SEXY STATISTICS? Sexy statistics?
Lies, damned lies and statistics. Why do people mistrust statistics yet use them to justify opinions? And why do journalists and politicians persist in quoting them without proper understanding? John MacInnes explains why statistics are routinely abused and what’s being done to improve standards
T
HE RECENT RSS/IPSOS Mori report shows that the public’s ignorance of many of the key statistics in public life is second only to their distrust of statistics as such. At least this is coherent: why bother about something you don’t believe in! Media professionals know this only too well:
they focus on ‘the stories behind the numbers’ and use statistics, if at all, as a fetish that warrants the ‘scientific’ credentials of their argument. Any number will suffice. ‘Nine out of ten cats prefer…’ is nonsense, but it makes little less sense than most of the numbers in public discourse, from the Sun to the Financial Times.
If you doubt this, ask yourself the following three questions. What is the approximate size of the national debt? And of the national deficit? And what is the difference between a debt and a deficit?* This has been the focus of British politics for the last five years, yet I doubt if one voter in a hundred has these figures to hand. Inspired by a Paul Krugman comment, Google recently surveyed US voters’ knowledge of its deficit. A vast majority thought it was rising, with 40 per cent saying it was rising fast. In fact it is falling rapidly. This should not surprise us. We know that
typically people form opinions and then become alert to ‘evidence’ that reinforces them, regardless of its quality or veracity. Cognitive psychologists such as Nobel laureate Daniel Kahneman and Amos Tversky have shown that the human brain can effortlessly construct convincing cause-effect stories from flimsy or just downright irrelevant material. We effortlessly ‘see’ stories in random noise. It is possible that the origin of this lies in the
evolutionary advantage of being alert to danger: that pattern of long waving grass might be a tiger and it is better to be safe than dinner. But such an account is, of course, itself an example of this cognitive bias. Conversely, the human brain finds statistical reasoning laborious, slow, error-prone work. An earlier Royal Statistical Society (RSS) enquiry
asked MPs a simple probability problem: what was the chance of two heads if a coin was tossed twice? MPs are continually making decisions that require them to weigh up competing probabilities, yet only 40 per cent answered correctly, even though the question was multiple-choice. I defy anyone to produce a copy of any newspaper that is free of some
16 SOCIETY NOW AUTUMN 2013
The mathematics curriculum is almost devoid of statistics – but that could be changing
statistical howler. The Telegraph recently reported that there were 100 cod left in the North sea. Presumably the copy editor had neither eaten fish and chips recently, nor paused to ponder how one might possibly estimate such a tiny number in such a vast expanse of ocean. Had (s)he fancied a glass of wine with their fish supper, the Daily Mail it came wrapped in might have put them off. It recently declared that a glass of wine increased ‘the risk of death by 15 per cent’. Before you conclude that a bottle of wine is lethal, keep in mind that we all have a 100 per cent risk of death. Statistical reasoning may be beyond journalists or MPs, but it lies at the heart of almost all science, because it is virtually the only way to produce forms of knowledge that can escape from human observers’ cognitive biases. One can think of the explosion of knowledge that has transformed the world over the last five centuries as the rise of statistical reasoning: the substitution of logical evidence-gathering, interpretation and testing for leaps of faith or dogma. This is especially true of the social sciences where the signal method of the natural sciences – the experiment – is rarely possible and the alternative – systematic observation – is inherently
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32