A situation in which there is a specific link between the two observations in each pair, for example if pieces of cloth are divided in two and the two halves of each piece make a pair. Paired-sample tests are generally much more powerful than the analogous two-sample test because they remove a major source of variation. By looking only at the difference in outcome between the two members of each pair, the variability between subjects has been removed, making it easier to identify any genuine difference in the experimental treatments.