Data audit (P1)
| file | SHA-256 | lines (wc -l) | matches README hash |
|---|---|---|---|
| data/README.md | 16ebdff389eeaee9508fe81c4edc43690f31f3463c469680c814134086ef9929 | 10 | n/a |
| data/sources/arxiv-2609.27732v1.txt | ee2d444d2f78f2af2ca2cf69b59a61fcd7ac5e10ea29e8ddcd67917bf07d6a8b | 134 | yes |
| data/sources/arxiv-2606.14621v1.txt | 27c463c0a144d9d5a3916dc641f15b6029f6c68b5e574ad9074e7df1aeb0a238 | 58 | yes |
Schema: plain-text extracts; tables one row per line, cells separated by " | "; DCR SE rows begin with "| (" on the line after the coefficient row (blank line between). RESEA Tables 2/3 re-set one row per outcome; each cell "estimate; p=...; 95% CI: [lo, hi]". Units: DCR ATTs in log points, Observations in region-years (thousands separators); RESEA wages USD/yr with "$" and thousands commas, reemployment pp, weeks. Missing: no exact p-values in DCR tables (stars only); no SEs in RESEA Table 2 (CIs only). No duplicated rows.
Anomalies / internal inconsistencies noted (none affects the requested cells; handled by using the table named in each key):
- The Table 6 5 km SE is (0.210) but Appendix Table A19 col CS-DR prints the same ATT -0.431** with SE (0.218). Key names Table 6 -> 0.210 used.
- Table 6/7 and Figure 6 say 608 never-treated control regions; A19 note says 640. Table 6 observation counts (16,175 / 16,875 / 17,250) equal 25 x (640 + treated regions 7/35/50), i.e. consistent with 640, not 608, controls. Reported as printed; not resolved.
- Abstract says fatal outcomes "fall by 35%"; Table 7 fatal 5 km ATT -0.354 corresponds to exp(-0.354)-1 = -29.8% (Sec. 6.1 says ~30%). Log points vs percent conflation in abstract; not used.
- Prose (Sec. 6) calls HIV estimates "close to zero" although 3 km/5 km ATTs are -0.239/-0.340 (insignificant). Not used.
- RESEA prose gives the weeks-on-UI CI as magnitudes [1.75, 2.24]; Table 2 prints [-2.24, -1.75]. Table 2 used.
- RESEA prose uses "p ¡ 0.001" (rendering defect for "<"). Not used. Availability: no input violates the rules; Table 3 of RESEA duplicates Table 2 ATE column, A19 duplicates Table 6 5 km - named tables used.