# Data audit
| file | SHA-256 | lines (wc -l) / items | notes |
|---|---|---|---|
| data/assertions.json | 2965ab3acc6f326653169b83a5345698b51fac88a556de65faf89210aea0cad4 | 32 lines; 6 items | ids A1–A6, no duplicates; A1–A5 cite arxiv-2609.27732v1, A6 cites arxiv-2606.14621v1 |
| data/README.md | 5348bd49912099735108dbbe19ae09fa72bd600cb3d12b878658e6b9806198a7 | 14 | provenance; extract hashes listed there match the computed hashes |
| data/sources/arxiv-2609.27732v1.txt | ee2d444d2f78f2af2ca2cf69b59a61fcd7ac5e10ea29e8ddcd67917bf07d6a8b | 134 | Table 6, Table 7, Sec 6.1, Table A19; log points, clustered SEs, stars |
| data/sources/arxiv-2606.14621v1.txt | 27c463c0a144d9d5a3916dc641f15b6029f6c68b5e574ad9074e7df1aeb0a238 | 58 | Table 2 & 3 re-set one row per outcome |

Hashes also match bundle.json. Not tabular data: no missing-value/duplicate concepts beyond the above.
Anomalies (all recorded in results.json `data_issues`): Table 6 SE 0.210 vs Table A19 SE 0.218 for the same CS-DR 5 km estimate; A19 note 640 vs 608 control regions; abstract "fatal fall by 35%" vs Table 7 −0.354 (−29.8%); text "HIV close to zero" vs 5 km −0.340; Table 6 3 km "−24.7%" vs computed −24.65%; RESEA text CI written as positive magnitudes; "p ¡ 0.001" rendering defect.
No input violates the information-availability rules (all are the admissible extracts / task files).
No exclusions: all 6 assertions assessed.