# Verification checks (P4)

- Clean re-run of out/analysis.py reproduced out/results.json byte-identically (cmp): PASS
- CI contains estimate: PASS
- CI half-width = 1.96*SE: PASS
- derived 95% CI excludes 0, consistent with ** (p<0.05): PASS
- wage CI contains ATE: PASS
- wage CI near-symmetric (|asym|<=2): PASS
- derived SE positive: PASS
- implied p from derived SE (0.0221) ~ printed p=0.022: PASS
- weeks CI ordered and contains ATE: PASS
- all SEs positive: PASS
- HIV: |t|=1.16 <1.645 consistent with no star: PASS
- dcr_narcotics_att_3km |t|=2.94>2.576 consistent with ***: PASS
- dcr_fatal_att_1km |t|=8.81>2.576 consistent with ***: PASS
- dcr_att_5km |t|=2.05 in (1.96,2.576) consistent with **: PASS
- p-value null with 'not reported' locator: PASS
- locators cover all keys except 2 booleans: PASS
- exclusion ledger: 25-0=25=n_analysed: PASS