Data audit (P1)
| file | SHA-256 | lines (wc -l) | notes |
|---|---|---|---|
| data/README.md | c6d375725356176fa7600ab663d9306c5f37c1097b619792bcbed166ccc792b2 | 17 | provenance table |
| data/reference_list.json | 7ab6cbf2606f90f3174215214feddd73533d29cda4befdc77ffdb595193ab5da | 32 | 5 items, keys id/entry/cited_for/source_file, no duplicates, no missing |
| sources/arxiv-2609.27732v1.txt | ee2d444d…6a8b | 134 | v1 only, 23 Sep 2026; matches README hash |
| sources/arxiv-2606.14621v1.txt | 27c463c0…a238 | 58 | v1 only, 12 Jun 2026; matches README hash |
| sources/arxiv-2603.03260v2.txt | 3cb9a4f3…1c55 | 107 | v1 3 Mar 2026, v2 2 Apr 2026; matches README hash |
| sources/arxiv-2505.04555v3.txt | b1731a9b…95c1 | 66 | v1 7 May 2025, v3 16 Mar 2026; matches README hash |
| sources/arxiv-2510.03658v1.txt | f8782937…1127 | 91 | v1 only, 4 Oct 2025; matches README hash |
All five extract SHA-256 values equal those listed in data/README.md. Schema: text extracts with provenance header, separator line, body. Not tabular, so "missing values/duplicates" concern only the reference list (none).
Anomalies noted (handled in results.json data_issues)
- 2609.27732: Table 6 5 km ATT SE (0.210) differs from Table A19 CS-DR column SE (0.218) for the same estimate (-0.431); A19 note says 640 never-treated regions vs 608 elsewhere; the "paragraph that discusses" Table A19 announced in the header is not present (extract ends at the A19 note). Abstract says fatal outcomes fall by 35% while Section 6.1 says ~30% (Table 7: -0.354 → exp−1 ≈ −29.8%).
- 2603.03260v2: Table 3 header garbled in the rendering; extract editor's note gives column order — used.
- 2606.14621v1: rendering defect "p ¡ 0.001" (line 39), read as p < 0.001; text gives the weeks-on-UI CI as positive magnitudes [1.75, 2.24] while Table 2 gives [-2.24, -1.75] (sign convention only).
- 2505.04555v3: SE of own-wage elasticity (0.091) printed identical to SE of affected employment (0.091); possibly a source copy issue, but the cited value matches the print.
- 2510.03658v1: abstract headline (0.0855, 18.79%, from Table 2, not in extract) differs from Table 3 col (4) (0.0962, 36.85%) — different outcome definitions; Table 3 "% Change" -0.3685 vs -0.0962/0.261 = -0.3686 (rounding).
Information-availability
No input violates the rules; all evidence comes from the extracts. No remembered content used.