Back to Search-quality version diff: v1.1.0 → v1.2.0
Search-quality version diff: v1.1.0 → v1.2.0

Paired Statistical Tests

Paired Wilcoxon signed-rank on per-query RR (B vs A):

  • N_nonzero = 22
  • W+ = 251.50, W− = 1.50
  • Two-sided p = 0.000049
  • One-sided p (v1.2.0 > v1.1.0) = 0.000025

McNemar on rank-1 outcome:

v1.2.0 rank-1v1.2.0 not rank-1
v1.1.0 rank-126 (concordant +)0 (regression)
v1.1.0 not rank-120 (improvement)4 (concordant −)
  • Discordant pairs: b = 0, c = 20
  • McNemar exact (binomial), two-sided p = 0.000002