Speed camera report sparks row
Expect 'bad science' says campaigner
Has the Gatso had its chips?
According to Safe Speed, last year's annual report was widely criticised for failing to account for 'regression to the mean effect' (RTTM) -- follow the first two links below for explanations. This year's report was expected in June, but has been greatly delayed. Safe Speed said it believed the delay resulted from attempts to account for RTTM effect -- and during the delay period, the DfT declared a moratorium on approving new speed camera sites.
Safe Speed said it "has no confidence that the new report will be accurate -- after all, look at the track record. Last year's report wildly exaggerated the benefits due to neglect of RTTM effect." According to the group, for the report to be meaningful and accurate, it must fully account for all the following effects and side effects:
- Regression to the mean effects at speed camera sites
- Reduction in traffic at speed camera sites
- Benefits of other engineering treatments at or near speed camera sites
- Correction for long term trends
- Confidence interval of results (are they 'statistically significant'?)
- Are site sizes realistic to isolate camera effects?
Side effects of speed cameras and speed camera policy including:
- driver attention diverted from road ahead towards speedo, speed limits and cameras.
- damage to the police / public relationship
- loss of confidence in official road safety messages
- reduced driver responsibility for speed choice
- more effective policies neglected or replaced
- the cash distorts local objectives
- traffic displaced to less safe routes as drivers seek to avoid cameras
- automated enforcement encourages some to operate outside the law (more improperly registered vehicles for example)
- drivers feel 'under pressure' and don't perform as safely
- heavy load on courts
- promotion of the illusion that driving within the speed limit will ensure safety (selected from a list of about 30 side effects)
Safe Speed said: "They will probably call the new report: 'independent' but it's far from independent because it will have been paid for by the DfT. Until the various side effects are fully accounted for, claims of benefits 'at speed camera sites' are worthless. For example we might apparently save 100 lives at camera sites but 'irritant' side effects may cost 1,000 lives elsewhere. Safe Speed believes that road deaths would have been reduced to about 2,000 per years by now if it wasn't for bad policy founded on speed cameras."
New scamera partnership handbook
The second document is the 2006/7 handbook for camera partnerships. Press reports suggest that there will be substantial changes in the way funds will be used, with the Sunday Times newspaper reporting that the cap on new camera sites will remain -- see related story link below. Reports also suggest that speed camera cash will be used for wider road safety purposes such as junction improvements.
Safe Speed said: "It is highly dangerous to tie necessary road safety improvements to speed camera cash. One item to watch out for is if the 'inverted' rule for speed camera placement remains. Up until now speed cameras have had to be placed where many vehicles are speeding. While this sounds superficially attractive, it has actually prevented speed cameras from being deployed in places where it is dangerous to exceed the speed limit (see link on rules below).
"No wonder we have all been complaining that speed cameras are in the wrong places. They are all in locations where it is normally 'safe to speed'."
Comment
What's clear is that the Government's moratorium on new cameras shows that it recognises the strength of the anti-camera campaigners -- and that it may even have got the whole policy wrong. But more importantly, it knows that it has lost the confidence of British motorists.
Last year's report claimed a 40 per cent reduction in 'Killed and Seriously Injured' (KSI) at camera sites. This year's report will claim a much lower figures, for which Safe Speed claimed credit by pulling the DfT up on last year's bad science.
Paul Smith, founder of the Safe Speed road safety campaign said: "The truth is that speed cameras make road safety worse. We will continue to chip away at their bad science until we dig down to the truth. Next year's report will show an even smaller 'benefit'. Anyway, if they said '40 per cent' last year, and this year they say much less, why should we believe a single word they say? They are clearly far too worried about a) face saving and b) looking after the dangerous and flawed camera partnership quangos. This policy must be stopped. Lives depend on it."
Smith added: "The speed camera programme has been an unmitigated disaster. The supporting science has been bunk, the foundation claims have been bunk, and the road safety results have been disastrous disappointing. The Department for Transport has failed in its duty to provide effective road safety policies, preferring to fiddle and fudge in a misguided attempt to convince the public that its fatally flawed road safety policies are effective. The only way to restore road safety values and public confidence is for DfT to come clean, admit their mistakes, and scrap the entire speed camera programme."
Links
Does anyone outside of the dft get access to the raw data to assess its authenticity and subsequent interpretation?
purpleheadedcerb said:
Didn't our illustrious leader state recently that government performance statistics were going to be taken away from the ONS and given to a non civil service, non-partisan company to assess and produce?
Nope, I think the ONS is going to be moved outside of core government in the same way that Ordnance Survey was.
The official reason being greater autonomy. The reality is it allows them to be centrally funded but to sell their data and make a 'profit'.
Expect a number of serious DPA breaches imo. They'll apologise, someone may even 'resign' (i.e. be promoted sideways), but your data will already be in the hands of others.
cymtriks said:
Copied from the report on the link above.
"after allowing for the long-term trend, but without allowing for selection effects (such as regression-to-mean) "
Eh?
Surely the RTTM is about allowing for long term effects? What are they saying here?
