Use rank-based methods when your data violates normality or when you have ordinal outcomes. Explore Mann-Whitney U, Wilcoxon signed-rank, and Kruskal-Wallis tests.
Sign in to track progress
Each dataset exhibits a different reason to prefer nonparametric methods.
n=40 observations · showing first 10
| Index | Group 1 | Group 2 |
|---|---|---|
| 1 | 2 | 5 |
| 2 | 3 | 6 |
| 3 | 4 | 7 |
| 4 | 4 | 8 |
| 5 | 5 | 9 |
| 6 | 5 | 11 |
| 7 | 5 | 14 |
| 8 | 6 | 17 |
| 9 | 7 | 22 |
| 10 | 8 | 29 |
Check normality before choosing a test
Distribution
Q-Q Plot
Click "Run Assumption Check" to test for normality using the Q-Q plot correlation (how closely the sorted data track normal quantiles).
Compare the t-test and Mann-Whitney U results side by side.
Parametric
Independent t-test
Nonparametric
Mann-Whitney U
| Aspect | Parametric | Nonparametric |
|---|---|---|
| Test Used | t-test | Mann-Whitney U |
| Statistic | t = -2.81 | U = 99 |
| p-value | 0.010 | 0.006 |
| Assumption Met? | NO | N/A |
| Reliable? | QUESTIONABLE | YES |
Answer 5 questions to find the right test
How many groups are you comparing?
Step-by-step breakdown of each nonparametric test
When to Use
Two independent groups (e.g. Group A vs Group B). Tests whether the rank distribution of one group is stochastically greater than the other.
Calculation Steps