Road Safety News
 

Thames Valley cameras have had 'no impact' on collisions

Monday 13th February 2012

A new report by an electronics engineer has concluded that safety cameras in the Thames Valley region have not ‘made any impact in preventing road traffic collisions’.

The report, ‘Does reducing traffic speed using speed cameras reduce the number of collisions?’, investigates the effect that cameras have had on the number of collisions that resulted in death or injury. It has been produced by Dave Finney, an electronics engineer and anti-camera campaigner.

The report covers all 212 fixed camera sites and 105 mobile camera sites that had been operating for two or more years (at the start of 2009) in the Thames Valley region (Oxfordshire, Buckinghamshire and Berkshire).

The report concludes that there was ‘no reduction in the number of collisions at fixed (Gatso) camera sites or at mobile camera sites after they started operating’.

Even after compensating for other influences such as rising traffic volumes and safer vehicle design, the report says that ‘cameras are still not demonstrated to have resulted in any reduction in collision rates’.

Mr Finney’s report goes on to say that ‘collision rates increased at mobile speed camera sites following deployment, relative to all roads in the Thames Valley area’.

He also found no relationship between vehicle speeds and the number of collisions, stating that ‘reductions in speeds at camera sites (both average and above the speed limit) did not result in any reduction in the number of collisions.’

Click here to read the report or contact Dave Finney on 07561 197 543.

Comments

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:

An interesting (worrying) development in Wiltshire: http://transportsurveyspecialists.co.uk/accidents-rise-after-speed-cameras-scrapped/

Russian Roulette?
Susan, Northamptonshire

Agree (1) | Disagree (9)
-8

Forget the cameras, forget the collision rate, it's simple; stay at a safe speed set for the area i.e the speed limit and you can totally ignore the camera. It's not going to go "off", you will not get points or fines, or be asked to attend a course.

The speed limit is there for a reason. Its law to stay within this limit, its against the law to go over. Simple!

Do you steal from shops, no, why? It's against the law. Then do not go above the speed limit and stay within the law. Cameras then can be forgotten about, not removed but be there for surveillance like they are in the shops and on the streets, trying to keep you and your communities safe!

Why do we get into this every year, it's a waste of time.

Keep within the law.
Anthony, Cardiff

Agree (2) | Disagree (9)
-7

It beggars belief that some are still trying to suggest that safety cameras do not work or worse still that cameras are 'in some way' making drivers less safe? I refer Mr Finney to a recently published RAC report here: http://www.racfoundation.org/research/safety/effectiveness-of-speed-cameras there is overwhelming support for safety cameras across the entire country despite the fact that nobody wants to get a ticket them self. All those not in favour of speed enforcement - raise your right foot!
Susan, Northamptonshire

Agree (2) | Disagree (8)
-6

A report so flawed it barely warrants consideration (but I will continue). It is littered with incorrect assumptions about the way speed cameras were installed and managed and chooses to ignore most of the data available.

If I were to pick just one problem though, it is the concept of the ‘Selection Bias Effect’. A concept virtually irrelevant when applied to the cameras installed in Thames Valley in the 1990s as there was no collision-based selection criteria with cameras installed wherever authorities wanted to place them and no ‘Site Selection Period’. The SSP concept really only applied to cameras installed in the 2000’s during the Hypothecation period. There is then no need to correct for the bias. Indeed, the only bias applied is carried out by Mr Finney who chooses to ignore the long-term data that most strongly supports the case for reductions at camera sites.

Rather than ignore the data, the report published by the Thames Valley Safer Roads Partnership in October 2009 (using the dataset available to Mr Finney) covers all available before and after installation data and comes up with some rather different conclusions.
http://www.saferroads.org/wpcms/wp-content/uploads/camera-site-report09.pdf

Don’t let this so-called independent analysis fool you; this is an anti-camera report concocted by a long-time campaigner against the use of speed cameras who doesn’t deserve the oxygen of publicity.
Richard Owen - Thames Valley

Agree (3) | Disagree (6)
-3

I am one of those people whose opinions Susan of Northampton dismisses without explanation other than reference to the RAC report.

