Replication Data and Code for: Measuring What Matters for Viability of Inclusive Insurance — Behavioural Diagnostics from 169 Schemes Across 38 Countries

Replication repository for Dror, "Measuring What Matters for Viability of Inclusive Insurance: Behavioral Diagnostics from 169 Schemes Across 38 Countries" (manuscript under review at the Geneva Papers on Risk and Insurance — Issues and Practice, Special Issue on Inclusive Insurance). This repository contains the data and code needed to reproduce the statistical results, tables, and figures reported in the manuscript. A single R script (replicate_all.R) reproduces the main findings from the source data without manual intervention; stopifnot() assertions verify each computed value against the manuscript text. DATASETS: schemes_dataset.csv — 169 named community health insurance schemes across 38 countries, with country-level trust measurements and sustainability outcomes (derived from Trust_as_Commons, Zenodo record 18244575). COMPLETE_274_ORGANIZATIONS_FINAL.csv — 274 organizations at three governance scales (community, local, regional) across five continents, with continuous consensus measurements and success outcomes (derived from Consensus Thresholds repository, Zenodo record 17219731). countries_dataset.csv — country-level trust availability and scheme counts. trust_data_sources.csv — survey attribution for each country's trust measurement (World Values Survey, Afrobarometer, Asian Barometer, Latinobarómetro, Arab Barometer, Wellcome Global Monitor). schemes_with_wgm.csv — 169 schemes merged with Wellcome Global Monitor 2018 trust data across the relevant survey items. WHO_Health_Expenditure_Global_2000-2021.csv — WHO Global Health Expenditure Database extract used as economic controls. wgm2018_country_level.csv — country-level WGM 2018 trust extract. ROBUSTNESS BATTERY (phase2_econometric_reruns.R): Cluster-robust standard errors with country-level clustering (Liang–Zeger CR1); non-parametric bootstrap confidence interval for the AIC-optimal trust breakpoint; Bonferroni correction across candidate thresholds; leave-one-country-out validation; post-hoc power analysis for the Asia local-scale subsample; wild cluster bootstrap; variance inflation factor diagnostics; falsification test across adjacent trust constructs from the Wellcome Global Monitor; economic controls; Sub-Saharan Africa subgroup regression. ADDITIONAL ANALYSES: phase2b_ssa_investigation.R (within-SSA sub-region analysis); q11e_derisk.R (channel-specific trust as candidate pre-launch screen); Consensus/02_asia_local_power.R (post-hoc power for the Asia local-scale result); Issue Briefs IB1–IB5 (self-contained arguments with reproducible evidence on threshold robustness, channel-specific trust, scale-and-consensus, the Asia subsample, and the choice of operational screening instrument). DOCUMENTATION: README.md (quick-start and methodology notes); CODEBOOK.md (data dictionary for all variables in all data files); LICENSE (CC BY 4.0). REQUIREMENTS: R ≥ 4.1. Main replication uses base R. The sandwich and lmtest packages are required for cluster-robust standard errors. The dplyr, boot, pwr, and car packages are required for the robustness scripts. ggplot2 is required for figures. readxl is required for data preparation.