# F4-conf-01 — Treatment effects and uncertainty extracted from two published tables

# Question

Extract specified point estimates, standard errors, CI bounds and counts exactly as printed from (i) Armendariz (2026), arXiv:2609.27732v1 (staggered DiD, Swiss drug consumption rooms, Tables 6–7) and (ii) Li, Pearson-Merkowitz & Yokum (2026), arXiv:2606.14621v1 (RESEA RCT, Table 2); attach locators; derive a 95% CI from an SE and an SE from a 95% CI; identify the requested quantity that is not reported. Nothing is re-estimated.

# Data and exclusions

Inputs: data/README.md (10 lines), data/sources/arxiv-2609.27732v1.txt (134 lines), data/sources/arxiv-2606.14621v1.txt (58 lines) (n_input_rows); SHA-256s match the README/bundle (input_sha256). All 25 requested items were assessed (n_analysed = 25); nothing was excluded (out/exclusions.csv). Five input problems were found and recorded in data_issues:

1. The 5 km CS-DR ATT −0.431 has SE (0.210) in Table 6 but (0.218) in Table A19; Table 6 is used, as the key definitions require.
2. Tables 6/7 say 608 never-treated control regions, while the Table A19 note says 640. The Table 6 observation counts equal exactly 25 × (treated + 640), not 25 × (treated + 608). The count is reported as printed and the discrepancy is left unresolved.
3. The abstract says fatal hospitalisations "fall by 35%". The Table 7 5 km fatal ATT of −0.354 implies about −29.8%. The text also calls the HIV estimates "close to zero" although the 5 km ATT is −0.340. Narrative statements were not used.
4. The RESEA text gives the weeks-on-UI CI as positive magnitudes [1.75, 2.24]. Table 2 prints [−2.24, −1.75], and the Table 2 values are used.
5. The RESEA text contains the rendering defect "p ¡ 0.001". It does not affect any key.

# Method

out/analysis.py parses the pipe-delimited tables, matches row and column labels, and pairs each coefficient row with the parenthesised SE line that follows it. It removes stars, $ and thousands separators. It prints every matched source line, then writes out/results.json. No randomness is involved (random_seed = null). Derived quantities use the normal critical value 1.96 (derivation_notes). The baseline, an alternative 2.0 rule, is kept as a sensitivity check (baseline_dcr_att_5km_ci_rule2, baseline_ri_wage_ate_se_rule2).

# Results

DCR (arXiv:2609.27732v1)

• Table 6, 5 km: ATT −0.431** (dcr_att_5km), SE (0.210) (dcr_att_5km_se), Observations 17,250 (dcr_obs_5km).
• Table 6, 3 km: ATT −0.283 (dcr_att_3km), SE (0.279) (dcr_att_3km_se).
• Table 7: Narcotics 3 km 0.549*** (0.187) (dcr_narcotics_att_3km, dcr_narcotics_att_3km_se); Drug-related (fatal) 1 km −1.207*** (0.137) (dcr_fatal_att_1km, dcr_fatal_att_1km_se); HIV 5 km −0.340 (0.294) with no star (dcr_hiv_att_5km, dcr_hiv_att_5km_se).
• The HIV 5 km ATT is not marked significant at 10% (dcr_hiv_att_5km_significant_10pct = false). Its exact p-value is not reported in the extract (dcr_hiv_att_5km_p_value = null).
• Derived 95% CI of the 5 km ATT: [−0.8426, −0.0194] (dcr_att_5km_ci_low, dcr_att_5km_ci_high).

RESEA (arXiv:2606.14621v1), Table 2

• Wage ATE $1,153 (ri_wage_ate), 95% CI [166, 2,141] (ri_wage_ate_ci_low, ri_wage_ate_ci_high). No SE is printed (ri_wage_ate_se_reported = false). The SE derived from the CI is 503.83 USD/yr (ri_wage_ate_se_derived).
• Control mean wage $33,043 (ri_control_mean_wage).
• Reemployment ATE 1.47 pp (ri_reemployment_ate_pp).
• Weeks-on-UI ATE −1.99 weeks, 95% CI [−2.24, −1.75] (ri_weeks_ui_ate, ri_weeks_ui_ate_ci_low, ri_weeks_ui_ate_ci_high).

JSON numbers do not keep trailing zeros: 0.210 is stored as 0.21 and −0.340 as −0.34. The numeric values are identical to the printed ones.

# Uncertainty

• Standard errors: the DCR paper (Tables 6, 7, A19) reports SEs clustered at the MedStat region level, in parentheses, plus stars (p<0.01/0.05/0.10). It prints no exact p-values or CIs for the table entries.
• Confidence intervals: the RESEA paper (Tables 2, 3) reports 95% CIs and two-sided p-values. It prints no SEs.
• The derived DCR CI (ATT ± 1.96 × SE) assumes the estimator is approximately normally distributed, with the printed clustered SE as its standard deviation. With 50 treated regions a t or bootstrap interval could be wider. A 2.0 rule gives [−0.851, −0.011] (baseline_dcr_att_5km_ci_rule2). Using the Table A19 SE of 0.218 instead would also widen the interval.
• The derived RESEA SE ((upper − lower)/3.92) assumes the printed CI is a symmetric Wald interval. This is supported by the near-equal half-widths of 987 and 988. The bounds are rounded to whole dollars, so the derived SE is accurate to about ±0.2. A 2.0 rule gives 493.75 (baseline_ri_wage_ate_se_rule2).
• For context only, a normal approximation of the HIV p-value gives about 0.25 (check_dcr_hiv_att_5km_normal_approx_p_not_printed). This is not a printed value and is not used for dcr_hiv_att_5km_p_value.

# Limitations and open issues

• The exact p-value of the HIV 5 km ATT is not printed, so it is null by design. The task is otherwise complete (feasibility = "complete").
• Two inconsistencies inside the DCR source are unresolved and are reported, not corrected: SE 0.210 vs 0.218, and 608 vs 640 control regions.
• The derived quantities depend on the normal-approximation convention stated above.