According to chi-square testing, interval data has to be tested using a cross-tabulation approach O each cell should have a minimum 5 observations O if more than half of the cells have actual counts less than 10, then results will not be reliable O when the differences between the actual and expected frequencies is small, then there is not a good fit