Permutation inference under exchangeability
Distribution-free procedures still require assumptions.
Key knowledge
Under an iid common-distribution null, a two-sample permutation test conditions on pooled observations and compares a statistic across uniformly reassigned sample labels. The resulting null calibration uses exchangeability.
Conditions and scope
- The permitted label permutations must match the null model; clustered or paired data need a compatible design.
- The continuous rank-sum version assumes no ties; nonparametric does not mean assumption-free.
Sources · 1
References
Sources retain their recorded scope. A source link alone does not establish verification of every statement.
Evidence and review
Sections 9.1–9.2: continuous iid rank-sum model and permutation null distribution conditional on pooled values.
2026-10-10 · Two-sample permutation null and continuous rank-sum scope.
Entry details
Permanent link to this entry ↗- Record identity
- math-nonparametric-inference
- Revision
- university-reference.1
- Structured reference record
- University reference