Sorry I couldn’t resist that. As someone who actually works with statistics, I can confirm almost any statistics you examine in detail turn out to be based on spurious data.<\/p>\n

The Bad Science article<\/a> was particularly effective in uncovering the mythical nature of the doubling (1.4 rounded down to 1 and 1.9 rounded up to 2). Ben Goldacre even managed to explain the impact of clustering in comprehensible terms. Wow.<\/p>\n

Rounding formed the basis of the oldest forms of computer crime (Round the results of millions of money calulations down and you can amass and, obviously, remove the invisible money that results. 100,000,000 times .5p is half a million pounds. Nowadays, the banks are wise to it of course – they keep the invisible money themselves)<\/p>\n

Statistics are used to support almost any argument. This is not a fault in the statistical procedures. (I am prepared to take their value as given, insofar as I will never understand half of them.) The problem comes from the way statistics are embedded in social relations. Pure research for its own sake is rare. Who would pay for it? So any published statistics are created for some purpose – to direct the use of resources, to evaluate the success of a policy, and so on.<\/p>\n

The “so on” includes influencing public opinion. The media are not interested in publishing dull statistics but they are very keen on inflaming public opinion. We have an infinte appetite for hearing that the young people are behaing worse than in the past, so how could a newspaper resist a headline grabbing statistic that involves youth, drugs and things getting worse?<\/p>\n

Many thanks to Ben Goldacre for paying attention to a use of statistics and for having the wit to ask the rare questions – what are these numbers? what do they mean? who collected them?<\/p>\n