It is not just about the Melody: How Europe Votes for its Favorite Songs

TL;DR

This paper analyzes Eurovision voting networks to uncover factors influencing voting decisions, identifies community structures, and evaluates predictive models, revealing limited improvements over betting tables.

Contribution

It provides a comprehensive analysis of Eurovision voting patterns, identifies community clusters, and assesses the predictive power of network-based models compared to betting data.

Findings

Communities of countries tend to award each other more points.

Network structure alone offers limited predictive power.

Betting tables outperform network-based models in vote prediction.

Abstract

The Eurovision Song Contest is a popular annual international song competition organized by the European Broadcasting Union. The winner is decided by the audience and expert juries from each participating nation, which is why the analysis of its voting network offers a great insight into what factors, besides the quality of the performances, influence the voting decisions. In this paper, we present the findings of the analysis of the voting network, together with the results of a predictive model based on the collected data. We touch upon the methodology used and describe the dataset we carry the analysis on. The results include some general features of the voting networks, the exposed communities of countries that award significantly more points among themselves than would be expected and some predictions on what the biggest factors that lead to this phenomenon are. We also include the model to predict the votes based on network structure of both previous votes and song preferences of nations, which we found not to offer much improved predictions than relying on the betting tables alone.