Friedman

The Friedman test is designed for situations where you need to compare three or more related groups—that is, when the same subjects, units, or entities are measured under different conditions. It is the non‑parametric counterpart to repeated‑measures ANOVA and is particularly useful when data is ordinal, non‑normal, or affected by outliers. 

Common applications include comparing operators performing the same task, evaluating multiple methods on the same parts, or assessing performance across repeated trials. Because the test accounts for the fact that measurements are related, it provides a more accurate assessment of differences than tests designed for independent samples. 

The Friedman test works by ranking the values within each subject or block. For each block, the lowest value receives rank 1, the next lowest rank 2, and so on. The ranks are then summed across conditions, and a chi‑square statistic is used to evaluate whether the rank distributions differ significantly. A significant result indicates that at least one condition differs from the others. 

As with Kruskal‑Wallis, the Friedman test does not identify which groups differ. Post‑hoc tests—such as pairwise Wilcoxon signed‑rank tests with appropriate adjustments—are needed to pinpoint specific differences. 

The strength of the Friedman test lies in its robustness. It does not require normality, equal variances, or interval‑scale data. It handles skewed distributions, ordinal ratings, and outliers with ease. This makes it ideal for real-world processes where repeated measurements often violate parametric assumptions. 

In the Analyze phase, the Friedman test helps you evaluate repeated‑measures scenarios with confidence. It provides a structured, assumption‑light way to compare multiple conditions and identify meaningful differences in performance. When used thoughtfully, it becomes a powerful tool for uncovering insights in complex, dependent data structures.