Data audit (P1)
| file | SHA-256 | lines (wc -l) | matches README hash |
|---|---|---|---|
| data/README.md | 16ebdff389eeaee9508fe81c4edc43690f31f3463c469680c814134086ef9929 | 10 | (not listed; matches bundle.json) |
| data/sources/arxiv-2609.27732v1.txt | ee2d444d2f78f2af2ca2cf69b59a61fcd7ac5e10ea29e8ddcd67917bf07d6a8b | 134 | yes |
| data/sources/arxiv-2606.14621v1.txt | 27c463c0a144d9d5a3916dc641f15b6029f6c68b5e574ad9074e7df1aeb0a238 | 58 | yes |
Schema: plain-text extracts with provenance header; tables rendered one row per line with " | " separators.
- DCR (Armendariz 2026): Table 6 (cols 1/3/5 km; rows ATT, SE line, % change, Treated regions, Control regions, Observations), Table 7 (rows Mental/Behavioral, Hepatitis, HIV, Narcotics, Drug-related (fatal) each with SE line; Treated/Control regions), Table A19 (5 km, four estimators). Units: log points; SEs clustered at Medstat region; stars *** <0.01, ** <0.05, * <0.10. No p-values printed. Panel 1998–2022.
- RESEA (Li et al. 2026): Table 2 (6 outcomes × Treatment, Control, Treatment-control difference, ATE with p and 95% CI), Table 3 (3 primary outcomes × 3 estimators). Re-set by extract editor (one row per outcome). No standard errors printed. RCT Feb 2022–Sep 2023, N = 23,549. Missing values: none in the requested cells. Duplicates: 5 km CS-DR ATT appears in Tables 6 and A19 (SE differs: 0.210 vs 0.218); RESEA Table 3 col 1 duplicates Table 2 ATE column (identical, verified). Anomalies (all recorded in results.json data_issues): SE discrepancy Table 6 vs A19; 608 vs 640 control regions (Observations arithmetically consistent with 640 × 25); abstract "35%" fatal vs −29.8% implied; "HIV close to zero" wording; A19 % change rounding; "p ¡ 0.001" rendering defect; sign convention of weeks-on-UI CI in text. Availability rules: no input violates them; only the two extracts are used; values taken from the tables named in each key (Table 6 not A19; Table 2 not Table 3 or text). Exclusions: none (see out/exclusions.csv).