A Bayesian Networks Approach for Analyzing Voting Behavior

Arbeitspapier

The problem of finding the factors influencing voting behavior is of crucial interest in political science and is frequently analyzed in books and articles. But there are not so many studies whose supporting information comes from official registers. This work uses official vote records in Spain matched to other files containing the values of some determinants of voting behavior at a previously unexplored level of disaggregation. The statistical relationships among the participation, the vote for parties and some socio-economic variables are analyzed by means of Gaussian Bayesian Networks. These networks, developed by the machine learning community, are built from data including only the dependencies among the variables needed to explain the data by maximizing the likelihood of the underlying probabilistic Gaussian model. The results are simple, sparse, and non-redundant graph representations encoding the complex structure of the data. The generated structure of dependencies confirms many previously studied influences, but it can also discover unreported ones such as the proportion of foreign population on all vote variables.

Angaben zum Objekt

• Sprache: Englisch

Verweise und Beziehungen

• Erschienen in: Series: CESifo Working Paper ; No. 10855

Klassifikation und Themen

• Klassifikation: Wirtschaft ; Specific Distributions; Specific Statistics ; Personal Income, Wealth, and Their Distributions ; Political Processes: Rent-seeking, Lobbying, Elections, Legislatures, and Voting Behavior ; Micro-Based Behavioral Economics: Role and Effects of Psychological, Emotional, Social, and Cognitive Factors on Decision Making‡
• Thema: Bayesian networks ; Gaussian distributions ; voting behaviour ; elections ; voter turnout ; political participation

Beteiligte, Orts- und Zeitangaben

• Ereignis: Geistige Schöpfung
• (wer): Calvin, Miguel ; Rey del Castillo, Pilar
• Ereignis: Veröffentlichung
• (wer): Center for Economic Studies and ifo Institute (CESifo)
• (wo): Munich
• (wann): 2023

Weitere Informationen

• Handle: http://hdl.handle.net/10419/282543
• Letzte Aktualisierung: 10.03.2025, 11:41 MEZ