
FISHER EXACT TEST 65
the alternative hypothesis is that column proportions are not all equal. For a single
sample, the null hypothesis is that there is no association between the measured vari-
ables (i.e., they are independent), whereas the alternative hypothesis states that an
association exists.
The assumption of validity of the chi-square test is that the expected count under H
0
for each cell in a contingency table is at least 5. The expected count in a cell is de-
fined as the product of the corresponding marginal totals divided by the grand total.
If this assumption is violated and some expected counts are below 5, a nonparamet-
ric Fisher exact ...