# Data audit (P1)
All SHA-256 values match data/README.md and bundle.json.

| file | sha256 | lines (wc -l) / items | content |
|---|---|---|---|
| data/README.md | c6d37572…792b2 | 17 lines | provenance table |
| data/reference_list.json | 7ab6cbf2…5da | 5 items (32 lines) | R1–R5: id, entry, cited_for, source_file; no missing fields, no duplicate ids |
| sources/arxiv-2609.27732v1.txt | ee2d444d…a8b | 134 | header + abstract, Sec 6, Tables 6,7, A19 |
| sources/arxiv-2606.14621v1.txt | 27c463c0…a238 | 58 | header + abstract, Sec 3 Table 2, Table 3 |
| sources/arxiv-2603.03260v2.txt | 3cb9a4f3…1551 | 107 | header + abstract, Sec 3 Tables 2,3, Sec 5 |
| sources/arxiv-2505.04555v3.txt | b1731a9b…95c1 | 66 | header + abstract, App. A.1 Table 2 |
| sources/arxiv-2510.03658v1.txt | f8782937…1127 | 91 | header + abstract, Sec 4.1, Table 3 |
(Full hashes are recomputed and printed by out/analysis.py.)

Every reference's source_file exists. Each extract has a "Title:" and "Bibliographic record:" header line and a body title line ("# ...") identical to the header title.

## Anomalies found (carried into results.json `data_issues`)
- 2609.27732 (R1): Table A19 CS-DR column prints SE (0.218) for the same 5 km CS-DR ATT −0.431 whose SE is (0.210) in Table 6; A19 note says 640 never-treated regions while the A19 body and Tables 6/7 say 608. Abstract says fatal outcomes "fall by 35%" whereas Sec 6.1 says "approximately 30%" (Table 7 fatal 5 km coefficient −0.354 → exp(−0.354)−1 ≈ −29.8%).
- 2505.04555 (R4): Table 2 SE of own-wage elasticity (0.091) is identical to the SE of "Affected employment"; since −0.732 = −0.268/0.366 is a ratio, an SE equal to the numerator's SE is implausible (possible typesetting duplication in the source). The value is nonetheless printed as such.
- 2510.03658 (R5): footnote markers rendered as "^9^9 9"; abstract headline (0.0855, 18.79%, from Table 2 — omitted from extract, mean 0.455) differs from Table 3 preferred estimate (0.0962, 36.85%, mean 0.261); these are different outcome definitions (count vs binary), not a contradiction. Table 3 "% Change" −0.3685 vs −0.0962/0.261 = −0.36858 (last-digit truncation).
- 2606.14621 (R2): "p ¡ 0.001" rendering defect; Tables 2–3 re-set by extract editor (declared in header).
- 2603.03260 (R3): Table 3 header row garbled (editor's note supplies the column names).

## Availability rules
No input violates the rules; only the extracts are used. No information from memory or other versions is used.