A non-parametric test (see non-parametric methods) for testing the null hypothesis that two independent samples of size n and m are from the same population. The observations are combined and ranked. If the null hypothesis is true, the sum U of the ranks of the observations in the sample of size n has mean n(n + m + 1)/2 and has variance nm(n + m + 1)/12. For even relatively small values of n and m, the value of U can be tested against a normal distribution. This is also known as the Mann–Whitney U test.