Like any statistical test, the Mann‐Whitney U has forces
and weaknesses. In terms of forces, like any non‐parametric
test, the Mann‐Whitney U does not depend on assumptions
on the distribution (i.e. one does not need to postulate the
data distribution of the target population). One can also use
it when the conditions of normality neither are met nor
realisable by transformations. Moreover, one can use it
when his sample is small and the data are semi‐quantitative
or at least ordinal. In short, few constraints apply to this test