Chi-Square and Contingency Table

The Chi-Square procedure tests if there is association between the classification of one variable and the classification of another variable. The data must be summarized and presented in a table form where the cells contains frequency counts for various levels. The null hypothesis states that no association exists. ProcessMA highlights to you the number of cells with expected frequencies less than 5. This is because the chi-square test may not be reliable is more than 20% of the cells are less than 5 and especially when they have a significant contribution to chi-square statistics (small p value). An alternative method is to combine categories so that cells with small expected frequencies are treated as one.