DfT skewed data and analysis

Edited 3rd July 11

This is a summary of how the DfT managed to get their comparison of the cost effectiveness of speed cameras and vehicle activated signs wrong by a factor of 50 to 1.

Having realised what should have been obvious from the beginning, that Partnerships whose income and jobs relied directly in speed camera fines would not be at all keen to use “perhaps more cost effective methods” the DfT ended the hypothecation scheme in April 2007 in favour of funding via local authorities. Taken by surprise by Transcom’s 2006 request for details of these “perhaps more cost effective methods” the DfT dared not submit the figures readily available from years of data on thousands of cameras and TRL 548 (a large scale study of signs) that would have exposed them as having pursued a speed camera policy that was hopelessly ineffective in cost.benefit terms they instead submitted – ludicrously – data for only one camera site and one sign site, data which as any student statistician, let alone the DfT must know to be so limited as to have no statistical significance.They then systematically, cynically – and clearly deliberately – massively skewed the results in favour of cameras by:

1/ Quoting nonsensical costs of £7,500 for a camera but £14,000 for a sign – a far less complex device. ( I later forced Stephen Ladyman to admit in writing that they had failed to include the £32,000 cost of the camera itself, and that the £14,000 paid for two signs not one.)

2/ Completely ignoring the massive costs of the enforcement system.

3/ Misrepresenting the falls in accidents. At the camera site they fell from 29 in 5 years to 18 in 5 years, i.e. by 38%. At the sign site they fell from 31 in 10 years to zero in 5 years, or by 100%. The  ratio of effectiveness was therefore 2.6 to 1 (100/38) in favour of signs but the DfT used a comparison of  3.1 (31/10) to 2.2 (29-18/5) or 1.4 – and in that way alone skewed the result by 1.8 to 1 (2.6/1.40) in favour of cameras. (In my day a third form arithmetic studend who made such mistakes would have failed his exam, but in 2006 we have statisticians at the DfT basing vital road safety decisions on such nonsense.)

4/ Basing the comparison only on First Year Costs – an insane method that no commercial company would dream of using. As anyone who has ever run a company knows, viability of such schemes must be assessed over the useful life of equipment – in this context of the order of 10 years. Few commercial organisations would expect any scheme of any size, involving research, design, tooling and sales, to be in proft in the first year. In terms of the cost effectiveness of cameras versus signs this is a very significant factor, because speed cameras continue to incur high annual costs through maintenance and recalibration but also in particular through the very expensive enforcement system. Signs on the other hand cost typically £200 a year to maintain, with no enforcement costs whatever. The overall effect on the bottom line was that while cameras cost about 10 times as much as signs to install and operate in the first year, they cost more than 50 times as much over 10 years due to their high recurring costs. (See the independent Accountant’s report)

By means of these skewed figures and analysis the DfT achieved its objective of misreprenting a cost effectiveness advantage for signs of some 50 to 1 into a 12% advantage for cameras. That Transcom and its advisers were prepared to believe and publish such nonsense was bad enough, but that both they and the DfT then flatly denied that the figures were wrong when I sent them clear evidence to that effect is even worse, and serves only to confirm intent to mislead, and to avoid at all costs admitting that they had both been wrong all along about cameras.

All of this is fully detailed in the documents attached to the Cameras v Signs page, including Ladyman’s admission to Transcom that the figures were indeed wrong – and their joint decision that this need make no difference to their preference for cameras. Now please read on. As always, if you see any error at all, or have any questions, please contact me at irfrancis@onetel.com

Comments are closed.