- #1
mnb96
- 715
- 5
Hello,
I have a 16x16 contingency table, and I would like to perform a test in order to reject the hypothesis of independence between the variable X (columns) and the other variable Y (rows).
Unfortunately many cells in the table contain zeros. The zeros are due to the fact that the sample size was not sufficiently big to properly populate the table (I had "only" 200 samples, while the cells of the table are 256).
In this case I cannot use neither the χ2-test, nor the χ2-square test with Yate's correction, nor the G-test, because I obviously get divisions by zero.
What kind of test can I perform in such situations?
I have a 16x16 contingency table, and I would like to perform a test in order to reject the hypothesis of independence between the variable X (columns) and the other variable Y (rows).
Unfortunately many cells in the table contain zeros. The zeros are due to the fact that the sample size was not sufficiently big to properly populate the table (I had "only" 200 samples, while the cells of the table are 256).
In this case I cannot use neither the χ2-test, nor the χ2-square test with Yate's correction, nor the G-test, because I obviously get divisions by zero.
What kind of test can I perform in such situations?