Safe Speed home
About Safe Speed
Safe Speed response to article in Which?
See Which? magazine, October 2004


Which Magazine published an article about speed cameras in the October 2004 issue. We recognise that the author of the article made reasonable attempts at balance and quoted the ABD and Safe Speed extensively. You can read the complete article here.

A series of arguments are presented, both for and against cameras. In our opinion, each and every one of the "for arguments" can be readily and safely dismissed based on our own investigations alone.

This page is our response to the article and should be sufficient to clarify most of the points.

Main pro camera arguments
  • The article says: "Overall figures show that speed cameras work." But no overall figures are offered to support the statement. It's argument by assertion, and it's just plain wrong.
  • The claim that accidents dropped by a third at speed camera sites is repeated. Until the regression to the mean error is properly evaluated we cannot know if the reduction or any part of it was due to the cameras or not. We have carried out a vast amount of work on this problem. You can read the arguments and find links to our extensive evidence from these two Safe Speed press releases. (PR126 and PR127)
  • It is said that: "Four other studies show that speed cameras cut accidents dramatically. Two of these allow for long term trends." After a decade of speed cameras, this is scant evidence indeed. The fact that the majority of studies do not allow for long term trends speaks volumes about the motivations of the authors - it simply isn't science if long term trends (especially the regression to the mean effect) are not allowed for. We are left with two apparently worthy studies - and we know of a third - that do seem to have sound scientific methodology and also found a benefit "at speed camera sites". But Safe Speed asserts that the real danger from speed cameras affects the entire road network - not just camera sites. The danger results from alterations in drivers' priorities and thought processes that are caused by cameras and camera enforcement. The following pages carry a description of the argument: (tiger) and (effects). When Professor Mervyn Stone acted as independent adjudicator and examined Safe Speed's arguments he wrote on this very point: "But no-one can say that these localised savings may not be outweighed by an irritation-induced increase on the 99% plus of the road network that is well away from any safety camera." (Stone Report) This is why it is extremely important to examine overall figures, and not just rely of any apparent benefits "at speed camera sites".
  • A TRL representative seeks to assert that "speed is a large and significant cause of accidents". We don't like his arguments and we have written to him. See below.
  • And then there's the graph. And what a graph it is...
The graph in the Which? article:

A fair and accurate graph:

A key argument in the Which? article is that deaths are rising on roads where speed cameras are used infrequently. A graph is offered in support of the theory. But the information is all wrong. Note these points:

  • It is blatantly unscientific to say “Speed cameras are mostly used on urban roads and rural A roads”. Mostly? What percentage is that then?
  • It is true that deaths have fallen more on the urban roads and rural A roads as claimed in the article, although the difference isn't great. Assuming and implying that the difference is due to speed cameras is a leap too far. No evidence is offered to support the assertion. 
  • The Which? graph is dishonest. With suppressed zeros and with a very carefully selected period a very misleading picture is displayed. Extend the data by a few years, and present a clear “Y scale” and the truth emerges. Note that the earlier beneficial trend - highly visible over the period 1991 to 1994 on the blue curve has been lost in later years.
  • The graph is labelled: “Deaths on UK Roads” but actually plots fatal accidents on UK roads – some of which involved multiple fatalities. The graph of fatal casualties is slightly different. We have graphed road deaths not road accidents.
  • But worst of all is the failure to consider traffic volume changes. The following percentages changes apply over the period 1993 to 2003:
Road group Traffic Volume change, 1993 to 2003
Rural A roads +18.2%
All Urban Roads +6.4%
Motorways +48.9%
Minor rural roads +43.9%
  • When one properly consider the effects of the traffic volume changes the following alterations in the risk values apply:
Road group Risk change (per bvkm), 1993 to 2003
Rural A roads -19.2%
All Urban Roads -24.7%
Motorways -27.5%
Minor rural roads -19.3%
  • Motorways show clearly as the best performing roads – this is the exact opposite of the effect claimed in the Which? article.
  • Minor rural roads and Rural A roads have very similar performance - if there are more cameras on Rural A roads, then they are not making a difference.
  • Much of the fall in urban areas may be due to reduced pedestrian activity.
  • The poorest performing roads are the road types most likely to be strongly affected by a reduction in average driver quality - exactly the sort of side effect that we expect from too many cameras.
  • The previous decade 1983 to 1993 showed an improvement in risk of 51% across all roads.

