Re: Fisher's Exact Test

by resonator (Initiate)
 on Apr 18, 2007 at 17:41 UTC ( #610806=note: print w/replies, xml ) Need Help??

in reply to Fisher's Exact Test

Hi.

Have you checked the results of your code for Fisher's Exact Test? I don't think it produces correct results.

Here's a utility that produces results identical to those I calculate: http://www.exactoid.com/fisher/index.php

This expression at the beginning of your code seems odd:

```\$test = \$a*(\$a+\$b+\$c+\$d ) - (\$a+\$b)*(\$a+\$c);
It simplifies to: \$test = \$a*\$d - \$b*\$c;

I don't follow why you have this:

```return 1 if \$test < 0 and \$ts;
For example, if a = 1, b = 50, c = 10 and d = 5, P(one-sided) = P(two-sided) = 1.4386e-07, not 1.

Nevertheless, thanks for the tip about calculating factorials only once. Let me know if I've misunderstood something.

Replies are listed 'Best First'.
Re^2: Fisher's Exact Test
by Anonymous Monk on Aug 20, 2012 at 23:34 UTC
I think it is correct. It is justified because \$ts implies one-side test, so the p-value is indeed 1. Hao
I think it is wrong. \$ts means two-side, not one-side.

Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: note [id://610806]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others chanting in the Monastery: (2)
As of 2023-02-08 04:50 GMT
Sections?
Information?
Find Nodes?
Leftovers?
Voting Booth?