title	output	date
Readme	html_document	2025-01-30

Replication Materials: [The Zweitstimme Forecast for the German Federal Election 2025]

This repository contains replication materials for "[The Zweitstimme Forecast for the German Federal Election 2025]".

Repository Structure

Data Files

Raw Data

• ./data/btw_2021_kerg2.csv: 2021 German federal election data
• ./data/btw_candidates_1983-2025.csv: Historical candidate data from 1983-2025 (csv format)
• ./data/btw_candidates_1983-2025.RData: Historical candidate data from 1983-2025 (R format)
• ./data/btw25_geometrie_wahlkreise_shp/*: Shapefile containing electoral district geometries
• ./data/pre_train_data_21.rds: Pre-training model data including elections until 2021
• ./data/germany-federal-polls.csv: Federal polling data
• ./data/germany-state-elections.csv: State election results
• ./data/germany-state-polls.csv: State-level polling data

Generated Data

• ./output/forecast_draws_2025-01-30.rds: Model forecast draws (2025-01-30)
• ./output/forecast_party_vote.rds: Party vote forecasts (2025-01-30)
• ./output/district_reg_predictions.rds: District-level predictions (2025-01-30)
• ./output/prediction_data_districts.rds: Prediction data for districts from regession (predicted party vote for each district)
• ./output/pred_probabilities.rds: Probability calculations for scenarios (e.g. coalition majorities)

Code Files

The analysis can be reproduced by running the scripts in the following order:

./auxiliary/functions.R: Helper functions ./auxiliary/packages.r: Install and load required R packages

./code/00_run-model.R: Main script to execute the full analysis

1. ./code/01_prepare-data.R: Data preparation and cleaning for pre-training model
2. ./code/02_ger_structural_pre_train_stan.R: Structural model for party vote
3. ./code/03_ger_combined_model_stan.R: Combined forecasting model for party vote
4. ./code/04_party-vote-data.R: Processing second vote forecast data
5. ./code/05_party-vote-figures.R: Creating figures for party vote analysis
6. ./code/06_district-model.R: Electoral district forecasting model
7. ./code/07_vacant-seats.R: Analysis of potential vacant seats
8. ./code/08_district-figures.R: Creating district-level figures
9. ./code/09_probabilities.R: Probability calculations for scenarios (e.g. coalition majorities)

./code/A_descriptives.R: Descriptives

Model Code

• ./model_code/combined_model_simple.stan: Combined forecasting model specification
• ./model_code/structural_pre_train_simple.stan: Structural model specification

Output Files

Figures

• ./output/plots/figure_forecast_party_vote.pdf: Main forecast visualization (Figure 1)
• ./output/plots/figure_forecast_party_vote.png: Main forecast visualization (png) (Figure 1)
• ./output/plots/figure_forecast_districts.png: District-level predictions (Figure 2)
• ./output/plots/state-rmse.pdf: State-level polling accuracy RMSE (Figure 3)
• ./output/plots/federal-rmse.pdf: Federal-level polling accuracy RMSE (Figure 4)

Tables

• ./output/tables/district_reg.tex: District model regression results (Table B.2)
• ./output/tables/forecast_districts.tex: District forecasts (Table B.3)
• ./output/tables/vacant_districts.tex: Vacant district analysis (Table B.4)

Requirements

Software Requirements

• R version 4.4.1 (2024-06-14)
• RStudio 2024.9.1.394

Required R Packages

The analysis requires several R packages for data manipulation, visualization, and statistical modeling. All required packages are listed in auxiliary/packages.r and will be automatically installed and loaded when running the code. Key packages include:

• Data manipulation: tidyverse, dplyr, haven
• Statistical modeling: rstan, MASS
• Visualization: ggplot2, plotly
• Spatial analysis: sf
• Output formatting: stargazer, knitr

The complete list of dependencies will be installed automatically when running auxiliary/packages.r.