hedgehog

Accidents rose at 70 sites in London

http://www.thisislondon.co.uk/news/a...0160?version=1

And in Birmingham a camera sited where there was no record of accidents has overlooked 7 people being killed or seriously injured:

http://icbirmingham.icnetwork.co.uk/...name_page.html

Jerome

I suspect if they widened the area around the camera for accident rates changes, I bet they would find the number of sites where increases on accidents occured would be much higher.

I for one knew numerous ways of avoiding certain cameras in Central London by taking back roads etc.

They should also include the volume of traffic in their equations. If the number of acidents have gone down 10%, but the level of traffic has decreased by 20%, then - if my maths isn't failing me - that means accidents have increased.

Apparition

You only have to look at the skid marks on the roads around cameras to see how they provide a potential accident black spot, which was quite possibly not there before.
Yve

hedgehog

The Birmingham site is likely to be an excellent example of the statistical artifact known as return to the mean or regression to the mean. It is likely that, on average, this road has no accidents but after the camera went up they got unlucky and there was a random spate of accidents.

Their claims of accident reductions of 40% at camera sites work on the same principle: they put the cameras where there have just been a spate of accidents and the situation returns to the normal of no accidents. Then they claim the camera caused this. If they had placed a garden gnome on a pole the accidents would have behaved in exactly the same manner.

This is why they are so keen on placing cameras where there has been a recent, but out of character, spate of accidents. The Birmingham example above shows how the need to hide behind this tactic just because the cameras may actually cause accidents if regression to the mean is removed from the equation.

It is a bit unfortunate that they are allowed to use our cash to keep lying to us like this.

AndyC_772

iTrader: (2)

Lol There's a controlled experiment I'd love to see.

Anyone care to join me in forming a Safety Gnome Partnership?

OllyK

Speed cameras reducing casualties is a classic case of the "post hoc ergo propter hoc" logical fallacy