Randomization tests are non-parametric statistical tests. They are based on the ranking of statistics, such as the t or F value for the comparison of two or more means, respectively. It has been shown that randomization tests combine the advantages of non-parametric and parametric methods. Because they are based on ranking as are other non-parametric tests, the influence of outliers is minimal and a normal distribution is not presupposed. On the other hand, they are as efficient as parametric methods, because they use all the information in the data.