Eurovision Research Network

As everyone in the ERN knows Eurovision uses  a Borda count to tabulate scores. Each country allocates points to their 10 favourites. However, rather than allocate 10 points to the first ranked and 9 to the second, first place earns 12 points; 2nd 10 points. The idea is to skew the vote in favour of top-ranked scores…and to reduce the possibility of tie scores.

Using the awesome data available from here (http://www NULL.myledbury NULL.co NULL.uk/eurovision/)I yanked the Borda count down to a normal count. Pretty simple really: convert the 10 point allocations to 9, the 12 point ones to 10.  Doesn’t change the results, but does give a clearer idea of how closely the songs scored.  The italicized songs are the top 10 for each night.

I’ve resisted the temptation to calculate z scores though.  

Semi-final one: (Note the gap between Iceland and Portugal – 32 points)
Belgium    152
Greece    128
Iceland    118
Portugal    86
Russia    74
Serbia    74
Albania    71
Bosnia & Herzegovina    57
Belarus    57
Moldova    52
Finland    48
Poland    44
Malta    43
Estonia    39
F.Y.R. Macedonia    34
Slovakia    24
Latvia    11

Semi-final two: (note how closely Azerbaijan, Romania, Georgia and Denmark are clustered)
Turkey    111
Azerbaijan    104
Romania    101
Georgia    99
Denmark    96
Armenia    76
Ukraine    75
Israel    69
Ireland    64
Cyprus    64
Sweden    57
Lithuania    42
Croatia    31
The Netherlands    29
Bulgaria    19
Slovenia    6
Switzerland    2

Final: (Germany still wins by 6o+ margin; Turkey and Romania much closer; places 4 through 9 separated by 8 points)
Germany    223
Turkey    159
Romania    156
Denmark    138
Azerbaijan    136
Belgium    135
Armenia    134
Greece    131
Georgia    131
Ukraine    106
Russia    83
France    82
Israel    69
Serbia    68
Spain    66
Albania    59
Bosnia & Herzegovina    48
Portugal    43
Iceland    41
Norway    35
Moldova    26
Cyprus    25
Ireland    25
Belarus    16
United Kingdom    10

I’d be curious about the thoughts of any others quantitatively bent: should we stick with a Borda count?