For before/after measurements on the same people, should you permute all measurements freely?
StatisticsLevel: IntermediatePermutation experiment design
For before/after measurements on the same people, should you permute all measurements freely?
Hints
- A person is a block.
- Consider sign flips of within-person differences.
Answer
No. Preserve pairs and swap treatment labels within each pair when justified by the null design. Free pooling removes the subject structure.
Source and licence: SciPy — resampling and Monte Carlo inference · BSD-3-Clause · answer: Quantoros editorial