Minimum n for Chi square Test?

#1
Hey folks,

I'm currently evaluating the data of a study we performed.
Just got a very simple question: Is the a minimum number of subject neccessary to perform a Chi square test. I mean, I believe there should be, as you are splitting the group in 4 groups when performing the test...
But does anybody know wheather there is such a cut - off number?

Would be glad for any help you might offer,

Regards

vier.gewinnt
 

TheEcologist

Global Moderator
#2
The minimum number of groups required to do a chi-squared is 2. 2 observed groups and 2 expected values.

Look at the formula for the degrees of freedom, it cant be less than 1.
http://en.wikipedia.org/wiki/Chi-square_test

Now there is a recommendation: namely the minimum expected value in any group should never be less than 5.

I hope I got your question because:

I mean, I believe there should be, as you are splitting the group in 4 groups when performing the test.
Is kind off vague.