Police poison speed debate with fuzzy figures
Not just crime figures getting spin dry treatment
This may be the week when the Department for Transport learns about the risks of making a case for road safety based on figures that every expert in the field knows to be untrue.
Swindon Council got a good old-fashioned clip around the ear on Thursday from David Ainsworth, Deputy Chief Constable of Wiltshire, who declared himself decidedly unamused at its announcement that it would remove funding from fixed speed cameras in the Swindon area.
Responding to this decision, he said: "Wiltshire Police will not compromise on public safety. Together with other partners in the road safety partnership Wiltshire Police remain committed to the support of a variety of methods including the use of cameras in speed enforcement."
In other words, the local police force will carry on regardless. Just in case we hadn’t got the message, he added: "Nationally 13 per cent of all fatal casualties in 2007 were due to exceeding the speed limit." This figure is rather different from the one quoted by Peter Greenhalgh, the Tory councillor behind the idea, who said annual figures from the Department for Transport published in September showed just six per cent of collisions had been caused by people breaking speed limits.
As we explain below, the police figure is almost certainly nonsense, but when we asked them to substantiate it they simply refused to do so. So much for debating the issues.
No wonder organisations such as SafeSpeed, a national anti-speeding camera organisation, continue to gain support in challenging the official view on cameras.
One problem with speed cameras is that the government is crying wolf in overestimating their effectiveness. Faced with criticism of its speed camera strategy, the government has insisted on using figures that it knows to be inflated, thereby handing ammunition and possibly also the moral high ground to the sceptics.
As we reported earlier in the year, the level of road deaths and fatal injuries at any given camera site can be allocated to three separate and independent factors. First, there is the effect of the speed camera itself. Second is the "trend effect" – the rate at which the casualty rate is moving downward over time anyway. Third there is the "regression to the mean" effect (rtm).
That one is a little more complicated – but is a statistical quantification of the common-sense view that where a speed camera has been set up due to above-average casualties in a previous time period, some of that "before-the-camera" increase was little more than random fluctuation. Therefore any item placed by the roadside would be associated with a subsequent fall in deaths. By this means, bad statistics could "prove" that locating bananas at crash sites would reduce the level of road deaths.
This view is reinforced by Dr Linda Mountain, Senior Lecturer in the Department of Engineering at Liverpool University, who has been involved in two in-depth studies of the effects of speed cameras. She concludes that speed cameras reduce fatal and serious accidents by around 20 per cent. It is a real effect – but it is not as big an effect as the 42 per cent claimed by government. As Mountain says: "This failure to take account of rtm effects leaves the door open for criticism."
Despite this, she is convinced that the Swindon decision is the wrong one. She adds: "Speed cameras are one of many measures that have been shown to increase road safety. They have a role to play, along with a number of other techniques, such as road layout and traffic calming measures. What really matters is the combination."
Mountain is not sure that there exists any reliable study of cost-effectiveness. However, she believes that the costs associated with road deaths, as well as the fact that most cameras return far more than their operating costs in terms of revenue mean that this is one of these questions where conclusions can be reached without recourse to detailed analysis.
A further muddying of the waters comes with a 2006 study by the DfT using STATS19 data – Traffic Police estimates of whether crash victims were "exceeding the speed limit" or "going too fast for conditions" – as a means to categorise the causes of road accidents. This gave exceeding the speed limt as a factor in 12 per cent of fatal road accidents and 5 per cent of all road accidents.
An independent UK-based controlled study then used STATS19 data to show that speed cameras are effective at reducing accidents and injuries.
This finding was subsequently challenged by a paper (pdf) in the British Medical Journal, which highlighted discrepancies between Police reporting of casualty figures and hospital admissions for road accidents over the same period, and concluded that "the overall fall seen in police statistics for non-fatal road traffic injuries probably represents a fall in completeness of reporting of these injuries". So much for the statistics. Unfortunately, politics and political accounting also play a part. Swindon Borough Council complains that the £320,000 it pays annually towards just three fixed cameras could be better spent on warning signs and better lighting. The council might have a point, in the sense that that much money can pay for a lot of other measures.
Worse, from the government's point of view, is that accidents in the Swindon area have been on the increase over the last couple of years. Remember that pesky rtm effect? The one that the DfT prefers to overlook?
Well, if the upward trend in Swindon is itself little more than random fluctuation, then removing the cameras could end up being associated with a fall in accidents over the next year or so. Which would "prove" that cameras actually cause accidents?
Not exactly. But in order to counter that assertion, the time might come when the DfT itself would have to invoke the dread rtm effect. Did we hear someone mutter "hoisted by their own petard"? ®