[Corpora-List] how to calculate Chi-square criterion correctly?

[Nicolas Mazziotta]

Le dimanche 18 mai 2008 22:01, Yuri Tambovtsev a écrit :
> Dear Corpora colleagues, do you know if the modern way of calculating the
> Chi-square criterion is different than it was some 30 years ago? Or is it
> the same? For example, I calculate the Chi-square between 22 and 8 as 6.53.
> Thus, at the 5% level the theoretical value is 3.84 at one degree of
> freedom. It is much less than 6.53. Therefore, the difference is
> statistically significant.Is it a correct way of calculating the Chi-square
> criterion? Looking forward to hearing from you soon to yutamb at mail.ru Yours
> sincerely Yuri Tambovtsev, Novosibirsk, Russia = yutamb at mail.ru

Dear Colleague,

It indeed seems that things have not changed much. But I am not sure this is 
relevant in your case. If your data is a simple 2x1 table, the chi-square 
coefficient might not be what you need. What do you want to show?

You can find the current state of the art about categorical data analysis in 
A. Agresti. 2002. Categorical Data Analysis. 

Best Regards,

-- 
Nicolas Mazziotta, Assistant
Université de Liège (bât. A2)
Langues et Littératures Françaises et Romanes
Linguistique française et dialectologie wallonne
Place Cockerill, 3-5/45; B-4000 Liège (Belgium)
nicolas.mazziotta at ulg.ac.be - tél. +32/(0)4.366.53.82

The contents of this e-mail, including any attachments, are confidential and 
are intended for the sole use of the individual or entity to which it is 
addressed. Any distribution, copying or dissemination of this message is 
expected to conform to all legal stipulations governing the use of 
information.