I base my assessment of camera effects on more than 50 years day-by-day involvement with mathematics, starting before my State Scholarship in Pure and Applied Mathematics and Physics, my engineering degree (and then, as it happened, overlapping for two years as a post-graduate research student at the Engineering Laboratory of the University of Cambridge with Richard Allsop, author of the RAC Report at the Mathematics Department) and 30 years running my own electronic engineering company. My views are not influenced by any financial interests whatever in speed cameras, other than the costs I incur. And, incidentally, 1m+ miles without hurting anyone, much of it in the much more dangerous 1950s and 1960s in cars far less safe than modern ones, and much of it still in pre-war and 1960s cars without ABS, air bags, stability systems or in some cases seat belts.

I would be interested to hear how Susan's background compares.
Idris Francis Petersfield

Agree (0) | Disagree (2)
-2

Again for John F and also for Dave Finney, this time on randomised control trials.

Sorry, this is nonsense - both in terms of carrying out such trials and because it has already been done on a very larege scale - by default!

Of course in basic principle it would seem to be a good idea to select one group of roads, of varying characteristics and another group of the same, and then install cameras on one group but not on the other and then compare what happens.

The many reasons this is reality nonsense include:

1/ What local or national politician, road safety officer or police chief is going to stand up and say to the control group “We think your roads are so dangerous that you should have speed cameras, but we are going to put them elsewhere to see how much worse your accidents are than theirs.” Volunteers? I thought not.

2/ Anyone who has studied road accidents for half an hour knows that one year is far too short a period for variations to be significant - harsh or mild winters, wet or dry summers, the economy, road improvements, RTTM etc all mean that it is impossible to separate out camera effect from the chaff of the rest.

3/ Who is going to decide which roads are sufficiently comparable to be compared with and without a camera – and on what basis? If his selection is wrong, what are the results worth?

4/ This is even more true for a relatively small area with its low accident numbers, inevitably highly volatile as they are.

5/ In summary, there is not and never will be any prospect of obtaining meaningful data from any such experiment that could realistically be set up - see also
http://conservativehome.blogs.com/localgovernment/2011/06/fact-and-fiction-in-measuring-whether-speed-cameras-save-lives.html

6/ But the good news is that however politically and economically impossible such an overtly selective scheme undoubtedly would be, because (as someone mentioned here) there can never remotely be a camera at every site which might (theoretically) need one, hundreds of thousands of accident sites have never had a camera anywhere near and can therefore tell us what we need to know about what typically happens without cameras without needing any more data - it already exists:

7/ I realised a few years ago when thinking about this problem and looking at data that the DfT have in their databases records at least from 1985 for every recorded injury accident details of not only when but also WHERE it happened, to within 10m x 10m areas as determined by the 5+5 digit Grid References. I therefore signed up to obtain the data, then for 4.7m accidents from 1991 to 2007 and extracted only what I needed.

8/ Using only the first 3 grid reference digits, I was able to build a database of every 1km sq area in the country – equivalent to a camera site – which had suffered even one injury accident in that whole period.

9/ From that I was able to extract a database containing only those sites – 131,303 of them - which had suffered at least one 3-year period of 4 KSI, and then a further databases showing the percentage falls in the 3 year “after” periods. Very few of those sites ever had cameras installed, and most of those that were, from 2000 onwards, but so few that they could not materially affect the results.

10/ These percentage falls in K, SI, KSI, Slight and All are of course averages for each area selected by police code number ranging from the whole of Britain, London (a special case due to congestion and volume), Britain excluding London and then every other police area from congested South to almost empty Scottish or North Wales areas.

11/ These percentage falls happen largely as a result of RTTM but also to an extent due to long term trend, itself due to a wide variety of factors such as improving roads, vehicles, tyres, systems and (slowing and recently falling) traffic growth – but with no meaningful camera effect.

12/ It is obvious that another accident is more likely to happen soon at that same spot on a busy rather than on a quiet road – ie RTTM will be low on busy roads but high on quiet ones. And indeed it is.

