The Stone Report - now available

[safespeed]

Hi All,

After all that stuff on Radio 4 last week, The Stone Report has finally been made public, but not by the BBC.

Professor Stone has made it available from UCL:
www.ucl.ac.uk/Stats/research/Resrprts/speed.pdf

Paul Smith's evidence document:
www.safespeed.org.uk/againstcameras.doc

Paul Smith's reply to the Stone Report, plus links and other information:
www.safespeed.org.uk/stone.html

Don't miss the audio links on the BBC web pages.

[Peter Ward]

Fascinating report! There's so much that's quotable, but I do like the following from page 5:

report said:

I give some details of the South Wales study in my Appendix C: the estimated effect (a 51% reduction in PIAs) was confined to within a few hundred yards of the cameras. The paper puts this nding nicely into context: In 2000 only 2.9% of all crashes [PIAs]in South Wales occurred within a 500 metre route of [101] mobile speed camera sites included in the present analysis. Thus the effect across the entire area would be expected to be a 1.5% overall reduction, a figure that is well within year-to-year variability. To have a much greater effect, cameras would need to be employed much more widely, and perhaps randomly."

As far as I can tell, this means that a fairly scientific study of mobile speed cameras in S Wales says:
- only 2.9% of injury accidents occur at mobile speed camera sites
- the average effect of mobile speed cameras is at the noise level.

Paul, have you made a precis of the report highlighting the important points?

[safespeed]

Peter Ward said:

Paul, have you made a precis of the report highlighting the important points?

I did send a PR: www.safespeed.org.uk/pr130.doc

And there's the "reply" on: www.safespeed.org.uk/stone.html

My favourite quote is:
"But no-one can say that these localised savings may not be outweighed by an irritation-induced increase on the 99% plus of the road network that is well away from any safety camera."

[pdV6]

Peter Ward said:

- only 2.9% of injury accidents occur at mobile speed camera sites

Blackspots? Hmmm. Didn't think so...

[safespeed]

pdV6 said:

Peter Ward said: - only 2.9% of injury accidents occur at mobile speed camera sites Blackspots? Hmmm. Didn't think so...

Blackspots are a bit like that these days. Obviously where there's more traffic there tends to be more accidents.

And we do have "special case" places where something in the environment causes many to make similar mistakes. But according to a traffic engineer I was speaking to:

In the 1960s 50% of all accidents used to take place at these sort of traditional blackspots but they have been treated and now blackspots account for less than 20% of all accidents.

Therefore over 80% of accidents are fairly randomly distributed across the road network.

[IAnReid]

safespeed said:

Hi All, After all that stuff on Radio 4 last week, The Stone Report has finally been made public, but not by the BBC. Professor Stone has made it available from UCL: <a href="http://www.ucl.ac.uk/Stats/research/Resrprts/speed.pdf">www.ucl.ac.uk/Stats/research/Resrprts/speed.pdf</a> Paul Smith's evidence document: <a href="http://www.safespeed.org.uk/againstcameras.doc">www.safespeed.org.uk/againstcameras.doc</a> Paul Smith's reply to the Stone Report, plus links and other information: <a href="http://www.safespeed.org.uk/stone.html">www.safespeed.org.uk/stone.html</a> Don't miss the audio links on the BBC web pages.

The Stone report is an example of what Orwell called doublethink. It is full of prose demolishing the evidence for speed cameras and then agreeing with the PACTS guy that they are a good thing. It is as if the whole establishment in this country have frimly inserted their arses up their behind on this issue.

To criticise the report in full would take a long time, but for a starter here are two gems that I have spotted. The first is talking about the questions that are asked to get the 80% of the public support speed cameras figure

With a little effort, you can see that, in the new seven there are now five statements favouring cameras, and only two against. If, as I believe to be the case, there is a sizeable majority of the population in favour of speed cameras,
why use such a loaded instrument to determine or inflate it? More significantly, why do something that could be destructive of public trust?

It doesn't seem to occur to the good Professor that perhaps the reason they load the question is because unlike him they don't believe a large majority of the public support their use. And isn't there perhaps a hint of his own bias in his statement.

