Road safety expert Idris Francis says don't rely on the data from individual speed cameras for their impact on accidents.
I wondered yesterday on seeing wall-to-wall media coverage of results from individual speed cameras, seemingly based on the assumption that any change in accident levels must be due to camera effect alone, whether the boy who exposed the Emperor or those who denied that the earth was flat ever suffered from self-doubt before speaking out. Given the majority view, might they be wrong? Or were they so certain that doubt never entered their minds?
Having checked my assessment with others and been reassured, I return here to the central point I made on 26th June in some detail, that because accidents usually arise from the near-random coincidence of several causal factors out of the 77 now recognised by police and DfT (and because many of those factors cannot be quantified even when they can be identified) the limited and often seriously flawed data from any one speed camera site cannot possibly have any statistical significance.
In simple terms, and because we can never know what accidents would have happened had the camera not been present, it is entirely possible (assuming for the sake of argument that cameras can sometimes be effective – a debate for another day) that accidents can rise despite an effective camera that prevented a larger rise, or fall where a dangerous camera prevented larger falls. (On that last point, the Highways Agency, Transport Research Laboratory and DfT all now admit that cameras can and do cause accidents, though the DfT to this day refuses to research this important issue).
Another fundamental point is that due to long-term downward trends and cameras usually being installed where there was a recent history of high accident numbers, accident numbers at sites would normally be expected to fall even if nothing were done, hence the absence of a fall implies adverse rather than nil effect.
Another is that because speeds above limits are in fact a relatively trivial causal factor (IAM report) even the maximum possible supposed camera benefit is so small that it cannot be separated out from the many other near-random causal factors, and all the more so at individual sites where the small numbers are especially volatile.
My first article on these issued was triggered in June by media reports such as that inter alia that "Mr Penning said …"This new data will end that by clearly showing whether a camera is saving lives or just making money." I asked my MP George Hollingbery to copy my detailed concerns about data for individual cameras to the Minister but was puzzled by the 15th July in his name that "The Government has not requested speed camera data for its own use and it is not true to suggest the Government is forcing local authorities to announce which speed cameras are cost effective."
I remain unable to reconcile either statement with media reports of late June or the DfT Press Release of 24th August stating that:
"All local authorities were asked to publish information about the effectiveness of their speed cameras as soon as practicable and provide a web link to this material by 20th July 2011……Local residents have a right to expect that when their council spends money on speed cameras, they publish information to show whether those cameras are helping to reduce accidents or not."
It might, I suppose, be argued that the last few words relate not to single cameras but average of many but this was certainly not the way it was widely reported in, for example, the Telegraph or the main television news channels whose reports related almost exclusively to individual cameras. Further, in at least one television news report Mr. Penning himself spoke in clear terms of results at individual sites as if they indicated success or failure of the particular camera.
Despite Mr. Penning's denial to me, the two reports also states that "officials from the Department of Transport will conduct a detailed statistical analysis of the data ….to assess the effectiveness of speed cameras in improving road safety". If by that the DfT means analysis of average results of large numbers of cameras, taking fully into account the whole range of factors which affect the results – well fine, not before time, previous attempts from the original eight-area trial of 2000 to the infamous 4th Year Report of 2006 to the 2010 RAC one consisted largely of wishful thinking mixed with large quantities of whitewash, weasel words and half-truths, as I am gradually documenting on www.fightbackwithfacts.com.
My message to Councillors, MPs, Ministers and the DfT is therefore a very simple one – please do not waste your time and our money analysing data which has no statistical significance, please stop pretending that whether accidents at a sites go up or down tell us whether the camera is effective or dangerous, please stop ignoring – and this to the Minister and DfT- please at long last investigating the adverse effects which kill and main and ensure that your Press Releases are more carefully worded so that numerically challenged reporters are not misled into even worse misrepresentation. Oh and one last thing – will you please, please now start listening to those who have been pointing out for twelve years or more that much of your information and analysis is seriously flawed?