ABSTRACT
As in parametric tests of comparison, the distinction must be made whether the two samples come from independent populations or whether the variates are paired in some way, perhaps by each subject acting as its own control. This obviously depends on the design of the experiment. Hence, here is another example of the design of the experiment and the method of evaluating the results being inextricably linked. Some of the tests which can be used are:
1. Sign test for paired data. 2. Wilcoxon signed rank test for paired data (the Mann-Whitney U-test is
very similar). 3. The Wilcoxon two-sample test for unpaired data.