They are saying: "We don't have the guts to admit that we were wrong and speed cameras are about 5 times LESS EFFECTIVE that we thought."
Safe Speed issued the following PR at 10:42 this morning:
PR266: Truth finally emerging about hopelessly ineffective speed cameras
news: for immediate release
In the report of the 4th year of the speed camera hypothecation scheme,
published today by the Department for Transport (DfT) we finally get an
estimate for THE major error in speed camera claims.
The error is a statistical effect called 'regression to the mean' (RTM or
RTTM). Appendix H contains: "Thus RTM accounts for about three fifths of the
observed reduction in FSCs (Fatal and serious collisions) with the effects of
the cameras and trend each accounting for a fifth."
This is a clear and unequivocal admission that the benefits of speed cameras
have been wildly exaggerated. Last year's claim of 40% reduction in killed and
seriously injured at speed camera sites becomes 8%.
Paul Smith, founder of the Safe Speed road safety campaign
(www.safespeed.org.uk) said: "With the truth about speed camera
ineffectiveness finally emerging it is staggering that the DfT has not found
the courage to pull the plug."
"Speed cameras are 21st century snake oil. They have made our roads far more
dangerous by focussing everyone on the wrong safety factor. They must be
scrapped immediately before more people die of 'bad policy'.
<ends>
safespeed said:
Last year's claim of 40% reduction in killed and
seriously injured at speed camera sites becomes 8%.
And the drop in Personal Injury Collisions becomes 16%.
Accepting your figures means cameras saved at least 25 lives.
Thus the statement that this report is 'clear proof' that cameras save lives is correct.
This report has generated a totally positive outcome for the motorist. "...from 2007/08 money from speeding fines will no longer go simply on more speed cameras. It will instead fund other road safety measures and better warning signs." This is what many people on PH have been campaigning for.
spnracing said:
Last year's claim of 40% reduction in killed and
seriously injured at speed camera sites becomes 8%.
And the drop in Personal Injury Collisions becomes 16%.
Accepting your figures means cameras saved at least 25 lives.
Thus the statement that this report is 'clear proof' that cameras save lives is correct.
Let's see. 6,000 cameras. 25 lives. A camera saves a life once in 240 years. In Essex, according to accounts, each camera costs in total about £100,000 a year. So saving a life might cost £24 million. I wonder how many lives we could save with £24 million spent on real roads policing? Or cancer research for that matter.
Except cameras and camera policy comes with side effects as far as the eye can see. Just think how much we could damage the police public relationship in 240 years of speed cameras. A list of side effects was provided with yesterday's briefing document:
http://groups.yahoo.com/group/SafeSpeedPR/message/11
cuneus said:
Please explain how you came to that conclusion (given that nowhere in the report are Killed separated from Seriously injured)
Yes they are.
DfT Report said:
At camera sites, there was also a reduction of over 100 fatalities per annum (32% fewer). There were 1,745 fewer people killed or seriously injured and 4,230 fewer personal injury collisions per annum in 2004. There was an association between reductions in speed and reductions in PICs.
and if 32% of X =100 then X = 312 so 8% = 25.
But we have a contradiction:
P4 "Both casualties and deaths were down – after allowing for the long-term
trend, but without allowing for selection effects (such as regression-to-mean)
there was a 22% reduction in personal injury collisions (PICs) at sites after
cameras were introduced. Overall 42% fewer people were killed or seriously
injured. At camera sites, there was also a reduction of over 100 fatalities per
annum (32% fewer)"
P46 "No adjustment was applied to account for long-term trend as the number
of killed did not drop substantially in this study period."
>> Edited by cuneus on Thursday 15th December 18:02
However, as the number of fines increases, people's disposable income reduces. The net result of all these fines may even be visible in the decrease in economic growth of the country and a reduction in the real tax that goes to pay for the beloved "health and education". Fewer fines would mean higher tax take, and therefore an additional increase in health spending.
Therefore, as SS says, this money could have been much better spent in life-saving than it has. We'd be better off overall with a few more deaths on the road and a lot fewer deaths from cancer, HIV, flu, etc. Or would pro-camera campaigners prefer to die slowly from cancer than quickly in an RTA?
They know that the figures are going to level out or get worse based on the creiteria they apply. So, they say, we won't put anymore up. Two years down the line, the figures show increase in KSI's, deaths etc so they then say that obviously they need to return to what they were doing as the new policy hasn't worked!
Call me cynical but I'm sure that's the direction they are going in.
SCott
rbryant said:
I listened to BBC news tonight and heard them trot out the headline 40% reduction in KSIs at camera sites. No RTM effect noted, and any lay observer would have listened and formed the view that a positive effect was proven. Go to it Paul, that Appendix H needs publisizing
I couldn't agree more. I hadn't read your post before I issued the following PR at 7:56 this morning:
PR268: Camera report defrauds media and public
news: for immediate release
The Department for Transport (DfT) "4th year" report issued yesterday is
worded in such a way to have widely misled media and public alike about the
true effects of speed cameras.
