Many Eurovision viewers believe there are a several ‘cliques’ in the contest in which a number of countries all vote in a similar way. Using a framework of complex networks researchers from the University of Oxford analysed voting behaviour in the contest over a period from 1992 to 2003.

'As a measure of each country’s actions, we form a data series consisting of the average number of points assigned to each other entrant in the years in which they both compete. The closeness of each pair of countries can then be measured by comparing these data series using Pearson’s correlation coefficient (ρ). <> The Pearson coefficients are then rescaled to produce a ‘distance’ between 0 and 2 using the relationship that the rescaled distance is equal to Sqrt(2(1 − ρ)). The most closely related countries have rescaled distances close to 0, while the least correlated countries have distances close to 2. This data is then used to plot a dendrogram which provides a visual aid for identifying clusters.The figure above shows the resulting dendrogram obtained by consecutively linking the most correlated countries. For example, Greece and Cyprus have the smallest rescaled separation and so they are combined first. The next smallest rescaled distance is between Denmark and Sweden and so they form the next cluster. Once two countries A and B have been combined into a cluster, they are considered to be at the same distance from another country C, which is equal to the shorter of the distances AC and BC. This construction is then generalized up for clusters with more than two countries. The distance between any two clusters is the shortest distance between any two countries in the two clusters. Progressively more countries and clusters are combined in this way, with some countries combining with existing clusters, until all the countries are united into a single cluster. The dendrogram shows quite explicitly that the voting patterns of certain countries are highly correlated. Greece and Cyprus have a very small rescaled distance which demonstrates a very strong voting correlation. This particular finding thereby confirms a long-held belief among regular Eurovision viewers. There is a slightly less correlated cluster that involves the Nordic countries, Denmark, Sweden, Iceland, Norway, Finland, and somewhat surprisingly, Estonia. Many other clusters also arise: Bosnia and Turkey, Croatia and Malta, UK and Ireland (who also show a correlation with the Nordic clique), Belgium and the Netherlands and France and Portugal. The high correlations between the way in which countries assign points provides evidence in support of the theory that voting groups exist'[1].

Further analysis of voting patterns between countries allowed the researchers to identify several countries that appear to be more ‘in tune’ with the rest of Europe, that is, countries that are compatible with a greater number of countries than others. Compatibility between countries was measured by analysing how often a given country exchanged points with another country. If this number exceeded that expected for a ‘random contest’, the countries were judged to be compatible. The country that was found to be compatible with the greatest number of other countries was the UK, whilst at the opposite end of the spectrum France, and to a lesser degree, Spain, were found to be the least compatible with the rest of Europe. |University of Oxford News|

via Daily Favorite

[1] How does Europe

'As a measure of each country’s actions, we form a data series consisting of the average number of points assigned to each other entrant in the years in which they both compete. The closeness of each pair of countries can then be measured by comparing these data series using Pearson’s correlation coefficient (ρ). <> The Pearson coefficients are then rescaled to produce a ‘distance’ between 0 and 2 using the relationship that the rescaled distance is equal to Sqrt(2(1 − ρ)). The most closely related countries have rescaled distances close to 0, while the least correlated countries have distances close to 2. This data is then used to plot a dendrogram which provides a visual aid for identifying clusters.The figure above shows the resulting dendrogram obtained by consecutively linking the most correlated countries. For example, Greece and Cyprus have the smallest rescaled separation and so they are combined first. The next smallest rescaled distance is between Denmark and Sweden and so they form the next cluster. Once two countries A and B have been combined into a cluster, they are considered to be at the same distance from another country C, which is equal to the shorter of the distances AC and BC. This construction is then generalized up for clusters with more than two countries. The distance between any two clusters is the shortest distance between any two countries in the two clusters. Progressively more countries and clusters are combined in this way, with some countries combining with existing clusters, until all the countries are united into a single cluster. The dendrogram shows quite explicitly that the voting patterns of certain countries are highly correlated. Greece and Cyprus have a very small rescaled distance which demonstrates a very strong voting correlation. This particular finding thereby confirms a long-held belief among regular Eurovision viewers. There is a slightly less correlated cluster that involves the Nordic countries, Denmark, Sweden, Iceland, Norway, Finland, and somewhat surprisingly, Estonia. Many other clusters also arise: Bosnia and Turkey, Croatia and Malta, UK and Ireland (who also show a correlation with the Nordic clique), Belgium and the Netherlands and France and Portugal. The high correlations between the way in which countries assign points provides evidence in support of the theory that voting groups exist'[1].

Further analysis of voting patterns between countries allowed the researchers to identify several countries that appear to be more ‘in tune’ with the rest of Europe, that is, countries that are compatible with a greater number of countries than others. Compatibility between countries was measured by analysing how often a given country exchanged points with another country. If this number exceeded that expected for a ‘random contest’, the countries were judged to be compatible. The country that was found to be compatible with the greatest number of other countries was the UK, whilst at the opposite end of the spectrum France, and to a lesser degree, Spain, were found to be the least compatible with the rest of Europe. |University of Oxford News|

via Daily Favorite

[1] How does Europe

*Make Its Mind Up*[pdf], Daniel Fenn et al.Mahalanobis - am 2005-05-23 05:20 - Rubrik: sociology