ABD's lonely factoid

'''How dangerous is speed? The ABD'S lonely 'factoid' and the real world'''

From Slowdown, the newsletter of the Slower Speeds Initiative, Spring 2001

The Association of British Drivers (ABD) likes to cite a (TRL) report as a source for the true contribution of speed to road crashes and casualties. ABD members use the TRL report to contradict the 'mainstream' figure indicating that at least one-third of crashes are speed-related. According to the ABD, the TRL report proves that the true figure is under 5%. This is the only source of such a low figure. The ABD and a few motor lobby journalists are the only people to use it, generally to support the argument that "it is not speed but bad driving that is dangerous". The ABD especially likes to use the figure in letters to local papers where highway authorities are implementing speed control measures in response to deaths and serious injuries or local demands for safer communities. (Their preferred technique is for one or two writers to flood papers with pseudonymously penned letters to make it appear they have widespread public support.)

The Slower Speeds Initiative wrote to the Transport Research Laboratory concerning the ABD's use of the study. The TRL referred us to reports on speed. This is because the TRL study cited by the ABD, TRL Report 323, concerns "A new system for recording contributory factors in road accidents". TRL 323 is not a study of crash causation. It is a study of how to collect data. It was not designed to draw statistically reliable conclusions about the causes of road crashes. The accidents included in the three month study were not a statistically representative sample of all accidents. There is no basis for using the study to generalise about the speed-crash relationship.

The very low figure quoted by the ABD comes from a table which showed pairings of factors: In 4.04% of crashes recorded in the study, the person filling in the form paired 'excessive speed' (nowhere defined) with 'loss of control of vehicle'. 4.04% is only a subset of all speed related crashes recorded in the study. This use of statistics has been described by a professional statistician as "extremely naughty" and by the Department of the Environment, Transport and the Regions as "mischievous". DETR go on to say "it is interesting that none of the many other TRL reports on speed and accident risk have been mentioned by those using this report as the basis for their argument." The TRL 323 methodology for recording contributory factors simply does not ask the questions which would reveal the inherent dangers of speed:


 * Would the factor still have been present if the driver, and/or all the other drivers involved, had been driving more slowly?
 * IF YES, Would the factor still have resulted in a crash?
 * IF YES, Would the crash still have been so severe?

It is obvious to almost everyone (with the exception of libertarian motorists with a soap box to drive) that higher speeds reduce the amount of time any driver has to respond to the unexpected and that higher speeds increase the force of any impact. The importance of reduced speeds to crash prevention and reducing crash severity is no mystery. In fact, the TRL study beloved of the ABD and its fellow-travellers, indirectly acknowledges the overriding importance of speed:

"Virtually the only factor that road accidents have in common is that all would have been avoided if those involved had known with certainty, a few seconds in advance, that an accident was about the occur."

Lower speeds provide those few extra seconds.

Among the TRL reports the ABD does not like to cite is TRL 421, "The effects of drivers' speed on the frequency of road accidents" published in March 2000. Unlike TRL 323, this study was designed to discover the speed-crash relationship.

The authors looked at 300 sections of road, made 2 million observations of speed and got 10,000 drivers to complete questionnaires. They found that


 * the faster the traffic moves on average, the more crashes there are (and crash frequency increases approximately with the square of average traffic speed)
 * the larger the spread of speeds around the average, the more crashes there are

Significantly for the ABDs argument, and for the rest of us, they also found that:


 * drivers who choose speeds above the average on some roads tend also to do so on all roads
 * higher speed drivers are associated with a significantly greater crash involvement than are slower drivers

For these reasons they conclude that the speed of the fastest drivers (those travelling faster than the average for the road) should be reduced. The study confirmed what is described as a 'robust general rule' relating crash reductions to speed reductions: for every I mph reduction average speed, crashes are reduced by between 2-7%. More specifically, the crash reduction figure is around


 * 6% for urban roads with low average speeds
 * 4% for medium speed urban roads and lower speed rural main roads
 * 3% for higher speed urban roads and rural main roads

To put the dangerousness of speed into perspective, how many drivers care about or would notice a 2mph reduction in their average speed? Yet, averaged across the entire road network, a mere 2mph reduction in average speeds would prevent more than 200 deaths and 3,500 serious casualties a year. The authors of TRL 421 suggest that this target (about a sixth of the overall speed related casualty figure) is a 'reasonable minimum' to aim for. More importantly they use it to show "the sensitivity of accident numbers to a small change in average speed". In other words, speeds that might not seem excessive. Speeds that TRL323's methodology wouldn't even record.

Thanks to Stephen Plowden, Rosamund Weatherall and DETR