It is also obvious that there being far fewer fatal than serious, and serious than slight, accidents, the chances of another serious accident in the same place will be much lower than another slight accident. And that turns out to be true too.

13/ My Excel spreadsheet shows that falls (due to RTTM and Trend combined) range from 10% in London to 20% in other major centres to 80% in remote areas. This obviously cannot be due to geographical location or different driving skill in those different areas, but must be due to the different “mix” of busy and quiet roads in those areas.

14/ It is therefore nonsense for academics to estimate RTTM as an approximate figure - whether 75% (KSI) in Appendix H or 25% (KSI) in the RAC report – RTTM clearly varies enormously from busy roads to quiet roads so any assessment that fails to take into account the characteristics of the roads in question is a waste of time and effort. Indeed, worse than that, because defective analysis is unlikely to lead to the best solutions.

16/ The work has been much delayed by unrelated problems but I have since re-done the first part of the process, for 6m+ accidents, now that data from 1985 is available. At Dave Finney’s suggestion but also with Professor Allsop’s support, the new analysis will adjust the raw data for national trend so that what remains should be the best possible measure of the effects of RTTM alone.

The final stage, comparing the figures with those of camera sites should be instructive.
Idris Francis Petersfield

Agree (0) | Disagree (2)
-2

Re the report on Wiltshire:

3 months data comparing one year with the next is not enough data over a long enough period to be statistically meaningful. That said, deaths fell from 8 to 6.

More importantly, the report fails to specify what happened at former camera sites, as opposed to overall - as Dave has pointed out it is universally agreed that cameras have no benefit outside their site area, usually about 1km square.

If I wanted to exceed the speed limit I would be inhibited more by knowing that police patrols, that can be anywhere, had increased as had fines, than encouraged by knowing that fixed cameras had been switched off.
Idris Francis Petersfield

Agree (0) | Disagree (2)
-2

re wider area benefits of cameras:

In the early days camera advocates used slogans such as "making speeding as socially unacceptable as drink driving" - and that implies camera benefit everywhere not on 1% or 2% of our roads.

The 1st nationwide report did mention the hoped-for possibility of benefits outside the immediate areas of cameras but as Dave wrote, that disappeared for good from all subsequent reports as they realised that bad overall trends - fatalities in particular, being the only truly reliable data - made any such claim impossible.

I confirm that as Dave wrote, the 4th year report makes no mention whatever of benefits outside camera sites, and given its unrealistic claims of benefits within sites it is difficult not to believe that even if there had been only slight glimmer of benefit outside sites they would not have seized on it.

A simple question for John if I may - would you please explain how and why official claims for accident reductions due to camera substantially exceed the involvement of speeding as causal factors in accidents? For example - how can cameras (which far from eliminate all speeding) lead to the reductions of 40/50/60% attributed to them, when only some 10% of KSI accidents involve speeding - even as a minor let alone the primary triggering factor?

I note that Susan of Northampton seems not to have noticed my request that she disclose any relevant involvement with speed cameras.
Idris Francis Petersfield

Agree (0) | Disagree (2)
-2

John F spoils a potentially constructive contribution and impartiality with his final sentence: "If he can determine which community should be DENIED a camera when another gets one".

It's as if speed cameras have magical powers. They do not. As I have posted here previously, there is more evidence of crashes contributing to cameras than there is evidence (or credible argument) as to how they could prevent them. Remember - they are cameras, on poles or tripods, which distract and cause sudden braking (hazardous behaviours).

Phrases such as "deny a community a camera" have been used for years to avoid explicit randomised controlled trials because the powers that be know that such trials would expose the ineffectiveness of speed cameras (and, I believe, show that they have a net negative effect on collisions amd casualties.

Dave Finney is independent of all vested interests, and so am I. My sole interest is in road safety for all road users.

I would like to see other contributors declare their interests in this matter.
Eric Bridgstock, Independent Road Safety Research, St Albans

Agree (1) | Disagree (2)
-1

In response to John F:

Without wishing to turn this into a battle of CV's - and while still hoping that Susan or someone else will tell us whether she has any direct interest in the continuation of speed cameras - it seems to me that owner-managers of small specialist companies whose capital and careers are at risk to a single bad decision tend to be more scrupulous in their arithmetic than are executives of large multi-product companies that can and do survive such mistakes, not infrequently even after paying rewards for failure to those responsible. No such rewards are available to the small businessman.

