Road Safety News

Average speed cameras cut KSI collisions by a third: RAC Foundation

Tuesday 4th October 2016

The use of average speed cameras has, on average, cut the number of collisions resulting in death or serious injury by more than a third, according to new research.

Produced for the RAC Foundation by Road Safety Analysis (RSA), the study found that, having allowed for natural variation and overall trends, the number of fatal and serious collisions decreases by 36% after average speed cameras are introduced. In addition, the number of collisions resulting in injuries of all severities is cut by 16%.

The findings will be presented today (4 October) at the TISPOL Road Safety Conference by Richard Owen, operations director at RSA. Mr Owen will also present the findings as part of the Fringe at the 2016 National Road Safety Conference in November.

Richard Owen said: “Measuring the influence of speed cameras in isolation from other road safety improvements over time has previously never been undertaken on this scale.

“The statistical results clearly show good collision reductions on the stretches of road where average speed cameras are used; often covering much longer distances than other enforcement systems.

“The findings and methodology used should be of significant interest to those considering the use of this technology, as well as those wishing to evaluate their own road safety schemes.”

At the end of 2015, there were at least 50 stretches of road in Great Britain, with a total length of 255 miles (410 km), permanently covered by average speed cameras. These stretches range in length from under half a mile in Nottingham to 99 miles (159 km) on the A9 between Dunblane and Inverness in Scotland. Many of these stretches of road are broken down into subsections (79 in total) monitored by several sets of cameras.

The report says that one reason for the increase in usage has been the reduction in the installation costs of permanent average speed cameras, which is now typically around £100,000 per mile, compared with around £1.5m per mile in the early 2000s.

Steve Gooding, director of the RAC Foundation, said: “All the indications are that compliance with average speed cameras is high; now this research reveals the sizeable impact they can have in reducing death and serious injuries.

“As the cost of technology continues to fall, more and more authorities are considering whether to install average speed cameras and so it will be important to ensure that casualty and compliance data is openly available so we can continue to assess and understand the road safety benefits they deliver.”

© Copyright Robert Struthers and licensed for reuse under this Creative Commons Licence.


Comment on this story
Report a reader comment

What's your view - comment on this story:

I confirm that I have read and accept the moderation policy and house rules relating to comments posted on this website.
Your comment:
Your name and location:
Your email:

Nick, this recent report from the UK Statistics Authority on Scottish Government speed camera statistics explains my point better than I can, and with examples:
Charles, England

Agree (7) | Disagree (1)

To back up your claim in the third para of your post below, perhaps you could provide a few links to evaluation reports which, in your view, do as you suggest? Thanks.
Nick Rawlings, editor, Road Safety News

Agree (6) | Disagree (1)

Andrea, will the 'evaluation' of your scheme include comparisons with identical, but unchanged, similar other "control" routes closeby to help rule out the inevitable variations due to a variety of other factors (weather, legislation, police activity, vehicle engineering, driver influences, etc.)?

Will the site of the pilot scheme be unchanged and remain unchanged for the duration of the trial other than for the addition of the new cameras (no new road markings, no new signage, etc.) to rule out the inevitable effects of these changes, or will identical changes (other than the addition of new cameras) be made to the control routes?

Or will this evaluation be like many others that have preceded it and naively conclude that the effect on KSIs of the new cameras = ( number of KSIs after - number of KSIs before ) / number of KSIs before * 100%, and ignore all the other variables which have helped change the casualty numbers over time?
Charles, England

Agree (7) | Disagree (4)

Birmingham City Council, Solihull MBC and West Midlands Police are undertaking a pilot of Average Speed Enforcement (ASE) replacing the redundant GATSO cameras at selected routes. The RAC research is positive and we will be undertaking an initial 21-month evaluation phase to assess the effectiveness of the cameras, equipment and the overall system. The initial pilot at these routes will do much to help inform the wider efforts to make the West Midlands region’s roads as safe as they can possibly be.
Andrea Johnson, Birmingham City Council

Agree (3) | Disagree (9)

Hugh - Fatal & Serious Collisions did indeed reduce by over a third once all other factors were taken into consideration. If you only looked at the straight before and after results they would be much higher. There are still lots of other questions still to be answered about how collisions are reduced, what cost-benefits look like and how other types of technology compare. I could keep going for years on the subject but I feel the most important question has been answered.

Dave - Yes, you were the first person who came up with the idea of excluding collision data from a 'site selection period' to account for RTM. This was then turned into a workable model by Professor Allsop and then further revised by ourselves in this study. Your previous work could be repeated using this more accurate methodology but you would need to use a much longer period of data to ensure your results were statistically significant. You would also need to obtain information about the actual site selection periods instead of guessing.

I will give you a shout-out at the RSGB conference when I deliver the presentation!
Richard Owen - Banbury

Agree (8) | Disagree (7)

It is fantastic to see Richard Owen and other experts using the method that I proposed such that we now agree on perhaps more than we disagree. Rather than explaining how to use the FTP method accurately, let's instead continue to close the gap still further and agree to settle the matter for good by running RCT scientific trials.

