# Contract — F4-conf-01

**Question.** Extract specified treatment effects and their printed uncertainty, verbatim, from two source extracts (Armendariz 2026, DCR staggered DiD, Tables 6/7; Li, Pearson-Merkowitz & Yokum 2026, RESEA RCT, Table 2), attach locators, derive two non-printed quantities (95% CI from SE for the DCR 5 km ATT; SE from the 95% CI for the RI wage ATE) under a stated convention, and recognise the requested quantity that is not reported (exact p-value of HIV 5 km ATT).

**Estimands.** The printed numbers themselves (no re-estimation). Derived CI assumes approximate normality of the CS-DR ATT estimator with the clustered SE (z = 1.96). Derived SE assumes the printed CI is a symmetric Wald interval estimate ± 1.96·SE (normal approximation; large N = 23,549 makes t ≈ z).

**Outputs.** out/results.json with keys: dcr_att_5km, dcr_att_5km_se, dcr_att_3km, dcr_att_3km_se, dcr_obs_5km (int), dcr_narcotics_att_3km(_se), dcr_fatal_att_1km(_se), dcr_hiv_att_5km(_se), dcr_hiv_att_5km_p_value (null if not printed), dcr_hiv_att_5km_significant_10pct (bool), dcr_att_5km_ci_low/high (derived), ri_wage_ate, ri_wage_ate_ci_low/high, ri_wage_ate_se_reported (bool), ri_wage_ate_se_derived, ri_control_mean_wage, ri_reemployment_ate_pp, ri_weeks_ui_ate, ri_weeks_ui_ate_ci_low/high, locators, derivation_notes, task_id, feasibility, feasibility_reason, data_issues, n_input_rows, n_analysed, random_seed (null). Also out/report.md (6 sections), out/analysis.py, out/claims.json, plus protocol files under out/state and out/exclusions.csv.

**Information rules.** Only the two extracts are evidence; values as printed (sign, decimals; thousands separators removed); where a value appears in several tables use the table named in the key (e.g. Table 6 not A19; Table 2 not Table 3 or text); locators in prescribed format; null + "not reported" locator for absent quantities.

**Invalidity.** Using memory/other versions; taking a value from the wrong table (e.g. A19 SE 0.218 instead of Table 6 SE 0.210; text's positive-magnitude weeks CI instead of Table 2); inventing a p-value; unstated derivation convention.