This paper analyses votes cast in the Eurovision Song Contest in the period 1975 - 2003. We test whether accusations of 'political' voting among participants can be substantiated by looking at geographical influences. Our approach differs in two ways from earlier studies. First, we take into account a variety of variables to distinguish political voting from preferences based on cultural, linguistic, ethnic, and religious differences and similarities between countries. Secondly, we analyse the determinants of the voting behaviour separately per country, instead of looking at average effects over all participating countries. We find that geographical factors substantially affect the votes. Even after correction for cultural, linguistic and other factors many countries prefer or dislike the songs of surrounding countries. This leads to the suspicion that the geograph¬ical preferences reflect political voting. Also, we show that several countries favour songs of participants with the same religious background, while others prefer the contributions of countries with a different religion. Moreover, using data on the amount of Turkish immigrants across European countries, we document that countries with a substantial Turkish population favour the Turkish songs ('patriotic' voting). Furthermore, we study the repercussions of opening up the voting system to the general public by the introduction of televoting. It turns out that religious and patriotic voting have become considerably stronger since the introduction of the new voting system. Finally, we confront our em¬pirical findings to the publicly debated accusations of political voting made against certain blocks of countries. Although our analysis uncovers significant geographical patterns (suggesting political voting), we do hardly establish any empirical evidence for the claims against these particular countries.
|Name||Applied Mathematics Memoranda|
|Publisher||Department of Applied Mathematics, University of Twente|