Bachelor Thesis — Impact of Third-Force Candidates on Voter Turnout
The Impact of Third-Force Candidates on Voter Turnout in the 2020 and 2015 Polish Presidential Election and a Forecast for 2025
Overview
A bachelor thesis analyzing how third-force candidates influenced voter turnout in Poland's 2015 and 2020 presidential elections, with a forecast for 2025 based on county-level data, t-tests, and linear regression. The study finds a significant negative relationship between third-force support and second-round turnout growth.
The thesis examines whether strong first-round support for third-force candidates predicts reduced second-round turnout growth in Poland's 2015 and 2020 presidential elections, extending the analysis with a 2025 forecast. Using PKW county-level data, the study creates turnout-delta variables, flags "above-average" support counties, and applies Welch's t-tests and linear regression to quantify relationships between protest-vote strength and runoff engagement. Visual outputs include classification maps, bar charts of mean differences, regression plots, and a forward-looking forecast map and table. The conclusions suggest systematic demobilization among protest voters after their preferred candidates are eliminated, with implications for campaign strategy and democratic legitimacy.
Research Objectives
- Test whether counties with above-average support for Szymon Hołownia or Krzysztof Bosak in 2020, and Paweł Kukiz in 2015, showed smaller turnout increases in the runoff
- Forecast whether similar demobilization patterns may occur in 2025 in areas likely to back Mentzen or Hołownia
- Quantify runoff turnout changes in high vs. low third-force support counties
- Model continuous relationships and forecast 2025 risk areas
Methodology & Technology Stack
The research employs quantitative analysis and statistical modeling:
- Data Sources: County-level results from Państwowa Komisja Wyborcza (PKW) for 2015 and 2020; contextual exit-poll references (IPSOS 2020)
- Statistical Methods: Descriptive statistics, Welch's two-sample t-tests (high vs. low third-force support), and linear regression modeling turnout change as a function of first-round vote share
- Tools: RStudio workflow for data cleaning, feature creation (turnout deltas, binary flags for "above-average" support), and visualization (choropleths, bar charts, regression plots)
- R Packages: dplyr, tidyr, readr, ggplot2, sf or tmap for mapping, broom for model tidying, readxl/openxlsx for data ingestion
Key Features
- Comparative county segmentation using national thresholds for Hołownia (13.45%) and Bosak (6.78%) in 2020, plus Kukiz's 2015 share
- Multi-chart visual analytics: migration/flow conceptual figure, choropleth maps, bar charts, regression plots, and forecast visuals for 2025
- Actionable insights for campaign strategy and policy reform to reduce second-round demobilization
- Reproducible county segmentation and modeling pipeline
Research Findings
The thesis presents key insights into electoral behavior:
- Counties with above-average support for Hołownia and Bosak had significantly smaller second-round turnout gains; similar patterns hold in regions with historical Kukiz strength
- Regression models indicate a statistically significant negative slope linking first-round third-force vote share to second-round turnout growth; effect sizes vary by candidate, with low but meaningful R²
- 2025 outlook: risk of repeated demobilization in specific counties unless engagement strategies improve
Visualizations
Counties with Above-Average Support for Hołownia and Bosak (1st Round, 2020 Election)
Change in Turnout Between I and II Round of Presidential Elections 2020
Did Hołownia's Voters Participate in the Second Round?
Linear regression: Support for Szymon Hołownia in Round 1 vs. Change in Turnout
Did Bosak's Voters Participate in the Second Round?
Linear regression: Support for Krzysztof Bosak in Round 1 vs. Change in Turnout
Counties Forecasted to Show Lower Turnout Growth in the 2025 Runoff
Based on historical behavior of third-force electorates in 2020
Turnout Data by County
Download Thesis
You can download the complete thesis document:
Related Projects
- Election Data Cleaning Project — data pipeline for cleaning and structuring the PKW datasets used in this thesis