E6PythonAdvancedAbout 150 min to complete
Non-parametric Tests
The t-tests and ANOVA of the last two notebooks assume your data is roughly normal with equal variances. Real data — skewed incomes, ordinal survey ratings, tiny samples, outlier-riddled measurements — often breaks those assumptions.
Notebook
About 150 minutes to complete
Enrol to read the notebook and download it to run.
What this module covers
The t-tests and ANOVA of the last two notebooks assume your data is roughly normal with equal variances. Real data — skewed incomes, ordinal survey ratings, tiny samples, outlier-riddled measurements — often breaks those assumptions.
By the end of it
- Explain when and why to prefer non-parametric tests, and the power trade-off involved
- Run and interpret the rank-based alternatives:
- Use Spearman correlation and Fisher's exact test as non-parametric tools
- Report appropriate effect sizes (rank-biserial, ε²)
- Run permutation and bootstrap tests with SciPy — flexible methods that assume almost nothing
- Perform post-hoc comparisons after Kruskal–Wallis with multiplicity correction