Electoral behavior in the Digital Era

Electoral behavior represents one of the central objects of political science, reflecting how individuals form electoral preferences, make voting decisions, participate in elections, and respond to political, social, psychological, and institutional factors. Its study has been shaped by influential theoretical approaches, including sociological, socio-psychological, and rational choice models, as well as behavioral approaches explaining voting decisions through individual characteristics, social environment, political attitudes, party identification, and rational calculations. However, digital transformation raises important questions concerning the explanatory capacity of these classical approaches. The emergence of social media, digital platforms, algorithmic personalization, big data, artificial intelligence, and evolving information environments has transformed the context in which electoral preferences are formed and voting decisions are made. Voters are increasingly exposed to personalized political information, targeted communication, algorithmically selected content, and rapidly changing digital narratives. Consequently, electoral behavior has become more dynamic, multidimensional, and measurable through large-scale digital data. This article examines the principal theoretical and methodological approaches to the study of electoral behavior and assesses their capacity to explain contemporary transformations under digitalization. Particular attention is paid to the limitations of classical models in accounting for real-time changes in electoral preferences, digital information consumption, algorithmic influence, online political interactions, and the growing volume of behavioral data generated by voters. Building upon this theoretical analysis, the article proposes an AI-assisted algorithmic approach to the analysis and forecasting of electoral behavior. The proposed framework integrates political, social, psychological, demographic, behavioral, and digital indicators into computational models capable of identifying patterns, correlations, and temporal changes in electoral behavior. Such an approach could facilitate predictive models designed to assess changes in electoral preferences and anticipate patterns of electoral activity. The article argues that digital transformation requires a methodological expansion of traditional theories of electoral behavior toward dynamic, data-driven, and predictive approaches. At the same time, applying artificial intelligence to electoral forecasting requires methodological rigor, transparency, explainability, data protection, and careful consideration of algorithmic bias and manipulation. The study therefore seeks to establish a theoretical and methodological foundation for further research into AI-assisted modeling and forecasting of electoral behavior in contemporary digital societies and democratic systems today.