Randomization tests are the procedures for determining the statistical significance directly from the data with- out recourse to some particular sampling distribution. For instance, in a study including the comparison of two groups, the data would be splittes (permuted) repeatedly between groups and for each split (permutation) the relevant test statistic (for instance, a t or F), is calculated to determine the proportion of data permutations which provide as large a test statistic as that associated with observed data. If that quantity is smaller than some significance level α, the results are important at the α level.