More importantly, another difference between us, Mr. F, is that I doubt that you have spent the ten thousand hours or so that I have spent over the past 12 years evaluating data, trends, policies and results. Or that you have several gigabytes of information immediately to hand on disc, much of it in my own head or, increasingly, on a web site such as my www.fightbackwithfacts.com.

To be more specific, if you are "bemused" by criticism of RTTM having been ignored, that can only because you are not aware, as I and others are that:

1/ The original 2 year experiment of 1999-2001, devised by PA Consulting and UCL Professors, completely ignored the effects not only of RTTM but also of another significant factor, traffic volume.

2/ The graph "Eight Area Trial Scam K " (see www.fightbackwithfacts.com/casualty-data/) shows a 7.6% overall rise in fatalities in 1999, the year before the trial, so the 10% fall in the first year of the trial was little different from what RTTM might have predicted even if nothing had been done.

3/ Worse, in the second trial year fatalities rose by 4% and by 2003 peaked 10% higher than the pre-trial figure - no different from 1993 and 1995, 10 and 12 years earlier - in stark contrast to the 1968 to 1993 national trend that fell by 3% pa compound (7% pa compound relative to traffic volume)

4/ Naturally "The pilot was scheduled to last for two years, but the evidence of speed and casualty reduction after one year was so compelling that the programme board decided to introduce the system to other areas" (source section 1.4, 4th Report) and the worsening later trend in the trial areas was conveniently forgotten.

5/ It is some time since I read the first 3 national reports, but I am assured that not one of them even mentions RRTM or adjusts its fanciful claims for that effect - this despite the reductions claimed - only "at our sites" of course - clearly being substantially greater than involve (let alone are caused primarily by) speeding in the first place. Does no one understand these days even such simple statements as "You cannot get a quart out of a pint pot"? Or litres of course, to be PC.

6/ One of the main critics of the reports' failure to mention, let alone seek to quantify, RTTM was the late Paul Smith of Safespeed, who told me that when he was campaigning on that point, prior to the 4th Report, a retired Professor of Statistics told him of receiving a phone call from one of the authors of those reports, urgently asking him whether RTTM should have been considered!

7/ Due to such complaints, not least by Professor Linda Mountain of Liverpool University who had then recently published her own analysis that - like Dave Finney's - concluded that cameras provided no benefit after allowing for RTTM, long term trend and (from memory traffic diversion) the 4th report not only mentioned RTTM but included in Appendix H Dr. Mountain's analysis of it, estimating that of the observed falls in KSI at the 200 sites for which relevant data had been available, trend accounted for 20%, RTTM for 60% and camera affect only 20% - 1/4 of the benefit previously claimed. However Appendix H also pointed out that with no data available it was not possible to detemine how much further the supposed 20% camera benefit should be reduced to adjust for those drivers - perhaps the faster and/or more dangerous ones - who divert to avoid cameras and take their share of accidents with them.

7/ Like many others who follow these numbers, I consider it utterly unacceptable and indeed shameful that the authors of the 4th Report then chose to make their headline claims - since circulated as Gospel - that cameras prevented 1,645 serious injuries and saved 100 lives a year "after allowing for the trend but without allowing for selection effects (such as regression-to-mean)"

8/ Why would PA Consulting and senior academics DO that – i.e. publish as their headline claims numbers that an Appendix (all be it buried away) in the same report shows to be seriously overstated? Perhaps because the stark contrast between the figures, adjusted down for RTTM and what they had been claiming for years would raise eyebrows?

9/ And how could they also write that “Whilst regression-to-mean does appear to account for some of the reduction at casualty sites, the safety effect remains substantial” When in terms of KSI, RTTM reduces apparent camera benefit by 75%? In what sense of the word does a 75% reduction equate to “some”? And how “substantial” is a safety effect that (in terms of KSI, much the most important) is 75% less than previously claimed?

