ISA advocates are making fallacious casualty reduction claims
By Nigel Humphries, Association of British Drivers
The case presented for Intelligent Speed Adaptation depends very heavily on the use of ‘risk curves’ to predict the reductions in casualties deriving from the imposition of external speed control (ISA: a panacea for speeding or a step towards a big brother future?
LTT 16 Jan). Since these have been elevated in importance above the reports on the causes of real accidents, they require some further investigation.
Risk curves have been presented by ISA apologists as robust, scientifically derived measures of the risks of driving at a certain speed in a certain place. A read of TRL report 421 The effects of drivers` speed on the frequency of road accidents
shows that they are nothing of the kind. Risk curves are created in this report by assuming that there is a progressive causal relationship between very small changes in speed and accident involvement and then referring back to other pieces of research to quantify this.
There are two basic types of research referred to: first the ‘road-based’ work cited to in the LTT article, which relates average speed to accidents on individual roads. This has been exposed as flawed by the Association of British Drivers for many years, ranging from our critique of Finch et al
, which was quoted in your article, to a more recent in-depth analysis of TRL511
— The relationship between speed and accidents on rural single carriageway roads
, which also appears on the ABD website.
The second type is ‘driver-based’ research. This refers to a large study where ‘spot’ speeds of drivers were measured in free flowing traffic and the drivers were then sent a questionnaire asking them how many accidents they had been involved in during the past three years. The claim that followed was that a 1% variance in the driver`s speed from the average was associated with a 13% change in accident liability. This is breathtaking — I suggest readers go back and absorb the last couple of sentences. Yes, they are indeed claiming that travelling 1% faster on an A road in Berkshire makes it 13% more likely that you have been involved in an accident, irrespective of the circumstances and blame, anywhere in the UK.
This is madness. As John McEnroe would say “they CANNOT be serious!”
My operational research background has taught me to be very suspicious of any study that attempts to find a link between one variable in the whole population and relatively rare events with vastly disparate and complex causes. The whole thing is wide open to common causes (it is dangerous to assume that when two variables appear to be linked, one causes the other — there may be a third variable causing both) and bias, especially when it involves questionnaires.
The report writers make two high-handed assumptions that are in my experience incorrect. First, that a driver`s speed relative to others in one place is the same as in another — i.e. that a fast driver is always fast. Anyone who has followed a slow driver along an open A road at 40mph only to see them continue at the same speed through a village can demolish that argument. So can someone who has been held up while driving a low powered car around a bend by a big saloon that floors it and leaves them behind as soon as an overtaking opportunity arises.
The second assumption is one of causality, which is just assumed to be true without any evidence being offered.
Without causality, the justification for ISA disappears. Even if it is true that someone driving faster than average on one road is more likely to be involved in a whole spectrum of accidents on other kinds of roads, many of which would occur well below the speed limit, then why would suppressing that speed choice some of the time using ISA make any difference to the number of accidents? The psychological causes of the accidents would still be there and this is the common factor causing the statistical link.
The true way forward for road safety is hinted at on page 14 of TRL421 where it says that, in the driver-based study: “The accident involvement was also related to hazard involvement (the frequency with which drivers report that they have found themselves in hazardous situations as a result of perceptual failure).”
In my experience, this ‘hazard involvement’ fell away to nearly zero in my late 20s when I worked out, with the help of others, what safe driving was all about. At the same time, my speeds increased drastically everywhere except the areas of ‘hazard involvement’, where they fell. The ‘hazard involvement’ only threatens to return when I am forced to drive slower than road conditions dictate. This is a common shared experience amongst skilled drivers and means that, for them, there is an inverse relationship between spot speeds and hazard involvement unless you happen to measure the spot speeds near a hazard.
ISA cannot make people competent drivers. It could have a role as part of a strategy to improve driving skills in the period immediately after passing the test, but as a panacea, no way.
Association of British Drivers