Verification checks (P4)
- Clean re-run: deleted out/results.json, re-ran analysis.py; byte-identical to previous output (cmp) — PASS.
- Counts: n_supported(2)+n_not_supported(3)+n_not_assessable(1)=6=n_analysed — PASS.
- verdicts/locators/reasons keyed exactly A1..A6 — PASS.
- n_input_rows equals
wc -l(README 14, DCR 134, RESEA 58) and assertions.json has 6 items — PASS. - Conversions: 100(exp(-0.431)-1)=-35.01% (Table 6 prints -35.0%) — PASS; 100(exp(-0.283)-1)=-24.65% (Table 6 prints -24.7%) — PASS (confirms source internal consistency).
- Hepatitis 5 km CI: -0.406±1.96·0.250=[-0.896, 0.084]; ±2.0 SE=[-0.906, 0.094]; both contain 0; absence of star (p≥0.10) consistent — PASS.
- A1 significance: z=-0.431/0.210=-2.05 (p≈0.040); with A19's SE 0.218 z=-1.98 (p≈0.048); ** in both tables consistent — PASS (A19 SE discrepancy logged as data issue).
- A6: 1.99/17.3=11.50%; Table 2 CI [-2.24,-1.75] contains -1.99 — PASS.
- A5: regex search for men/women/male/female/sex/gender in DCR extract: 0 hits — PASS (supports not_assessable).
- Baseline (literal number matching) differs from final verdicts on A2 (ignores sign) and A5 (no numbers to check) — expected, documents why element-wise checking is needed.