# Verification checks (P4)

- Clean re-run of out/analysis.py: results.json byte-identical (cmp) -> PASS
- se positive: PASS
- CI contains tau: PASS
- CI symmetric: PASS
- tau = mean diff: PASS
- significance consistent with p and CI: PASS
- n_analysed = n_raw - removed (1000-0): PASS
- n_schools*25 = n_analysed: PASS
- 0<=icc<=1: PASS
- se_cluster/se_iid near sqrt(DEFF) (within 5%): PASS
- CR1 SE close to school-means Welch SE (within 5%): PASS