I got a data that can be categorized in two different ways and I like to test their association. The initial thought is Chi-squared test. However, the data is not normally distributed. What is the prefered test on such non-normally distributed contingency table?

  • $\begingroup$ contingency tables are counts; you should be fine to use a chi-squared test on 2x2 table of counts as long as none of the expected values are quite small. $\endgroup$ – Glen_b Aug 2 '19 at 6:47
  • $\begingroup$ What if my data is not normally distributed? Would that violate the chi-squared distribution assumption? $\endgroup$ – unicorn Aug 2 '19 at 8:40