Loading...
Loading...
Non-parametric test to compare two independent groups. No normality assumption needed.
Enter two groups of data and click Run Test.
Non-parametric alternative to the independent samples t-test
Use Mann-Whitney when: (1) data is not normally distributed, (2) you have ordinal data, (3) sample sizes are very small, or (4) data has significant outliers. Use t-test when data is approximately normal and you want maximum statistical power.
U counts how many times a value in one group exceeds a value in the other. U₁ + U₂ = n₁ × n₂. If groups don't overlap at all, U = 0 (or n₁n₂). If perfectly mixed, U ≈ n₁n₂/2.
r = |z| / √N is the rank-biserial correlation. Guidelines: r < 0.1 = negligible, 0.1-0.3 = small, 0.3-0.5 = medium, > 0.5 = large. It measures how much the distributions are separated.
Yes. Tied values receive the average of the ranks they would occupy. A continuity correction is applied to the z-approximation. For many ties, an exact test or permutation test is preferable.
Yes, they are mathematically equivalent. Mann-Whitney uses U statistic; Wilcoxon rank-sum uses W (sum of ranks). scipy's mannwhitneyu() and ranksums() give the same p-value.