Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It only takes a minute to sign up.
Sign up to join this community
I have a question about conducting independence tests for categorical data with Fisher's Exact when I have zeros in my data. Since I observed no data for a category in both groups, is it right to include zero? Is this the best I can do with some empty categories?
Take the following R code for instance which represents my two groups of 7 categories:
v1 <- c(31, 9, 6, 0, 5, 0 27)
v2 <- c(13, 0, 1, 0, 1, 2, 7)
v <- cbind(v1, v2)
fisher.test(v)
p-value = 0.08061
A row (or column) of $0$'s in a Fisher test don't contribute anything to the calculation (it shouldn't change the probabilities of the various tables with the same margins as the sample table if you remove any such rows and/or columns).
As a result it should make no difference, but personally I'd remove them (it may help avoid a number of possible issues, including programs that might balk at a particular table size but not one a little smaller).
