# Number of the Month

## April 2000

This month's shock horror statistic comes from the Daily Telegraph, April 25th.

13th contestant doomed to be a Eurovision flop
By David Derbyshire, Science Correspondent

 EVEN the most hardened Eurovision Song Contest cynic must surely take pity on Serafin Zubiri, this year's Spanish entrant. The pianist and vocalist has been doomed by scientists before he has had a chance to sing a single note at Stockholm on May 13. Mr Zubiri has been drawn at number 13 in the show - and that, according to a study into the chances of winning Eurovision, means he is at a serious disadvantage. Dr Richard Wiseman, a psychologist at the University of Hertfordshire, has discovered that songs at number 13 in the final do much worse than number 12 or 14. Although the reasons why 13 is so unlucky are unclear, the discovery sheds light on a few Eurovision mysteries - particularly the abysmal record of Norway. Since it entered the contest in 1960, it has been at 13 four times, including its infamous nil points of 1981. Dr Wiseman suspects that the curse of 13 is a self-fulfilling prophecy. The number is regarded as unlucky in most European countries and as entertainers tend to be superstitious, their performances could suffer, he said. Those at 13 have come bottom four times and second from bottom four times. Dr Wiseman said: "Only two countries have won the contest from the 13th position - Belgium in 1986 and Norway in 1985. Belgium is one of the few European countries to regard 13 as lucky."

So 13 won only twice in 45 years. Just how staggering are these numbers? How many times would we expect it to win? Without getting into the banalities of the contest itself, we can try some rough numbers. There are a couple of dozen countries in each final and the competition has been running for 45 years, so at random the average number of wins would be 45/24, which is about 1.9, or in round numbers 2. As for coming last (or indeed any rank) four times, we can calculate the probabilities using the Poisson distribution. The probability of four occurrences of any given rank then turns out to be 7.9%. Even an epidemiologist would blush at this level of significance. The probable number of occurrences of any rank is:

Clearly, the assumptions are crude. For example, the total number of contestant nations has no doubt varied over the years, but even if we reduce our assumed number to, say, 15, the average number of occurrences of any ranking becomes 3 and the chance of getting 2 is 23%, while the chance of four is 17%. Whichever way you look at it the statistical molehill refuses to grow into a mountain.

Consider, instead, my experiences in the village pub swindle. It is based on the weekly bonus ball in the National Lottery. It so happens that my birth date is 13, so that is the number I always choose. With a few occasional absences abroad I have paid my pound every week for a year and a half, but have never won. Some of my neighbours win frequently; one in three consecutive weeks. Furthermore, I always put in a pound for my wife for her birth date, which is 11. She has never won either. The probability of neither of these numbers coming up in that period is less than 5%, which for an epidemiologist is significant enough to publish a paper.

I used to frequent the other village pub and entered their draw every week for two years without winning. Another of my neighbours won that six times, including three in a row. Which all brings me to a thought for the month in the words of Slartibartfast in the Hitch-Hiker's Guide to the Galaxy – "That's just perfectly normal paranoia. Everyone in the Universe has that."

Footnote (added September 1st). The number 13 came up on the last Saturday in August. I was at the Eurosensors conference in Copenhagen, so somebody else got the ticket and the forty pounds prize. I wonder whether this will affect my chances of publication in the Journal of Epidemiology?