In one of the other appendices he talks about the West London Speed Camera Demonstartion project. Here is an excerpt from his critique of this report

The study used what it called "rigorous statistical analysis" to suggest that this reduction was a real effect of the cameras. The report misused the widely-taught chi-square test of statistical significance in its attempt to make a
"rigorous statistical analysis" of the big differences between the percentage changes (in accident numbers) for the 10 different stretches of road.

He concludes with

From 1992 onwards, cameras were introduced all over London, and the project was therefore without any scientifically respectable control data.

What cameras were introduced with no evidence of their effectiveness, misued statistics. Surely not. Quite apart from all the statistical fiddles he documents in this study, he fails to mention the even bigger flaw. During the course of the after period of the study the A40 was upgraded to a three lane dual carriageway, which surely had a lot to do with the reduction in casualties on this stretch of road, which just happened to be the most dangerous in the study. Here is the exact wording from the study

Old Western Avenue Original alignment of the A40 Western Avenue in the vicinity of the Long Lane junction in Hillingdon. (Note: the alignment of the A40 changed during the period of this analysis, but the old alignment is included for consistency and to provide a complete view of data for the area.

Perhaps the data was included because it made the figures look better for speed cameras? No, again surely not.

The study can be viewed at
[url]www.dft.gov.uk/stellent/groups/dft_rdsafety/documents/pdf/dft_rdsafety_pdf_023366.pdf[/url]

So as I said at the start the report is full of rubbishing of all the statiscal claims that speed cameras actually work, nonetheless he comes down on their side. Still we are preaching to the converted here. Those who wish to believe will continue to believe, no matter what the evidence layed before them.

>> Edited by IAnReid on Wednesday 30th June 14:07

[safespeed]

IAnReid said:

The Stone report is an example of what Orwell called doublethink. It is full of prose demolishing the evidence for speed cameras and then agreeing with the PACTS guy that they are a good thing.

I think you misunderstand the way that Professor Stone perceived his duty. Within the time available he did his best to sift the evidence in order to illuminate the debate. I think he did a fair job of that and strongly rejected some of the rubbish evidence on which the government relies. He rejects TRL421 and TRL511 utterly - which is very important - but he chose his words very carefully (and he told me this on the phone) "to protect the innocent". He actually says:
"A second strand of Mr Smith's case rests on his questioning, verging on the censorious, of the propriety of attributing causality to the role of speed in empirically established relationships between speed-based measures (such as its mean and coefficient of variation) and accident numbers. In this, I think he has a valid point. But it is one whose proper place is in the scientific discourse that organisations such as TRL Ltd should now be willing to entertain."

[kurgis]

Surprised they used chi-squaring and not polynomial trend analysis - polynomial analysis tends to take out stuff like seasonal variations and other minor fluctuations. Does help in some cases, but even that you have to be very careful with attaching to much significance to it...

Good old fashioned plain numbers still do the trick here

(with a little polynomial trend stuff )

[bogush]

IAnReid said:

In one of the other appendices he talks about the West London Speed Camera Demonstartion project. Here is an excerpt from his critique of this report

But isn't it the "West London Speed Camera And Red Light Camera Demonstration Project?!

And weren't there three or four new roads/major junction improvements, not just the bypass?!

[IAnReid]

bogush said:

IAnReid said: In one of the other appendices he talks about the West London Speed Camera Demonstartion project. Here is an excerpt from his critique of this report But isn't it the "West London Speed Camera And Red Light Camera Demonstration Project?! And weren't there three or four new roads/major junction improvements, not just the bypass?!

The title of the report only mentions speed cameras. Examining the text further there are mentions of the red light cameras in there. You are right about other changes at the time. The A406 NCR was improved at about the time that the cameras went in, I know I can remember sitting in the jams as the roadworks were done. And I beleive the Hayes bypass was built. Amongst all the statistics is the fact that at one camera location the accident rate increased. This is explained by the fact that the traffic levels increased. Everywhere the rate reduced it was the cameras wot done it.

So in fact the before and after comparisoms are completely invalid. These invalid statistics then had to be fiddled to get the right result. Remember it was this "evidence" which was used to inflict speed cameras on the rest of London, and ultimately the country.

And they wonder why we are sceptical.