Hi Samir
Thursday, May 17, 2007, 9:22:54 PM, You wrote:
SO> just wondering about Yates correction. Is it possible to do it for
SO> 2x3 tables? What should I do if I have few cells in 2x3 table with
SO> expected counts less then 5. I am doing Chi square, but you know
SO> that probably.
Although Yates' correction for contingency tables greater than 2x2 was
used in the past, it is considered an incorrect practice now. If you
have more than 20% of the cells with expected counts less than 5
and/or expected counts less than 1, you should use the extension of
Fisher's exact test (I think that someone in this same list said its
correct name was "Fisher-Freeman-Halton"). With SPSS, you need the
EXACT TEST module installed. If you don't have it, then you can use a
freeware program called EXACT2xk.EXE (you can imagine its purpose...),
one of the PEPI programs, downloadable at:
http://www.sagebrushpress.com/pepibook.htmlYou can download the whole package (43 programs) or individually.
If you are going to use PEPI, then I recommend you to select "mid-p
exact p-value" (more sensitive) instead of Fisher's (too
conservative). A good reference and worked example of the use of mid-p
exact p-values can be found at Statistics at Square One (chapter 9,
right near its end):
http://www.bmj.com/collections/statsbk/9.dtlIf, nonetheless, you still want Yates correction for your contingency
table, just let me know, because writing some MATRIX code for it is a
piece of cake.
--
Regards,
Dr. Marta García-Granero,PhD mailto:
[hidden email]
Statistician
---
"It is unwise to use a statistical procedure whose use one does
not understand. SPSS syntax guide cannot supply this knowledge, and it
is certainly no substitute for the basic understanding of statistics
and statistical thinking that is essential for the wise choice of
methods and the correct interpretation of their results".
(Adapted from WinPepi manual - I'm sure Joe Abrahmson will not mind)
Locked
