Dear Councillor,
I have spent many thousands of hours since 2001 analysing road casualties near speed camera sites. I can assure you that the claims made for them have always been nonsense, and that the similar figures submitted to you recently (as quoted in the Local Transport Today, below) are abject nonsense both in terms of the reductions achieved and their notional values.
I am now nearing completion of a detailed but esssentially simple analysis of KSI from 1987 to 2011 near speed camera in the 32 police areas for which the relevant data is available, and I expect to publish it within a matter of weeks. I will send a more detailed email to arrive by Monday morning, but for the moment are a few graphs which showing that there is no sensibly quantifiable reduction in KSI near camera over and above what would be expected anyway due to trend and regression to mean.
1/ London, some 688 fixed camera sites. Graphs drawn from TfL's recently published data
The blue line shows how KSI inside the areas which received cameras at some time during those years, and the maroon line shows KSI where no cameras were installed. The increase in the blue graph in the middle years was because those sites were (for the most part) explicitly selected for recently higher KSI numbers, and the return to trend as the number of new cameras dwindled in the late 00's was entirely due to regression to the mean aka return to normal. There being no logical reason why camera effect would be significantly different there than in Northumberland, this exposes the claims of 45% and 71% of the observed falls having been due to cameras as utter nonsense.
2/ 1/ London, some 294 red light camera sites. Graphs rom TfL's recently published data +
As before, no quantifiable benefit. The absence of the mid-term higher numbers is presumably because red light camera sites are not selected in the same way.
3/ London Fixed camera, but trend adjusted and time-shifted
.
This graph uses the same data as the first but shows directly the comparison between camera sites and other areas. The data is also time-shifted to appear as if all cameras had been installed in the same year, no 14.
The raised mid-term site selection numbers are clear, as is the fact that numbers fall back accurately to normal levels due to regression to mean, by year 14 - and do not then change at all. I should also mention not only that (competent) statisticians have known for 100 years that this is what always happens when a group of higher than normal numbers is selected from a larger group, and that those numbers always fall back to normal the moment the selection period ends. And I have graphs of KSI by month for 220,000 notional 1k sq areas showing that that is indeed what does, and must happen.
4/ 3-month and 12-month average KSI at Fixed camera sites in 26 police areas including London, Wales and Scotland
This is a graph by quarter and by year of KSI numbers from 96 months before installation to 36 months after installation. In the absence of site limit data from most police forces this graph is based on KSI within 500m of each camera, approximating to the supposed area of influence of those cameras. Numbers have again been adjusted for trend ie compared to non-camera sites, so that nil camera effect would be a horizontal line.
Here you can see again the higher numbers in the middle section due to site selection criteria, the lower normal level before those sites were selected, that numbers had started to fall before camera installation -- but (crucially) failed to fall back all the way to the prior normal level, as RTM would have predicted. Nor is there any sensibly quantifiable reduction after installation. The net effect therefore appears to be more accidents after installation than would otherwise have been expected.
Please note too that very large numbers of accidents across the country contribute to this graph and others like it, the numbers have not been cherry-picked to arrive at a predetermined results (as happens all too often)
As I say, I will send more information for Monday morning, and urge you not to be taken in any longer by incompetent analysis and vested interests. Please understand too that I am not asking you to stop spending public money on road safety, what I am asking you to do is to stop spending public money on speed cameras that at best do not work and at worst cause accidents that would not otherwise have happened, and to spend it instead on better and more cost effective methods. And on methods that do not waste money on hundreds of people sitting in offices filling in data and processing penalties, or for that matter penalise millions of safe drivers, cost thousands of them their licenses and many of those their jobs and, in far too many cases, their liberty.
I should also say that the DfT's "values" for accidents supposedly prevented are tripe. Some 60% of it is a purely hypothetical figure for what the public would be willing to pay to prevent accidents (ie not cash at all) and the rest of it is the "lost output" that casualties would have produced had they not been killed or injured. Again, nonsense - when someone is killed his output does of coures end - but so does his consumption, which more than cancels out the output!
As well as providing this information by email I would be happy to visit you at my own expense, to clarify all of these issues.
Yours sincerely