# Verification checks (P4, 2026-10-30)

| # | check | outcome |
|---|---|---|
| 1 | Clean re-run: results.json deleted, script run twice, SHA-256 identical | PASS |
| 2 | n_harmonised (4) + n_not_harmonisable (1) = 5 items = n_analysed | PASS (asserted in script) |
| 3 | Every item with value null has harmonisable=false and non-null missing_quantity; others have missing_quantity null | PASS |
| 4 | E1 computed -35.01% agrees with printed '% change' -35.0% (Table 6, 5 km) | PASS |
| 5 | E5: printed % Change -0.3685 vs b/mean -0.0962/0.261 = -0.3686 (consistent; confirms 0-1 scale, so x100 to pp is right) | PASS |
| 6 | E2 magnitude: 1.47/75.0 = 1.96%; sign positive like the ATE | PASS |
| 7 | E3 magnitude: -0.388*0.047 = -1.82%, same order as abstract's "2% employment decline" | PASS (sanity only) |
| 8 | Signs preserved vs baseline printed estimates (baseline_printed_estimates) for E1,E2,E3,E5 | PASS |
| 9 | Printed numbers parsed verbatim with string assertions (-0.431**, 75.0%, 1.47 pp, -0.388, 0.047, -0.354**, -0.0962) | PASS |
| 10 | n_input_rows equal wc -l (README 16, sources 134/58/66/91) and effects.json 5 items | PASS |
| 11 | No CI/SE/chronology pairs required by the task (no uncertainty propagation) | N/A |
| 12 | Source-internal inconsistencies detected and recorded in data_issues (none affects the harmonised values) | recorded |