DATA FOR: "Multiscale Spatial Heterogeneity of Electoral Determinants in the 2022 Brazilian Presidential Election "

Replication package (code and data) for a manuscript under double-blind review at The Annals of Regional Science. Author identity withheld pending completion of peer review; full attribution will be added upon acceptance. This package accompanies a study of the spatial heterogeneity of electoral determinants in the 2022 Brazilian presidential election, estimated via Multiscale Geographically Weighted Regression (MGWR) across all 5,570 Brazilian municipalities. The dataset (novos_dados_2022.csv) contains municipality-level aggregate covariates only, no individual-level records, drawn from publicly available sources (TSE, IBGE Census 2022, ANS, RAIS, DATASUS, FIRJAN); full variable definitions and sources are documented in CODEBOOK.md. reproduce.py performs the complete pipeline from raw data to all reported tables and figures, with a pinned-bandwidth mode for exact reproduction and a full re-estimation mode for independent bandwidth selection. MANIFEST.md, generated on each run, cross-checks 46 numerical targets against the manuscript. Environment: Python 3.11, scikit-learn, mgwr, libpysal, spglm, esda, spreg (full specification in environment.yml).