Let's work towards running the trials that can bring all parties together to agree not just on the methods, but on the final results as well. That would be a day well worth all of us working towards.
Dave Finney, Slough

Agree (10) | Disagree (3)

Isn't the heading a bit misleading therefore? I interpreted it that KSIs had been reduced by a third....due to the presence of the cameras hence my first sceptical comment, however Mr Owen says "The results show that indeed, a significant proportion of reductions reported at the sites were not attributable to the presence of the cameras". If so, perhaps it's a more realistic and reasonable conclusion and as I said before, the true benefits of speed limit enforcemnt cannot be accurately measured, but in the meantime I'm happy with reasonable presumptions as to their likely benefit.
Hugh Jones,Cheshire

Agree (2) | Disagree (10)

The best way to identify the site selection period is to ask the people who carried out the analysis what period they used. This was achieved in the study. In some cases where the period was not identified directly it was calculated using the DfT operational case rules present at the time.

Guessing what the period might have been by subjectively trawling through data until you find a pattern that fits your own preconceptions is not a satisfactory technique. Neither is it acceptable to apply a FTP method to sites that were not selected due to a specific collision analysis.

The generalised linear model used in the report allows for overall improvements in road safety achieved through engineering, education and enforcement and produces a result that estimates the effect of the cameras themselves over a significant period of time. The results show that indeed, a significant proportion of reductions reported at the sites were not attributable to the presence of the cameras. It also produces a figure that estimates the ‘site selection effect’, which is considerable.

Given all of this I would have thought that Mr Finney would have welcomed the methodology with open arms and not picked an irrelevant point to argue the entire report was flawed.

As with all RACF publications, the report was peer reviewed by eminent road safety statisticians and those with a working knowledge of the systems.
Richard Owen - Banbury

Agree (10) | Disagree (6)

This RAC report contains some good news, but does not achieve the accuracy that we need when lives are at stake.

I developed the FTP method so that the effect of site-selection (RTM) could be completely excluded. In order to do this, however, the SSP (Site Selection Period) must be identified from the data and I proved that this was possible to do in two reports.

The RAC report recognizes the importance of the SSP. The SSP was estimated for some of the average speed camera sites, the SSP was given by the installers for some of the other sites, and some sites had no evaluation of the SSP. Unfortunately, though, the report does not perform the crucial step of identifying the SSP from the data. This means that we cannot be confident that the RAC analysis has eliminated RTM from it's results.

Also, many average speed camera sites had other interventions installed as well, and many of those sought to relieve the underlying causes of high crash rates (such as closing the central reservation gaps, changing speed limits, early warning messaging signs, removal of fixed speed cameras etc). I could not find any analysis of the effect of those other interventions in the RAC report.

Overall, therefore, the RAC report fails to prove their assertion that average speed cameras have caused reductions in serious crash rates.

The RAC report has taken important steps forward with a better understanding of RTM, and how to remove it's effect, but the report actually acknowledges that speed cameras are "a contentious issue" that have "long been a subject of significant debate and analysis". Why are we still debating and analyzing? We could have ended all of that years ago by simply by running RCT scientific trials. Let's learn from past mistakes and start running scientific trials.
Dave Finney, Slough

Agree (12) | Disagree (8)

It's a sad state of affairs if, as Pat is suggesting, that some drivers have so little confidence in their abilities to drive properly, that they have to choose alternative and longer routes, just because they are afraid of getting caught - if that's what he meant. How does he know that actually happens?
Hugh Jones, Cheshire

Agree (2) | Disagree (12)

I may have missed it in the report, but I couldn't see any 'before and after' speeds at the sites. Are the authors assuming camera installation and reduced casualties as an irrefutable link, or have they linked camera installations to slower overall speeds which has then led to reduced collisions? Could the general climate of enforcement and monitoring on these roads induce drivers to concentrate more anyway and up their game a bit? Feedback from the police for example on whether they perceived an overall improvement in driving on these stretches would have been useful.

Having said that, if at the end of the day the end result is better driving all round and therefore fewer collisions, perhaps it doesn't really matter. Perhaps signs advising of speed cameras only (without the cameras) would have had the same effect - in the short term at least.

Hope that doesn't sound too negative - I'm all for speed enforcement - but I think the long term effects on driver behaviour should not be ignored, the benefits of which can go beyond the original sites themselves, but as they cannot be easily measured, they possibly get overlooked in favour of concentrating too much on the road where the camera was.
Hugh Jones, Cheshire

Agree (2) | Disagree (6)

I have no difficulty in agreeing that average speed cameras will reduce KSI casualties on the routes that the cameras are installed. But to repeat my post on the article on 31 May, ,

When will a piece of work be done on the "displacement effect" i.e. the alternative, often rural, routes that some drivers use to avoid sections of roads that have average speed cameras. About time some pre- and post installation data assessments were done to measure the unintended consequences of such cameras on increasing traffic on some cross-country rural 'rat-run' routes. Or does this report look at that too?
Pat, Wales

Agree (15) | Disagree (3)