# Data audit
All SHA-256 match data/README.md and bundle.json.
| file | sha256 | lines (wc -l) / items | content |
|---|---|---|---|
| data/effects.json | 708c1ff0… | 5 top-level items | 5 effect specs, keys id/source/effect/reported_definition/target_definition; no duplicates, no missing |
| data/README.md | 7cf7cb38… | 16 lines | provenance table |
| sources/arxiv-2609.27732v1.txt | ee2d444d… | 134 | DCR: Tables 6, 7, A19; log rates per 100k; no level/mean of outcome printed |
| sources/arxiv-2606.14621v1.txt | 27c463c0… | 58 | RESEA Table 2 (means in %, effects in pp), Table 3 |
| sources/arxiv-2505.04555v3.txt | b1731a9b… | 66 | MW Table 2: elasticity -0.388, % MW changes 0.047 |
| sources/arxiv-2510.03658v1.txt | f8782937… | 91 | Benin Table 3 LPM; Mean Dep. Var. 0.261 |

Anomalies (also in results.json data_issues):
- DCR: CS-DR 5 km ATT -0.431 has SE (0.210) in Table 6 but (0.218) in Table A19.
- DCR: control regions 608 in tables, but A19 note says 640 never-treated regions.
- DCR: abstract says fatal hospitalisations fall by 35%; Table 7 5 km -0.354 implies -29.8% (Sec 6.1 says ~30%).
- DCR: printed % change rows for some A19 columns differ by 0.1 from 100(exp(b)-1) (rounding of b).
- DCR: no level (mean/baseline) of fatal hospitalisation rate printed -> E4 not harmonisable.
- MW: note says "about 4% minimum wage increase" whereas table row "% MW changes" prints 0.047 (4.7%). Target says use table value.
- RESEA: control mean 75.0% is the raw aggregate control mean, not block-adjusted; ATE 1.47 pp is block-adjusted. Abstract rounds to 1.5 pp.
- Benin: abstract/text headline 0.0855 (18.79%, mean 0.455) refers to Table 2, not in extract; Table 3 col (4) is -0.0962 with mean 0.261; "% Change" row -0.3685 consistent.
No filtering applied: all 5 items analysed (exclusions.csv records zero removals).
Availability: no input violates rules; E4 requires an unprinted quantity -> declared not harmonisable.