The Friedman test is used to detect any difference between subjects under test measured variously multiple times. More precisely, this non-parametric test states whether there is a significant difference in the location parameters of k statistical samples (>= 3, columns candidates , treatments , subject ), measured n times ( rows , blocks , participants , measures ), or not. The data in each row is ranked, based on which a resulting test statistic Q is calculated.
If n > 15 or k > 4, the test statistic Q can be approximated to be Χ 2 distributed. With the given significance level α, a corresponding p -value (null hypothesis H 0 : there is no difference of the location parameters in the samples, alternative hypothesis H A : the samples in the columns have different location parameters) can be given.
Please refer also to the Wikipedia description of the Friedman Test .