Which of the following is a reason not to use a chi-square test of homogeneity to analyze a set of data?
a) The data is categorical
b) The sample size is small
c) The variables are independent
d) The expected cell frequencies are too low