Friday, July 22, 2011

More Media Speed Camera Nonsense

Yesterday I commented on a report that the BBC had been criticised for giving climate sceptics too much airtime. In the firing line for sloppy journalism today, it's the Lancashire Evening Post, whose journalists have a weak grasp of science and no idea of balance.

"Cash-grabbing speed cams will be reviewed"

is the headline, although the article comments

"one of them, in Eastway, close to Glencourse Drive, Fulwood, caught just nine people speeding"

Not much cash grabbed there, then.

The article quotes chapter and verse from the Association of British Drivers (ABD), which is an unrepresentative group of petrolheads with a very small membership, dedicated to seeing the back of speed cameras, speed limits and anything else that slows them down. Balancing opinion from a Road Safety Partnership, a road safety charity like Brake, or indeed anyone who knows the first thing about road danger? No thanks- don't want to spoil a good story.

What's most irritating is the total ignorance of statistical methods. If you take any group of camera sites, there'll be a spread of collision rates. Some will have more collisions and some will have less. Over time there will be variation. At some, the collision rates will go up in real terms or in relation to the average trend, and at others, collision rates will go down. The journalist writes:

The Evening Post has found three speed cameras in Preston alone where the number of accidents and casualties has increased since the cameras were installed.

Nothing remarkable in that. But the Post jumps to the conclusion that because collisions have gone up, the cameras "don't work". This is completely unscientific. Normal statistical variation could explain the anomaly, but there could be other factors. If conditions at a particular camera site have changed, this could lead to more collisions. For example:
  • nearby roadworks causing rat-running and changed traffic patterns;
  • changes to linked roads which increase traffic through certain junctions;
  • road markings such as white lines becoming unclear through wear and tear.
The increase in collisions may have been much higher but for the presence of a speed camera. Just because collisions go up does not mean it's the camera's fault. To make an assertion like that, you need either a solid statistical evidence base, or a causal link, neither of which they have. In fact, quite the opposite: there is plenty of evidence that speed cameras do reduce collisions, and no real evidence that they cause them, despite the ABD's efforts to suggest that they distract drivers or make them concentrate on their speedometers to the exclusion of all else.

The ABD spokesman offered their usual assertion that with collisions falling nationally "it's impossible” to separate the effect speed cameras have had from other factors. In fact it is possible to separate the speed camera effect by turning the cameras off and seeing what happens. This is exactly what they did in Oxfordshire. Result: a 50% increase in fatalities and a 10% increase in serious injuries. The ABD can't blame that on speed cameras.

No comments:

Post a Comment