Mann-Whitney U Test Calculator
| Group | n | Median | Mean | Mean rank | Rank sum | IQR |
|---|
How this is calculated
Rank table
| Observation | Group | Average rank |
|---|
Assumptions and reporting notes
Use the Mann-Whitney U test for two independent groups when observations can be ordered. Observations should be independent within and between groups. The test evaluates whether the two population distributions differ stochastically. Interpreting the result specifically as a difference in medians or location is most appropriate when the two distributions have similarly shaped spreads.
Exact p-values in this calculator are used only when there are no ties and the browser state space remains suitable for live calculation. The asymptotic method uses the standard tie-corrected variance and can optionally apply the 0.5 continuity correction. For paired or repeated measurements, use the Wilcoxon signed-rank test instead.
Verification example
Built-in sample data: Group A = 19, 22, 16, 29, 24 and Group B = 20, 11, 17, 12. With the exact two-sided method, the expected result is UA = 17, UB = 3, and p = 0.1111111111. With the asymptotic method and continuity correction, the expected p-value is 0.1113468865.
