How to compare categorical data for more than 2 groups?

#1
I'm trying to compare categorical data between 4 groups
i.e :
smoking history (yes or no) between 4 groups of cancer staging (stage I to IV)
Some of the cells has 0 value in it (every person in stage I never smoke, while stage II has 5 smokers, etc)
Can I still use Chi-squared test for this hypotheses? I understand that you can't have 0 value in one of the cells for Chi-squared test.
I can't use Fisher test also because it is not a 2x2 table.
Can anyone show me the appropriate test?
What i'm trying to find out is there a difference in smoking history between persons with different stages of cancer.
Thank you
 
Last edited:

trinker

ggplot2orBust
#2
Two ideas come to my mind: You could combine rows or columns (that's the easiest way) or you could use a randomized test.

I think multinomial logistic regression could also work.
 

Karabiner

TS Contributor
#4
Can I still use Chi-squared test for this hypotheses? I understand that you can't have 0 value in one of the cells for Chi-squared test.
That is wrong. It does not matter if you have cells with frequency = 0. But you must
not have cells where the EXPECTED frequency is = 0.

Regards

K.