Replication package for "Designing Empirical Studies of AI Infrastructure, Behavioral Nudges, and Citizen Satisfaction: A Reproducible Monte Carlo Framework for Smart-City Research"

v3.0.0 (2026-09-28) accompanies revision R1 of the manuscript "Designing Empirical Studies of AI Infrastructure, Behavioral Nudges, and Citizen Satisfaction: A Reproducible Monte Carlo Framework for Smart-City Research" (Sage Open, under review). The package regenerates every table and figure of the manuscript from fixed seeds. The paper asks a design question: under what combinations of sample size, effect magnitude, predictor correlation and measurement reliability can a comparative municipality-level study reliably identify associations between a focal predictor (for example an AI-infrastructure or nudge-intensity index) and citizen satisfaction? It answers it with an 80-cell scenario-matrix Monte Carlo framework (06_scenario_matrix.py; 2,500 replications per cell) validated analytically (07_checks.py), and, in this version, with the analyses added in revision (08_revision_analytics.py): analytic noncentral-t sample sizes for an 80% rejection probability with dense-grid Monte Carlo brackets; the sensitivity of covariate-signal leakage (structural misattribution) to correlation level, reliability, unequal correlation structures and covariate-coefficient patterns; measurement error in the municipal satisfaction estimate; and a two-focal-predictor arm in which AI and nudge indices are modeled jointly. 09_make_figures.py draws the five manuscript figures from the CSV outputs. The package also contains the fully documented worked example used as an end-to-end calibration check (scripts 01–05, canonical synthetic dataset of 200 municipalities, 10,000-replication Monte Carlo, power curves, correlated-predictor check) and the materials of the measurement-feasibility audit of Santander, Spain: an audit note documenting the three institutional sources (strategic-plan executive summary, 2015; the municipality's indicator sheet for the InfoCiti reporting system of Red.es, December 2024; a technical questionnaire designed for this study, May 2026), the written institutional authorization to cite and publish them, the coding protocol, and the coding of the 86 InfoCiti indicators (reported / not reported, reference year, source-note type). No indicator values beyond those cited in the manuscript and no personal data are included. The audit establishes measurement feasibility only; it does not calibrate the simulation and claims no external validity. All simulation outputs were verified to regenerate byte for byte on the pinned environment (Python 3.11, NumPy 2.3.5, SciPy 1.17.0) on 2026-09-28. See README.md, CHANGELOG.md and Reproducibility_Report_v3.pdf.