A statistical test that does not make any assumptions about the distribution underlying the sample data. It is generally used to analyze data from a process whose distribution is unknown or obviously non-normal, or when the assumptions of the corresponding parametric test are violated. Although a non-parametric test is generally less powerful than its parametric counterpart when the parametric test is valid, it provides a robust alternative to the parametric test in case of such violations.
Non-parametric tests may not make distributional assumptions, but they do make other assumptions, namely equal Variances across Populations and independence of the sample observations and in some cases, symmetry of the population distribution(s).
Examples of non-parametric tests:
1. Binomial Test
2. Sign Test
3. Wilcoxon Signed Ranks Test
4. Mann-Whitney-Wilcoxon (Wilcoxon Rank Sum) Test
5. Kruskal-Wallis Test
6. Friedman Test