So do they work or not

[gfun]

From Dft web site - see link

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

The tables published reveal that for 743 locations, casualties increased rather than decreased.

However, this does not mean that cameras have not been effective at all these locations:

There are 269 locations for which the camera has not been in long enough to judge it's success - these camera sites were installed during 2001-02 and 2002-03 and the after data may be based on limited data for example one or two quarters;
There are 229 sites established before partnerships joined the netting-off scheme where the before data is taken during the period immediately preceding the partnerships entry to the scheme rather than when they were first introduced. It would therefore be wrong to conclude that these camera sites have not been effective and we will be asking partnerships to provide true baseline figures in due course to provide a more meaningful comparison.
We therefore believe that there are 245 instances where there have not been casualty reductions at locations, the majority of which (228) pre-date the scheme. We intend to discuss these locations with the partnerships to determine what should be done in respect of those locations.


Errm?

www.dft.gov.uk/stellent/groups/dft_rdsafety/documents/page/dft_rdsafety_029198.hcsp


So if casualties have gone up how can they still be considerd to work?

>> Edited by gfun on Saturday 13th August 12:13

[autismuk]

gfun: all quoting the DfT said:

The tables published reveal that for 743 locations, casualties increased rather than decreased. However, this does not mean that cameras have not been effective at all these locations:

Cue result picking and excuses.

We want a result "x", so we take our dataset, remove all those that are "x", and then try to explain away the non "x" cases.

It's rather like saying a coin always comes up heads, because you only count the times when it comes up heads when you are tallying it.

gfun said:

There are 269 locations for which the camera has not been in long enough to judge it's success

This never bothers the DfT or the Scamera partnerships. There's a strong statistical argument that *ALL* the Scamera "results" have this problem ; that is the core of the RTTM argument.

Figures fluctuate, and small numbers fluctuate a *lot*. KSI's "doubling" may well be from 2 to 4, which actually could be a reduction in the number of accidents (2 small vs 1 big !)

Why, out of interest, are we only analysing those datasets which don't produce the results we want ?

Aren't some of the ones which are "successful" of dubious value ?

gfun said:

- these camera sites were installed during 2001-02 and 2002-03 and the after data may be based on limited data for example one or two quarters;

Or, it may not. One presumes they do actually know ; if they don't the whole exercise is pointless. And if they do know, why not say ? Such things aren't complicated, and the maths is well known.

As my local Scumera Pratnership was touting improvements after six months perhaps I am not as impressed by this argument as I might be. Especially given the before count covered a six times longer period.

And again, a vast number of Scameras must have been installed post 2001 (that's four years ago), and their data too may be "based on limited data for example one or two quarters"

It's blatant statistical fiddling ; pick the results you like, try to explain away with spurious theories the ones you don't.

gfun said:

There are 229 sites established before partnerships joined the netting-off scheme where the before data is taken during the period immediately preceding the partnerships entry to the scheme rather than when they were first introduced.

What does the "netting off" scheme have to do with anything ? This is a decision about what happens to the money !

The conclusion one draws from this is the DfT genuinely thinks that accidents are more likely, or maybe less likely, when the money goes to the government rather than the pratnerships.

This is obvious nonsense. What happens to the money affects the number of cameras and so on, but *doesn't* affect the effectiveness of an individual camera.

Also, again, doesn't this apply to some of the allegedly "successful" cameras as well ?

gfun said:

It would therefore be wrong to conclude that these camera sites have not been effective

It would also be wrong to conclude that other camera sites have been effective, for exactly the same reasons.

gfun said:

and we will be asking partnerships to provide true baseline figures in due course to provide a more meaningful comparison.

Translation is probably ; pick your baseline as a year when accidents were very bad (maybe even on a per camera basis) so that the results "improve".

gfun said:

We therefore believe that there are 245 instances where there have not been casualty reductions at locations, the majority of which (228) pre-date the scheme.

What scheme, the netting off scheme ? What does this have to do anything ?

gfun said:

We intend to discuss these locations with the partnerships to determine what should be done in respect of those locations.

Well, how about lying with figures ?


Errm?

www.dft.gov.uk/stellent/groups/dft_rdsafety/documents/page/dft_rdsafety_029198.hcsp


So if casualties have gone up how can they still be considerd to work?

>> Edited by gfun on Saturday 13th August 12:13[/quote]

[smeggy]

Dft web site said:

However, this does not mean that cameras have not been effective at all these locations: ……….. we will be asking partnerships to provide true baseline figures in due course to provide a more meaningful comparison.

“So if casualties have gone up how can they still be considerd to work? “


I wonder if the partnership’s figures will factor in effects of 'regression to mean' error (as proven by Linda Mountain’s study).


I also wonder if the partnership’s figures will also factor in improvements from other unrelated safety measures implemented at these sites, instead of giving all the credit to the speed cameras only (like they usually do).

[gfun]

So I gather you are both a bit cynical about this little gem from the Dft?

Funny so was I

G :-)

[autismuk]

smeggy said:

I also wonder if the partnership’s figures will also factor in improvements from other unrelated safety measures implemented at these sites, instead of giving all the credit to the speed cameras only (like they usually do).

Yeah. The scameras "big thing" in Norfolk was the downturn in accidents at the Ketteringham Junction on the A11 near Norwich (The other one was the infamous Grapes Hill scamera).

Only problem was that the westbound turning, which was previously a very common ratrun has now been closed completely, so it only takes a very few local cars.

Even the thickos in the Scumera Pratnership should be able to work out this reduces the movements rather a lot.

The real reason it's dangerous ? Well, there's a weighbridge, so the poor HGV drivers have to do a right turn off and onto a dual carriageway, cutting across a lane of traffic doing 70mph.

The camera ; well actually there isn't one there .... how they could claim the credit for that is beyond me. There was occasionally a mobile one on a bridge a mile away.

[smeggy]

That’s a class example.

Don’t assume these partnerships are run by ‘thickos’.
I reckon only a small percentage of the general public understand or are aware of these mathematical aberrations; I also reckon the partnerships know this and deliberately use it to their advantage. Thank God for people like Paul Smith!

[catso]

gfun said:

So if casualties have gone up how can they still be considerd to work?

Because success is measured in a different way!

have you figured out what it is yet?

£££'s