In chi-sequre test of contingency table, is appropriate to perform test without clubbing up expected values which are less than 5?? I mean in exam do we take care of less than 5 expected values as we do in chi-sequre Fit Data test. Please clarify
NO... In goodness of fit and contingency test, we need to combine when expected values are less than 5! Since smaller expected values will give higher value of our test statistic
