A chisquare test may not be valid when options The expected

A chi-square test may not be valid when...

options: The expected frequencies are quite different across categories (one is 4 times greater than another). The observations are independent of one another. The observed frequencies for some cells are quite small (<5). The expected frequencies for some cells are quite small (<5).

Solution

A chi-square test may not be valid when The expected frequencies for some cells are quite small (<5). Then Yates correction is needed. This is done by subtracting 0.5 from the absolute value of O-E before squaring.