But there's more:

It is noteworthy that there has been a continuing fall in pedestrian fatalities over the last 10 years. Much or all of this fall appears to be associated with reduced pedestrian activity. Obviously if there are fewer pedestrians about, we should expect fewer to be killed in accidents. We don't believe that accurate figures are available, but the growth of the school run over the last decade is widely recognised, and is a well known example of "reduced pedestrian activity". 

These reductions in pedestrian fatalities are mainly concentrated in urban areas. – for example, in 2003 74% of pedestrian fatalities were in built up areas. (1998: 76.2%).

Refer to the table below to see how the patterns of road deaths are changing: 

road deaths pedestrian deaths all road deaths other road deaths pedestrian percentage
1993 1,241 3,814 2,573 32.5%
2003 774 3,508 2,734 22.1%
change -37.6% -8.0% +6.3% -

So is the reduction in pedestrian deaths a triumph for cameras? No, of course it isn't and more importantly, it cannot be. The following logic applies:

  • We know that there has been a reduction in pedestrian activity, which should be expected to lead to a reduction in deaths.
  • Doubtless some will propose that camera enforced reduced traffic speeds have made pedestrian impacts more survivable. And it sounds reasonable – but, and there's a very big “but”, any benefit of reduced traffic speeds must also improve the survival chances for vehicle occupants. Clearly that isn't happening, because more vehicle occupants are dying now than they were in 1993. The imagined benefit of reduced impact speeds cannot single out pedestrians - there is no credible mechanism. 
The reduction in pedestrian accidents alone accounts for the better overall road safety performance in urban areas. Over the decade 1993 to 2003, annual pedestrian fatalities fell by 467. Urban fatalities fell by 342 over the same period. 
new PACTS claims:

PACTS said: "The sharp fall in road deaths up to 1993 are associated with the 1983 law making front seatbelt wearing mandatory, better car design and major reductions in drink driving offences." Equally, PACTS says, increased road use, the sharp rise in motorcycle casualties and the use of mobile phones while driving are among the many reasons that the sharp fall hasn't continued."

Let's take those six claims one at a time:

  • In 1983 front seat belt wearing jumped from under 15% to over 90% overnight. Yet there's not even a ripple in the fatality stats for 1983/84. If front seat belts overall had the expected overall effect we'd see a clear step in the graph. (graphs)
  • Better car design is progressing FASTER since 1993 than it was in the 1980s.
  • The drop in drink drive fatalities hasn't continued because we've forgotten that roads policing is important. Anyway drink drive deaths were 540 in 1993 and 560 (official provisional estimate) in 2003. Policy against drink drive is proved ineffective, and it's policy we're complaining about.
  • Road use is increasing MORE SLOWLY now than it was in the 1980s.
  • Motorcycle casualties have increased by less than 200 per annum since 1993. We're looking for 1,200 fatalities per annum, and anyway some of those extra motorcycle casualties are a result of anti-nannying rebellion. In other words they are caused by policy.
  • When RoSPA trawled Coroner's Court reports looking for mobile phone fatalities they found 19 over 12 years. We're looking for 1,200 fatalities per annum. 
So clearly, PACTS is clutching at fake straws. Not one of the six claims stands up.
Some inaccurate statements...
  • Safe speed does not claim that the main purpose of cameras is to fleece motorists. We do believe that the intention is to save lives. The great shame is that it simply does not work.
  • Our claim about a relationship between the roads fatality rate loss of trend and speed camera convictions certainly does not depend on incorrect assumptions about the number of cameras around in the 1990’s. The correlation between speed camera convictions and “fatality gap” deaths is very accurate. Here's the graph we've been publishing since 2002:
  It appears on our "fatality" page.
  • The article claims that a DfT report earlier this year “reported that excessive speed contributed to less than a third of accidents”. The actual figure was 12.5%, and was the 7th most frequently recorded contributory factor.
Letter to TRL
13th October 2004
Professor Rod Kimber
TRL Limited
Old Wokingham Road
RG45 6AU 

Dear Professor Kimber,

Which? Magazine

I refer to quotations attributed to you in an article entitled “A Fine Idea” in Which? Magazine,  October 2004. I have important questions, and would very much appreciate a detailed answer to each numbered question.

