# Verification checks
1. Clean rerun: deleted results.json/table1.csv/figure1.png and reran analysis.py; `cmp` with the first run gives IDENTICAL. PASS
2. Input SHA-256 matches README. PASS
3. n_input_rows 2498 = README row count. PASS
4. Exclusions: 2498 − 1 − 2497 = 0 rows analysed. This matches n_analysed=null (no estimate). PASS
5. table1 n_days sums to 2498 (250×8 + 249×2). PASS
6. Baseline −0.0429 bp per day × 2497 changes ≈ −107 bp, and (1.68 − 2.75)×100 = −107 bp. PASS
7. Conclusion rule: n_treated=0 → premise_invalid, feasibility infeasible, effect/CI/model_metric/n_analysed null. PASS
8. CI-contains-estimate and SE>0 checks: not applicable, because no estimate exists.