Sensitivity analysis of elections for Dutch House of Commons (Tweede Kamer)

This thesis presents a comprehensive mathematical sensitivity analysis of the proportional representation (PR) system, as applied to the Dutch House of Commons (Tweede Kamer, hereinafter referred to as TK), focusing on the relationship between electoral fairness, party fragmentation, and seat-allocation stability. The study also examines how counting errors and voter uncertainty may affect vote distributions and final seat allocations. While existing studies often focus on deterministic measures of electoral fairness, this research uses a formal hybrid approach. By combining structural parameter analysis with the stochastic modeling of uncertainties using the 2025 Dutch Tweede Kamer (TK25) election data, this study provides a quantitative framework and theoretical foundation for evaluating electoral sensitivity. First, dynamic threshold simulations (from 0.67% to 5.0%) and comparative analyses of six seat allocation algorithms reveal that the currently implemented D’Hondt method mathematically favors large parties. Additionally, the analysis identifies a sensitive instability interval between 1.8% and 2.7% in which the number of parliamentary parties changes sharply. Second, a probabilistic spatial counting error model is developed to evaluate the possibility of human error and how that affects the stability of seat distribution for major and minor political parties. The findings indicate that even if 10,000 ballots were misclassified, the corrected true error-free votes count would not have much impact on the final seat allocation compared to the official Dutch TK25 election results. The voter-transfer simulation shows that uncertainty mainly redistributes votes within ideologically related blocs. The centrist bloc is particularly affected. Finally, Ordinary Least Squares (OLS) regression analysis of TK23 to TK25 public opinion polls shows a linear decay in polling errors as the election day approaches. These results provide insights and suggestions for discussing the robustness of proportional representation under both structural and stochastic uncertainty