You are quoted as claiming the TRL323 results cannot be trusted because the initial police report cannot properly assess the cause. “In nearly all cases the Police only have enough information to make a brief, subjective record.” But the DfT are in the process of rolling out a huge national scheme based on TRL323 methodology. Have you informed them that they are about to embark on an enormous waste of money? Or perhaps the truth is that the data the Police gather at the scene is extremely useful?

Question 1: Is the data gathered by Police officers and recorded in TRL323 style systems useful or not?

The quotes continue: “Although the police report excessive speed in less than a third of accidents, speed is obviously key in many other accidents put down to factors such as aggressive driving, following too close or failing to stop.

The TRL has made this assertion before, and I refer to the September 2002 newsletter article “Speed and accidents, lets put the record straight.” by Marie Taylor. (link)

The assertion is neither fact nor science. It does not illuminate the debate. It does not refer to speed in excess of a speed limit, and I am quite certain that you are fully aware that aggressive driving and following too close are accident causes that frequently occur irrespective of excessive speed.

Question 2: What percentage of road crashes do you believe are caused or contributed to by speed in excess of a speed limit?

Question 3: Exactly what research and what data do you base your answer to question 2 on?

Question 4: Within the excessive speed crash percentage in your answer to question 2, what proportion are caused or contributed to by “normal responsible motorists” simply exceeding a speed limit (or conversely, what proportion are caused by drivers who are drunk, unlicenced, reckless, driving stolen cars or otherwise far outside “normal responsible behaviour”)?

The article also contains: “TRL also points to a vast amount of evidence demonstrating the strong link between vehicle speeds and road accidents. One TRL study linked accident statistics with hundreds of thousands of speed observations on almost 300 roads. The higher the average speed for a given road type, the more accidents there are.

Clearly this refers to TRL421 and its stablemate TRL511.

Question 5: Both TRL 421 and TRL511 make a giant leap of assumed causality for the claimed speed accident relationship in their headline conclusions. What proportion of the claimed speed accident relationship do you believe to be causal?

Question 6. Exactly what research do you base your answer to Question 5 on? 

Finally you are quoted as saying: “I draw a distinction between coverage and credibility. I can't take seriously the arguments put forward by the ABD and Safe Speed unless they step forward into the scientific domain and produce data or arguments that are subject to scientific analysis. Everyone else does – I don't see why they don't.

Question 7. Our work is published via the Internet for all to examine. What exactly would you like us to do to make it “available to scientific analysis”?

I may have further questions, depending on how we get on with these.

I look forward to hearing from you in due course.

Yours sincerely

Paul Smith

Which Magazine
Colin Meek

We will publish any reply to this page.

Summary points

Traffic has increased. But so has vehicle safety and post crash medical care. Best estimates put the improvements in vehicle safety as being worth 4% per annum and post crash medical care as being worth about 1% per annum. (both in terms of fatality reduction). The increase in traffic is running at around 1.5% per annum. Given these figures, we should be looking for an annual net improvement of around 3.5% per annum without considering the effect of policy changes or road engineering changes. 3.5% per annum compounds up to 41% over a decade. So without policy or road engineering interventions we believe that roads fatalities should have followed the pattern of earlier decades and dropped by at least 40%. Dropping by just 8% represents a policy failure of epic proportions. As for the benefit of speed cameras – well – if they worked to save lives we should have seen fatalities drop by rather more than 40% over the last decade. 

It is vitally important to look at national total figures when considering the overall effect of a road safety policy. If we look at any narrow group we risk failing to notice that casualties have been “moved around” by the policy. Any effective policy should be expected to reduce overall road deaths.

Links and further reading:

Safe Speed encourages comments, further information and participation from our visitors. See our (forums).

Read about our comments policy (here).

Many pages (including this one) have a specific associated forum topic. You can (view) or (add comment) to the forum topic for this page. Posting in the forum requires simple registration.

Calling for real road safety, based on truth

We have a strict editorial policy regarding factual content. If any fact anywhere on this web site can be shown to be incorrect we promise to remove it or correct it as soon as possible.
Copyright © SafeSpeed 2004
Created 13/10/2004. Last update 18/10/2004
Safe Speed navigation:
front page forums join Safe Speed press / media email
main page site guide Paypal donate contact comments
See our new user's 'home page'

Note new address and telephone