Thursday, September 2, 2010

Anti Speed Camera Arguments - #1 Deaths and Camera Convictions

You might think, given the current Government's stance against speed cameras, that there was a large body of evidence that they don't work. Curiously enough, I've not been able to find any evidence to that effect from an authoritative road safety source. I'd love for someone to point me at anything relevant.
So I took a look at 'Safe Speed', which is an anti-camera website that has got a fair amount of press coverage, in the hope that there was a good, properly-argued and properly-sourced case.
I wasn't convinced. So I thought I would try to deconstruct their arguments. I'll do this in a series of posts so you don't get too bored. Their arguments can be viewed at http://www.safespeed.org.uk/effects.html .

#1 Deaths and Camera Convictions.

In this section, Safe Speed try to argue that the decline in road fatalities has flatlined, so that the long-term trend in reduction of road fatalities has ceased. They argue that this is correlated with the increased use of speed cameras. There would appear to be a correlation based on the graph they present, but they've cherry-picked their reporting period - it ends in 2002. Since 2002, fatalities have declined from 3127 to 2538 in 2008 (source).
Safe Speed argue that 'if cameras saved lives' there would have been a decline to below 1500 deaths by 2002. However, they do not state their sources, assumptions or methodology, so it's impossible to know how they've arrived at this figure other than by drawing a more-or-less straight line. If only statistical analysis were that simple.
It's convenient that I have the figures for 2008, because that fills in the gap in their analysis. They give the following analysis:
Period               %Decline in Road Deaths (that is, the end total as a percentage of the starting total)
1978-1982                         86.8

1982-1988                         85.1
1988-1992                         83.7
1998-2002                       100.3

But between 2002 and 2008, we have a decline to 81.2% of the 2002 total. What does that work out at per year? 2002-2008 gives an average annual 3.13% improvement. Between 1982 and 1998, there was an average annual 3.03% improvement. Now bear in mind that the period 1982 - 1998 included two recessions, and recessions improve road safety. I am not trying to prove a causal link, but I am pointing out that the Safe Speed's graphs don't extrapolate. It's not possible to conclude from any of these statistics what the effect of speed cameras is. There are many factors that affect crash rates: increases in car ownership, increases in distances driven, a thriving economy, social factors, demographics, increase in foreign drivers, improvements in vehicle safety, highway engineering (including lower speed limits and speed cameras), lower numbers of traffic police, increases in mobile phone use, changes in public attitudes, and so on. Safe Speed's argument is that because they cannot explain the departure from a linear trend with a single factor, it must be down to speed cameras.
Safe Speed's assumption of a linear decline in death rates is clearly flawed. According to them, if speed cameras saved lives then by linear extrapolation there would be no road deaths by 2010. It seems obvious that a linear extrapolation is both statistically invalid and also isn't realistic: speeding continues despite the presence of speed cameras, and there are other factors besides speed that cause road deaths.

According to Devon and Cornwall Safety Camera Partnership, "In the first two years of [camera] operation the number of injury collisions in this area has fallen by 27% and the number of people killed or seriously injured has reduced by 7%. However on those roads where cameras are used there has been a 44% reduction in crashes and a 17% reduction in casualties. Also the average speed of vehicles on the roads has reduced by 12% and the percentage of drivers exceeding the speed limit has been reduced from 64% to 33%." It is very difficult to square that with Safe Speed's world view.

In conclusion, the arguments put forward by Safe Speed to suggest a causal link between the use of speed cameras and an end to the decline in UK road deaths are selective in their use of evidence and use very questionable methodology and statistical techniques. In other words, it is not scientific.

Next, I'll be looking at Safe Speed's analysis of how speed cameras affect drivers.

2 comments:

  1. Thank you. The serious road safety people don't bother to answer the likes of safespeed, presumably because they don't want to give them publicity. And also perhaps, because the safespeed-types appear to be somewhat unpleasant people.

    But these contrarian pillocks get quoted by the media (I once heard Edward Stourton on the Today programme refer to them respectfully as a "road safety charity") so somebody has to point out that all they are doing is throwing statistical sand in people's faces.

    ReplyDelete
  2. Indeed. I am not impressed by the BBC's reporting of transport issues in general. They seem to treat all their sources as authoritative, neutral and unbiased (particularly people like Taxpayers' Alliance), they don't seek out critical balancing opinion and they don't recognize bad science when they see it.

    ReplyDelete