Statistical detection of systematic election irregularities – three essays on supervised and unsupervised machine learning approaches and the attitudinal consequences of exposing cheating

Behrens, Lion

PDF
BL_Dissertation_20231212.pdf - Published
Download (5MB)

URN:	urn:nbn:de:bsz:180-madoc-662088
Document Type:	Doctoral dissertation
Year of publication:	2023
Place of publication:	Mannheim
University:	Universität Mannheim
Evaluator:	Gschwend, Thomas
Date of oral examination:	28 August 2023
Publication language:	English
Institution:	School of Social Sciences > Social Data Science and Methodology (Keusch 2022-) School of Social Sciences > Politische Wissenschaft, Quantitative Sozialwissenschaftliche Methoden (Gschwend 2007-)
License:	Creative Commons Attribution 4.0 International (CC BY 4.0)
Subject:	310 Statistics
Classification:
Keywords (English):	election forensics , electoral integrity , Monte Carlo Simulation , machine learning , survey experiment
Abstract:	This dissertation (i) develops statistical methodology for the detection of systematic irregularities in fine-graded election results (ii) and seeks to enhance our understanding of the attitudinal consequences of exposing individuals to information on electoral malpractice. While the range of methods for statistical fraud detection—a field often referred to as 'election forensics'—is constantly expanding, the same holds for the knowledge and strategies of micro- and macro-level agents of interference. Hence, this dissertation aims to contribute to the need of continuous innovation in the methodology of electoral anomaly detection. Without doubt, communicating negative findings about the integrity of electoral events is in itself likely to lead to a legitimacy loss of political institutions among the citizenry. This is why—next to making methodological contributions—this dissertation empirically investigates these decays in support, and hence also reflects on the role of statistical fraud detection in the tension between safeguarding democracy and producing democratic backlashes.