Non-parametric Test Suite

Overview

The Non-parametric Test Suite provides a comprehensive collection of rank-based statistical tests that work without assuming normal distributions. These powerful alternatives to parametric tests are especially valuable when dealing with ordinal data, small samples, or distributions with outliers. Explore Mann-Whitney U, Wilcoxon signed-rank, Kruskal-Wallis, and Friedman tests to see how rank transformations enable robust hypothesis testing.

Open in new tab

Tips

• Mann-Whitney U tests if two independent groups differ - use it instead of a t-test when normality fails
• Wilcoxon signed-rank tests paired differences - perfect for before/after comparisons without normality
• Kruskal-Wallis extends Mann-Whitney to 3+ groups - it’s the non-parametric alternative to one-way ANOVA
• Friedman test handles repeated measures across 3+ conditions - use when repeated measures ANOVA assumptions fail
• All these tests work by ranking the data, making them resistant to outliers and skewness
• View the rank transformations to understand how the tests work with ordinal information
• Effect sizes (like rank-biserial correlation) help interpret the practical significance
• Compare to parametric equivalents to see when non-parametric tests are more appropriate
• These tests lose some power when data truly are normal, but gain robustness otherwise