Pairs of experimental units chosen to resemble each other as closely as possible. Alternatively, the phrase may refer to single experimental units measured at two points in time, or in two different situations. See also experimental design; hypothesis test; Wilcoxon signed-rank tests.