Electoral Preferences and Regional Economies in Romania’s 2024 Presidential Elections: Local-Level Results, Sectoral Indicators, and Spatial Models

Abstract: This repository contains the full dataset and model implementation for the analysis of voting patterns in Romania's 2024 presidential elections, focusing on the relationship between territorial economic structures and electoral preferences. The models estimate vote dominance at LAU level using sectoral, demographic, and regional predictors, including spatial autoregression. Particular attention is given to the overrepresentation of Bucharest in national-level FDI statistics, which is corrected through a GDP-based imputation model. For reproducibility, the repository includes: Cleaned and structured input data (LAU, NUTS3), all modelling scripts in R, Tableau maps for visual analysis and public presentation.File DescriptionsLAU.csvThis dataset contains the local-level electoral and socio-economic data for all Romanian LAU2 units used in the spatial and statistical analyses. The file is used as the base for all models and includes identifiers for merging with the shapefile or spatial weights. It includes:- Electoral results by presidential candidate (2024, simulated),- Dominant vote type per locality,- Sectoral employment categories,- Demographic variables (ethnicity, education, age),- Regional and metropolitan classifications,- Weights for modelling.NUTS3.csvThis dataset provides county-level economic indicators (GDP and FDI) over the period 2011–2022. The file supports the construction of regional indicators such as FDI-to-GDP ratios and export structure. Notably, the file includes both original and corrected values of FDI for Bucharest, following the imputation procedure described in the model script.model.RThis R script contains the full modelling pipeline. The script includes both a model variant with Bucharest excluded and an alternative version using corrected FDI values, confirming the robustness of coefficients across specifications. It includes:- Pre-processing of LAU and NUTS3 data,- Imputation of Bucharest FDI using a linear model on GDP,- Survey-weighted logistic regression models for vote dominance per candidate,- Multinomial and hierarchical logistic models,- Seemingly Unrelated Regressions (SUR),- Spatial error models (SEM),- Principal Component Analysis on SEM residuals,- Latent dominance prediction using softmax transformation,- Export of predicted latent vote maps.Maps.twbxThis Tableau workbook contains all final cartographic representations.The workbook uses a consistent colour palette based on candidate-typified economic structures (industry, services, agriculture, shrinking).- Choropleth maps of dominant vote by candidate,- Gradients reflecting latent probabilities from spatial models,- Maps of residuals and ideological factor scores (PCA-derived),- Sectoral economic geographies per county and per locality,- Overlay of dominant vote and sectoral transformation types.