None of 999 random permutations is as extreme as the observed statistic. Should the one-sided p-value be reported as…
StatisticsLevel: IntermediatePermutation experiment design
None of 999 random permutations is as extreme as the observed statistic. Should the one-sided p-value be reported as zero?
Hints
- Include the observed arrangement.
- Use both numerator and denominator corrections.
Answer
No. The plus-one correction gives (0+1)/(999+1)=0.001. Finite Monte Carlo sampling cannot justify an exact zero probability.
Source and licence: SciPy — resampling and Monte Carlo inference · BSD-3-Clause · answer: Quantoros editorial