Fisher's exact test

n. A test of the significance of relationship between two categorical variables which makes no assumptions about the population distribution and is appropriate for small samples for which a chi-square test would be inappropriate.