a statistical test of one hypothesis against another that has the greatest
power among all tests available at a given
alpha value. Suppose the two hypotheses are the
null hypothesis of no difference between two groups (Sample 1 mean = Sample 2 mean) and the
alternative hypothesis of some difference between the groups (Sample 1 mean ≠ Sample 2 mean). Because the latter is actually a composite of several possibilities, a UMP test is one defined by a
critical region that is best suited to test the null hypothesis against each of the simple hypotheses comprising the larger alternative hypothesis.