A
Al_mossah
Guest
There was an article on Radio 4's "More or less" programme this pm (Thurs 29th Jan) , which
purports to take a look at statistics. Today they were looking at speed cameras, and had a
soundbite from our hero on it. Distressingly, probably because they limited him to a few dozen
words, he sounded almost sane.
http://news.bbc.co.uk/1/hi/programmes/more_or_less/
He argued that improvements in road safety following the placement of cameras at accident blackspots
were often entirely coincidental, and a garden gnome would have had the same effect due to the
effect known as "regression to the mean". In other words, a bad couple of years is likely to be
followed by years which are nearer to the mean.
I shall write to the BBC to suggest that if they want to use authoritative sources they may be able
to do better than our hero.
Btw, A logical extension of his argument is that speed cameras are wasted at accident blackspots.
But I thought that the anti-camera lot objected to them being anywhere but at accident blackspots!
Oh, well, why should I expect logic.
Peter.
purports to take a look at statistics. Today they were looking at speed cameras, and had a
soundbite from our hero on it. Distressingly, probably because they limited him to a few dozen
words, he sounded almost sane.
http://news.bbc.co.uk/1/hi/programmes/more_or_less/
He argued that improvements in road safety following the placement of cameras at accident blackspots
were often entirely coincidental, and a garden gnome would have had the same effect due to the
effect known as "regression to the mean". In other words, a bad couple of years is likely to be
followed by years which are nearer to the mean.
I shall write to the BBC to suggest that if they want to use authoritative sources they may be able
to do better than our hero.
Btw, A logical extension of his argument is that speed cameras are wasted at accident blackspots.
But I thought that the anti-camera lot objected to them being anywhere but at accident blackspots!
Oh, well, why should I expect logic.
Peter.