Fisher exact test* I found (Web) a software
(author: Oyvind Langsrud) providing:
For both 2x2 Tables
_____T1 = [2, 17 ; 60, 80900]
_____T2 = [60 , 80900 ; 2 , 17]
___p(right tail) = 0.00009´77984
Then you can estate (with high probability) that
____w1 = 2 / (17 + 2) = 0.1053
____w2 =60 / (60 + 80900) = 0.0007
______w1 > w2.
______licas (Luis A. Afonso)
I can state with probability one that 0.1053 > 0.0007. This inequality comes from simple arithmetic.
Jack
.
- References:
- Re: Fisher exact test with large N and rare events
- From: \"Luis A. Afonso\"
- Re: Fisher exact test with large N and rare events
- Prev by Date: Re: Fred Mosteller [1916-2006]
- Next by Date: Re: Simple question
- Previous by thread: Re: Fisher exact test with large N and rare events
- Next by thread: Expected Value of R^2 for Fixed Regressors
- Index(es):
