# Data audit (P1)
| file | SHA-256 | lines (wc -l) | matches README/bundle hash |
|---|---|---|---|
| data/README.md | c6d375725356176fa7600ab663d9306c5f37c1097b619792bcbed166ccc792b2 | 17 | bundle: yes |
| data/reference_list.json | 7ab6cbf2606f90f3174215214feddd73533d29cda4befdc77ffdb595193ab5da | 32 (5 top-level items) | bundle: yes |
| data/sources/arxiv-2609.27732v1.txt | ee2d444d…a8b | 134 | yes |
| data/sources/arxiv-2606.14621v1.txt | 27c463c0…0b3 | 58 | yes |
| data/sources/arxiv-2603.03260v2.txt | 3cb9a4f3…1f | 107 | yes |
| data/sources/arxiv-2505.04555v3.txt | b1731a9b…c1 | 66 | yes |
| data/sources/arxiv-2510.03658v1.txt | f8782937…27 | 91 | yes |
(Full hashes are recomputed and verified by out/analysis.py against data/README.md.)

Schema: reference_list.json = list of 5 objects {id, entry, cited_for, source_file}; no missing fields, no duplicate ids; every source_file exists.
Each extract: provenance header (Title, Authors, Bibliographic record, Retrieved, License, contents, rendering note) then body with title line, abstract, cited sections/tables.

## Anomalies found
1. arxiv-2609.27732v1: Table 6 prints SE (0.210) for the 5 km CS-DR ATT −0.431, while Table A19 prints (0.218) for the same CS-DR 5 km estimate; Table A19 note says 640 never-treated regions while its body and Table 6 say 608. Abstract says fatal outcomes fall 35% while Section 6.1 says ~30% (Table 7: −0.354 log points ≈ −29.8%). Extract truncated mid-sentence at the end of the A19 note.
2. arxiv-2606.14621v1: rendering defect "p ¡ 0.001" (should read p < 0.001); Tables 2–3 re-set by the extract editor (documented in header).
3. arxiv-2603.03260v2: Table 3 header garbled (editor's note gives column order); τ=3 coefficient not reported in text.
4. arxiv-2505.04555v3: Table 2 prints identical SE (0.091) for "Affected employment" and the own-wage elasticity (a ratio of affected employment −0.268 to affected wages 0.366, whose SE is 0.171); a ratio SE equal to the numerator SE is unusual, but the value is as printed.
5. arxiv-2510.03658v1: abstract/§4.1 headline (0.0855, 18.79%) refers to a Table 2 that is omitted; Table 3 %change −0.3685 vs −0.0962/0.261 = −0.3686 (rounding of the printed mean).

## Information-availability
No input violates the rules: all evidence is in the extracts. No exclusions are needed (all 5 references analysed).