nonparam
Mann–Whitney U
Compare two independent groups when rank/ordinal behavior is more appropriate than means.
Requirements
- ordinal or continuous outcome
- two independent groups
Avoid when
- paired data
- many ties without exact handling
Formula
U₁ = n₁n₂ + n₁(n₁ + 1)/2 − R₁
Interpretation
Tests a shift in distributions; often summarized with rank-biserial correlation.
Worked example
Compare ordinal satisfaction scores between two independent care pathways.
Open example datasets