10/ It is worth noting that in 2006 I wrote repeatedly to the authors of the report to challenge them on these points, but the only reply I received was from one minor contributor several years later when he told me in an email that they had all agreed to ignore me.

11/ There is of course another simple yet vital reason for not admitting to the much reduced supposed benefit – their claim of 2.7 to 1 benefit to cost ratio. Start cutting that down by 75% for KSI and 25% for Slight and it soon reaches negative territory – a game not worth the candle.

12/ But as always, there is more and worse – the DfT valuations of fatalities and casualties avoided turn out to be wholly fanciful and seriously overstated. My article about this was published some months ago on Conservative Home, and as soon as I find it again I will put it on http://www.fightbackwithfacts.com/bogus-dft-values/ Essentially the point is that one third or so of the supposed cost of a fatal road accident is the “lost output” of that casualty for the rest of his working life. This is utter nonsense – no output is lost because (a) others step forward to ensure that as always supply meets demand and (b) to the extent that the death of that individual reduces his supply, it also reduces his demand with no significant net effect. For non-fatal casualties demand continues of course but his place in the supply chain is filled by others so there is no loss of output.

13/ A similar proportion of the valuation is taken up by an entirely theoretical figure for pain and suffering – but whatever use that figure might be in determining policy, it appears in any known ledger this side of the Pearly Gates and must not be treated as it if did.

14/ For both these reasons the true cost to the State in cash terms is a small fraction of the DfT’s figures, and so (even without allowing for RTTM) the benefit/cost ratio of cameras is not remotely 2.7 to 1 but well below 1/1.
Idris Francis Petersfield

Agree (1) | Disagree (2)
-1

This in response to Anthony, who suggests that we all obey the law because it is the law.

He is perhaps unaware that under our Constitution – written and unwritten – it is not only the right but also the duty of everyone to challenge bad law – and that is what camera critics are (increasingly) doing.

I suggest that those of us who are convinced by our analysis that cameras cause more problems than they solve, with (even on the basis of wholly unrealistic claims of accident reductions) costs that exceed benefits and more lives and limbs lost than saved, we have no alternative but to argue our case.

Especially when, surely for the first time in 180 years of policing, we now have the utterly unacceptable position of police officers having a direct financial interest in imposing as many penalties as possible.

Perhaps Anthony is also unaware of the Principle of Jury Nullfication, by which a jury is allowed to acquit in defiance of the law and the evidence if they believe the law to be wrong. One of the most famous such cases was of William Penn who later founded Pennsylvania, repeatedly acquitted in London by juries who defied the prosecution and the judge because they believed the law prohibiting preaching Quakerism was wrong.

If it were not for brave and principled men like them and many others who defied bad law and provided the last ditch defence of the common man against authoritarian governments this would have been for centuries the most free country in the world – though sadly it no longer is, as we know to our cost, not least because too few are prepared to stand up for fundamental principles.
Idris Francis Petersfield

Agree (1) | Disagree (2)
-1

Having read the report and studied the charts, the conclusions seem to be based on some assumptions which may not really be valid and do not really seem to tie up with the data presented. Taken at face value fig 7.1 for example shows mobile camera sites are remarkably effective based on the trend data for their introduction and collision rates.

It is disingenuous to suggest the report is 'independent' when it is compiled and supported by anti camera campaigners, who despite wishing to be seen as weighty academics, have not satisfactorily demonstrated they have considered all potential explanations equally objectively.

I was bemused to see the work of a number of well respected statisticians dismissed as ignoring selection bias – having listened to robust debates between these individuals they have worked very hard to provide a sensible and considered view on this complex topic. It is not appropriate to rule out analysis purely on the basis it used complicated formulae, especially when the subject under discussion requires a multivariate analysis of the topic. The analysis seems to be superficial and subjective in many areas.

I would not consider experience in non–aligned fields to immediately qualify anybody to have an expert opinion on a topic, having run a 100M Euro automotive electronics programme in the past – it does not necessarily make you a good researcher.

