A test of a statistical hypothesis in which this hypothesis is rejected only for sufficiently large values of the test statistic, or only for sufficiently small values. It is employed when the direction of the effect tested is known a priori. See also two-tailed test.