The important headline figure of '42% reduction in killed and seriously
injured' contain a gross statistical bias - yet has already been very widely
quoted as if it were the true benefit of speed camera operation.
The gross statistical bias is called 'regression to the mean' (RTTM or RTM).
It is a large effect. Although the report attempts to suggest that it cannot
be accurately estimated on available data, two self similar estimates are
included. (Table 4.9 and appendix H)
According to table 4.9:
0.36 RTTM + 0.11 benefit = 0.47 total. Applying this to the 42% headline
conclusion suggests:
42% reduction in KSI at speed camera sites is comprised of:
32% RTTM benefit illusion and
10% camera benefit
According to appendix H
* three fifths RTTM
* one fifth trend
* one fifth benefit
Since the trend has already been calculated in the 42% headline figure, this
becomes:
* three quarters RTTM
* one quarter benefit
This would suggest:
42% reduction in KSI at speed camera sites is comprised of:
31.5% RTTM benefit illusion and
10.5% camera benefit
Paul Smith, founder of the Safe Speed road safety campaign
(www.safespeed.org.uk) said: "Regression to the mean effect is acknowledged as
providing a gross distortion of the benefits claimed. It is no surprise to me
that the report's authors make many excuses for failing to provide a
substantial and accurate estimate of the effect because clearly it will reduce
the benefit claimed by about a factor of four. The benefit of cameras
expressed in KSI isn't 42% at all - it's about 10%."
"This effect means that if we had installed garden gnomes at the roadside
instead of speed cameras they would have been about three quarters as
effective. We could have had headlines stating that 'Gnomes cut fatal and
serious crashes by 32%'"
"I would be proud to be associated with a road safety initiative that cut
fatal and serious crashes by even 10%. However speed cameras and supporting
policy have many negative side effects on wider road safety which
comprehensively swamp the benefits. The net effect of speed cameras has been
to make our roads much more dangerous."
"I am disgusted that the executive summary of the report fails to make this
adequately clear. Newspapers and broadcasters have presented the 42% figure as
if it were truly the benefit of speed cameras. Clearly it is not and the
public is being misled."
RTTM explained:
Regression to the mean arises because we like to apply a safety treatment - in
this case a speed camera - to the worst places. This means we tend to place
cameras in places where crashes are peaking. After the camera has been
installed the 'peak' conditions pass and the crash rate at the site tends to
return to its long term average.
It should be obvious that many of these peaks arise through random chance and
random peaks always pass.
<ends>
Notes for editors
=================
I urge you to bring this vital information to the attention of the public. I
urge you to consider how it has been be possible for this issue to have been
largely glossed over. I urge you to consider who is responsible, and what
their motivation has been.
My congratulations to the few have already identified and reported the problem.
RTTM primer:
www.safespeed.org.uk/rttm.html
Yesterday's DfT report:
www.dft.gov.uk/stellent/groups/dft_rdsafety/documents/downloadable/dft_rdsafety_610816.pdf
The Times recognises the problem:
www.timesonline.co.uk/article/0,,2-1934085,00.html
(We don't agree with the figures the Times used.)
rbryant said:
I listened to BBC news tonight and heard them trot out the headline 40% reduction in KSIs at camera sites. No RTM effect noted, and any lay observer would have listened and formed the view that a positive effect was proven. Go to it Paul, that Appendix H needs publisizing
I heard it too and felt like putting my foot through the screen.
How the hell can the BBC not know by now that the figures are BS?
It is surely now clear that the BBC are the mouthpiece of the government and camera businesses and the thus happy to regurgitate whatever information is given to them by these crooked organisations.
james_j said:
I listened to BBC news tonight and heard them trot out the headline 40% reduction in KSIs at camera sites. No RTM effect noted, and any lay observer would have listened and formed the view that a positive effect was proven. Go to it Paul, that Appendix H needs publisizing
I heard it too and felt like putting my foot through the screen.
How the hell can the BBC not know by now that the figures are BS?
It is surely now clear that the BBC are the mouthpiece of the government and camera businesses and the thus happy to regurgitate whatever information is given to them by these crooked organisations.
But of course Liebchen.... they are. Ever wondered why you hear something "criticising Drony" on one bulletin und thereafter "all go quiet in snug" as ist never reported again.
Und I read very interesting piece in "Le Figaro" (back in late October - I think?) about the Evin law relating to advertising booze in France. (Apparently they are after tightening up advertising booze on telly und lowering drink drive limit ..in France) They reckon that targetting drink driving ist real reason for 50% drop in their KSI figures. They are not placing all success at "radars" und piece was a bit negative (bof
about their contribution to French road safty as well. Und at least one of the "Sacre Bleu- Merde ZUt Alors camp" say booze advertising should not be about drinking more - but drinking with a better attitude towards drink und spelling out dangers of driving after one large "ballon" of "honeysuckle-like chablis!" But it seem French are bit more honest over their stats....
Gassing Station | Speed, Plod & the Law | Top of Page | What's New | My Stuff