I would like to see this report properly reviewed by peers to validate whether professional analysts would reach the same conclusions, personally the only area I am in agreement with Mr Finney is that a randomised controlled trial would be helpful. If he can determine which community should be denied a camera when another gets one, then raise the funding required to do this and recruit some truly independent and objective expertise then I will take some notice.
John F Hertfordshire

Agree (2) | Disagree (2)
0

I don’t have the expert analysis that Eric, Dave, and Idris have, but I do have evidence and common sense. Look at this page if you want to understand some dangers of speed cameras: http://www.dorsetspeed.org.uk/news/neg.aspx.

How on earth can a few fixed yellow boxes and the occasional predictable stripy van seriously get drivers on side to obey all speed limits everywhere, and even if they did, how could this possibly result in the reductions claimed? What about the comfort that they give to boy racers and thugs who can’t believe their luck that the authorities have been so daft to use their precious resources in such a way, and that all they have to do is drive at a certain speed for a few yards in a very few locations, other than that it’s a free for all. And because most areas think they can reduce proper policing because of cameras, it’s plainly obvious that the overall effect is likely to be zero or negative.

I think it is possible that speed cameras COULD improve safety, but most certainly not in the way that they have become to be used.
Ian Belchamber

Agree (3) | Disagree (3)
0

Some drivers in the years shown in the data believed there was “a camera around every corner” and drove accordingly. Here the contribution of anti camera campaigners to road safety is probably underestimated, the publicity they initiated with the media may well have increased compliance levels with speed limits – well done.

The Thames Valley Partnership was one of the initial partnerships and the combined effect of other partnerships being established may well also have had a cumulative effect on driver behaviour, combined with the way that some drivers moderate their behaviour as they get points on their licence - this effect is unlikely to be a site specific one, so an area, or even nationwide effect which does not follow camera introduction rates locally is quite credible – predicting the phase lag however would be extremely difficult.

These factors would necessarily explain the data trends but failing to consider their effects undermines the credibility of the analysis in the report. Road safety research is complex - not everybody who has an opinion is necessarily making a useful contribution.

The article launching your report describes you as an anti camera campaigner, your research methods appear to consistent with that group so you have some work to do convince anyone you are entirely objective in your assessment of the data.

The Royal Statistical Society debate on the evidence on cameras and in particular RTM considered that although there was a variety of views on the size of the RTM effect, they concluded that on balance cameras were effective, to the best of my knowledge they did not trawl data that suited their arguments and publish it as ‘research’ which I have to say does seem to be the approach of anti camera lobbyists.

To be a disappointment to Mr Bridgstock is a burden I will have to live with, however he has fails to understand that although he disapproves of cameras, many communities if told they had an equal case for a camera intervention to another would regard not getting one, because of an experiment, would see that as being “denied” – not a personal view merely a reflection of the thinking of many communities, rightly or wrongly.

For an evaluation to be meaningful to establish control sites, the individual sites would have to be very similar (this is exceptionally rare in practice), hence the comments on adjudication despite the cynicism about this if it was easy it probably would have been done.

During the periods shown in the charts traffic growth across the UK was on average 2% pa, now who is going to tell me whether the impact of increased congestion and potentially reduced speed will offset the increased collision exposure rate and whether that counters or increases the effectiveness of cameras?

If you are serious about peer review you have to do it properly – perhaps you could start by asking someone like the RSS to advise you on this?
John F Hertfordshire

Agree (2) | Disagree (2)
0

• A new report has concluded that safety cameras in the Thames Valley region have not 'made any impact in preventing road traffic collisions'.

Having studied road casualty data, trends and claims for thousands of hours over some 12 years I find Mr. Finney's analysis entirely credible. I am currently uodating my own analysis of 4.7m injury accidents from 1991 to 2007 to 6m from 1985 to 2007 since the earlier data became available and comparing the 3 years after installation not only with the 3 years before but also with the 3 years before that, to highlight the extent to which selection bias skews the results and gives rise to the largely bogus claims of effectiveness we have seen for many years.

