dirtylitterboxofferingstospammers said:From this week's New Scientist - a review of a book, "More Damned Lies and
Statistics" by Joel Best - £12.95University of California Press
"DID you realise that schools in the US have been getting steadily safer,
despite terrible incidents such as the massacre at Columbine?
That statistic - that schools are getting safer - may feel like a comfort
blanket of truth in an uncertain world. In More Damned Lies and Statistics Joel
Best breaks the bad news that this is false comfort. He has bad news for anyone
who wants to misuse statistics, especially in the sphere of public policy,
often a stronghold of bafflement."
and
"Best concludes that statistics should be taught as a social science, not as
part of mathematics. In the meantime, he advises anyone seeing accounts of,
say, a 10 per cent increase in the risk of a particular disease to pay no
attention. Anything less than 200 per cent is just noise."
The review is by Marrtin Ince is a freelance journalist.
The review echoes what I personally feel about statistics, especially having
studied them at uni for a while. They are *so* easily skewed to fit any
particular viewpoint. I tend to take many with a very, very large pinch of
salt.
Just my sort of digressing £0.02 really
Cheers, helen s
Thanks for the reference to the review of the book Helen.
All this badmouthing of statistics seems to miss an important point. Substitute our everyday langauge for statistics - if you follow my drift - people manage, without any difficulty, to lie using our written and spoken language - but we don't suggest that everyday language can't be trusted because it's so often misused or used in the service of deception.
The point I'm trying to make is that people misuse language and they misuse statistics. We don't eschew the use of language because it's so often used in the service of deceit. People use language in deceiving not only themselves but in deceiving others - same goes for statistics. Just as we can become adept in spotting other's attempts at deception in their use of language - similarly we can do the same with statistics - all that it needs is the desire and the effort to find out whats in the bag of the dirty tricks that are so often tried on. I'm harping on here because I get peeved with this silly talk suggesting that we can't trust any statistics - what we do need is to get wise to the basics so that we don't get taken in by statistics from snake-oil vendors (social "scientists" come in a close second).
Speaking of which - I was flitting through a book by John Allen Paulos "A Mathematician Plays the Stock Market" - and he makes mention of one particular dirty trick, of which Guy (I think?) was complaining about recently - the BHIT lies of vastly exaggerating the HI deaths - anyway Paulos mentions the caper often used by groups of this ilk - how they publicly overstate by an order of magnitude or two, some horrifying statistic in order to arouse public concern for their cause. Even though they are inevitably hauled up on their exaggerated claim the damage is done in that most people who hear the exaggerated claim don't tend to revise it downward to anything like what it should be.
What Paulos points out is how people using this trick take advantage of what is known as the anchoring heuristic which seems to be one of the many foibles of our error prone faculty of intuitive reasoning. The first figure that is presented tends to act as an anchor, and from which subsequent revisions don't seem to deviate from much.
To illustrate this efect we could make a hugely exaggerated claim, that for example find out how many helmeted cyclists die from head injuries and multiply it by say 20 or 50 (why not- let's not let truth get in the way of a good story) - so we circulate this claim widely and inevitably we get hauled up that it is vastly exaggerated. The anchoring heuristic will suggest that for most people who hear this first exaggerated claim any subsequent downward revision that they make will make will not deviate radically from the first (false) figure that they were exposed to.
In Paulos's other books his general theme is to illustrate and decry statistical illiteracy - he coined the term innumeracy - check out the book by that title (Innumeracy, Penguin 1989 )
Roger