It is in any case clear that these claims have to be bogus because it is beyond rational dispute that speed cameras cannot lead to falls in accident numbers far greater than ever involve speeds above limits in the first place - 5% for slight, 6% for all, 9% for SI, 10% for KSI and 14% for K (approx). Even these figures are inherently overstated however because they include accidents where speeding "might have been" involved as well as those where it probably was. Further and in any case, in many of those accidents, speeding might only have been a minor or indeed irrelevant factor compared to the many other triggering factors. For all these reasons the potential benefits of cameras are relatively trivial at camera sites, (2% of our roads) and utterly trivial in terms of national figures.
Idris Francis Petersfield

Agree (4) | Disagree (2)
+2

Susan

I refer you to the report from your own county
http://www.roadsafetygb.org.uk/news/1995.html

There is evidence showing that speed cameras contribute to more crashes than they could ever prevent.

Finally "overwhelming support" from a population fed on the ill-judged and fatally flawed "speed kills" mantra counts for nothing. Either cameras work or they do not. My research over the last four years, and that of Dave Finney, suggest that they do not.
Eric Bridgstock, Independent Road Safety Research, St Albans

Agree (5) | Disagree (2)
+3

Interestingly John, what you say is what both the authorities AND objectors were saying in the early years of cameras, namely that speed cameras installed around the road network could change the behavour of all motorists everywhere and this had the potential to dramatically change road safety over regions, or even nationally.

The "area-wide" claim then seemed to be dropped by the authorities after it was noticed that road safety trends had turned for the worse when cameras started to be used, I gave the link before. I notice that some authorities suddenly started claiming the "area-wide" effect again after Oxfordshire switched their cameras off.

You are right that my report does not investigate any area-wide effect, just as neither did the largest report into speed cameras in Britain (4YE) nor the RAC report. In fact, would you like to name the reports that do consider the area-wide effect of cameras as these could be useful references?

My report is very simple. TRL research suggests that collision rates should reduce when speeds reduce, 5% for every 1mph. But this did not happen in practice.

We know traffic slowed down as a result of the cameras, but there was no reduction in the number of collisions at camera sites in Thames Valley when this occurred. This is consistent with Police investigations into the contributory factors to accidents and is also consistent with the RTM estimate for KSI in the largest report into speed cameras in Britain, though some way from the PIC estimate, again I gave the link before.

Please don't jump to the conclusion that I am an anti camera campaigner just because the article says so. Also I don't believe that the method I used in this report has ever been used before, not by the authorities, nor by camera supporters NOR by objectors. I was actually careful that neither side even knew my report existed prior to the press release but since then I've had comments from both sides expressing surprise that such a simple, straightforward and accurate method was possible.

The Royal Statistical Society is a good suggestion, I shall email them. Thank you.
Dave Finney - Slough

Agree (7) | Disagree (2)
+5

Is it worrying?

The title is "Accidents rise after speed cameras scrapped" but it could also have been "Deaths fall after speed cameras scrapped". It could also have been "Accidents rise after speed enforcement stepped up".

And the article looks at "area-wide" effects again rather than effects at the camera sites and is also only over 3 months. If cameras do have "area-wide" effects, how can the change in trend for the 1st 12 years of speed cameras be denied?

Doesn't this highlight that it seems easy to select the figures that support a political belief and ignore those that show it to be false? The only solution I can see is to implement scientific trials so that we can all find out. That is what my report recommends.

Can I ask again Susan, how would you explain figure 7.2 (the above graph)?
Dave Finney - Slough

Agree (7) | Disagree (2)
+5

That is the reason I produced my report. Most official reports do not even consider the RTM effect and those that do, eg the RAC Foundation report and the four year evaluation, only estimate its influence.

Rather than estimate the value of RTM, my report takes a new approach of simply excluding most of its effect, thereby aiming to obtain the most accurate evaluation of the effect speed cameras to date.

Putting aside for a minute what we may wish to believe, how would you explain figure 7.2 (the above graph)?
Dave Finney - Slough

Agree (8) | Disagree (2)
+6

Hi John, thank you for your considered comments. In order:

1) Do speed cameras affect collisions over entire areas? If so then you are right that figure 7.1 suggests mobile cameras do reduce collision rates, but it also suggests fixed cameras increase them. Indeed, in the early days it was thought area-wide improvements due to cameras could be demonstrated but this theory was dropped after it was discovered that road safety trends had turned for the worse when cameras started to be used:
http://speedcamerareport.co.uk/05_gb_road_safety.htm

2) No "anti camera campaigner" was involved in compiling this report. The analysis is all my own work and the engineers, officers and proof readers who assisted in the checking and editing are not "anti camera campaigners" so far as I'm aware.

3) I didn't wish to give the impression that I dismissed the official estimates of RTM, quite the opposite. If accurate estimates were showing RTM to be small (say 5 to 10% of fall), then there would be little need to be concerned but that is not the case. The 4YE estimate of RTM means that there is no evidence that cameras have saved any lives, nor prevented any serious injuries, although there may be a reduction in slight injuries:
http://speedcamerareport.co.uk/03_government_reports.htm

4) I agree but figure 7.2 (above) does show the actual collisions that occurred, that's not "superficial and subjective". It is a fact that a significant reduction is seen to occur a year before the cameras, but no reduction at all is seen after camera enforcement starts.

5) There have always been far more sites that meet requirements for cameras than there are cameras available so some communities are already denied a camera. In fact it was those very location choices by the camera operators that led to RTM in the first place, hence my report.

I have investigated peer review but it takes a long time (typically over a year). I'm not even sure that the official reports such as 4YE or RAC report were actually peer reviewed anyway. Have you better information on that?

Can you suggest a trusted and independent expert or organization willing to spend the time and effort to peer review my report? I am willing to spend my time and effort to pursue this.
Dave Finney - Slough

Agree (8) | Disagree (2)
+6

I would like to thank Road Safety GB for running this news story and for allowing a free discussion.

The response to my report so far does seem to demonstrate a very different culture in road safety to that in my area of engineering.

If we received a report challenging some of our engineering safety practices, we would examine the report, contact the author, hold a meeting to review all evidence and invite the author to that. We would make certain we understood why the new report came to different conclusions and would cover all available evidence before commenting publicly or modifying safety practices. Eg, I have contacted the authors of the major road safety reports I have commented on and just took for granted that this is standard practice in safety engineering.

Of the comments I’ve seen in the press, none of the critics of my report have contacted me.

Ps. I contacted the Office for National Statistics but they referred me to the DfT. And the Royal Statistical Society referred me to the Transport Statistics User Group. I am now contacting both the DfT and TSUG. Perhaps this is too much of a political hot potato?
Dave Finney - Slough

Agree (9) | Disagree (2)
+7

I'd like to thank Mr Owen and his team at TVSRP for supplying the database on which my report is based.

If there were no selection bias, collisions should have continued at the rate of around 960 per year right up to the camera sites being installed. But they didn't, and this is what my report investigates.

If the people who installed speed cameras genuinely had not known the collision history when they selected the locations, then there would be no RTM (regression to the mean) when comparing the 3 years before and after. But it is likely that not a single camera site was chosen without that knowledge, whether specific guidelines existed at the time or not. Therefore there was an SSP for every camera site, although the lack of formal guidelines prior to 1999 does make it more difficult to determine what that SSP was for the earlier cameras.

But even though RTM (arising from selection bias) seems very difficult for many to understand, the fact remains that cameras are not performing as we have been led to believe. The theory is that collisions are related to the speed of traffic and we know that speed cameras slow traffic down. But collisions did not reduce when the speeds reduced so what happened in practice did not fit the theory.

I have been told that Mr Owens team checked the data in my report and found no errors so this raises 2 genuine questions:
1) Why did collisions reduce a year before the camera sites were installed?
2) Why was there no reduction in collisions after enforcement started?

Road safety is complex. It seems that the effect of cameras may be so small that it can be swamped by other influences, particularly RTM, but I do not recommend removing all cameras. I don't see myself as an "anti-camera campaigner", but a campaigner for honesty and good engineering practices in road safety.
Dave Finney - Slough

Agree (10) | Disagree (2)
+8