A protest campaign over poor walking and cycling safety
Effective road safety planning: what should be happeningWhat the vision should be
Road safety planning and policing: what we have a right to expect
Poor Merseyside road casualties should be considered
Other relevant information
Options that should be considered
Merseyside road safety failures
Merseyside / national road safety failures
Merseyside road safety concernsSo-Mo project on pedestrian casualties
Merseyside road safety improvements
Taking action on poor road safety
The 2018 Atkins/AECOM/Maher 20mph Research Study was seriously flawed - draft
Two reports from a three-year project were published in November 2018:
- a 68-page Headline Report, with a 13-page Executive summary,
- a 221-page Technical Report.
There are several major concerns about the reports.
- The analysis of road casualty data in the reports is fatally flawed.
- There are also many other deficiencies in the reports.
- The reports also raise the question of the continung unconstitutional refusal of many police forces to enforce 20mph speed limits which has not yet been addressed.
Analysis of road casualty data: fatally flawedReduction of road danger is a major aim of 20mph speed limits, and so a key question about the 11 case studies that the project examined is what was the overall change in casualties following the introduction of 20mph speed limits.
The reports do not answer this directly, but using the figures in the longer report, I have done a rough estimate that there was an overall reduction in casualties of 11%* (95% confidence interval 6%* to 17%*, p < 0.001*). [The 95% confidence interval indicates the uncertainty around the estimate of 11%. It can be thought of as meaning that the true reduction in risk is probably between 6% and 17%, but there is some uncertainty because of random fluctuations in casualty numbers. ]
So the headline result should be "Across all areas, there was a statistically significant reduction in casualties of 11% (95% confidence interval 6% to 17%)".
But the reports did not do this. Instead of pooling the figures from the 11 case studies, which is the recommended approach, they analysed each of the case studies individually, and then totted up how many were statistically significant at a cutoff of p = 5%, which is known to be a flawed approach. They said that the majority were not statistically significant (one was), leading to headlines that "No evidence that 20mph limits cut casualties, says DfT study" in Local Transport Today. This approach is fatally flawed because
- many of the case studies were very small
- it throws away the advantage of spending a lot of money on a major study and collecting data in a way that it can easily be pooled
* These figures are estimated from the published reports. To calculate the exact figures requires access to the original data. this should have been done by the authors of the reports, and hopefully will be done by them or in an independent re-analysis.
- The report does not point out that many police forces are refusing to enforce 20mph speed limits, using a criterion of whether police officers feel 20mph is the appropriate limit. This is unconstitutional, it is an abuse of police power, and it needs to be eradicated - see below.
- The extent of enforcement is crucial, but was not well documented
- the extent of compliance in other countries and the amount of enforcement and engineering measures is hardly mentioned
- the total size of the areas studied was small
- none of the areas studied was in London
- Scope: some essential background is that the key research question is how can we ensure the known benefits of 20mph speed limits can be achieved, but the reports appear to be asking the question of whether 20mph speed limits work
- the starting point seems to be that 30mph limits are the norm - and that we might introduce 20mph "schemes" - whereas it should be that 30mph limits are the wrong limits on many streets, and that the limits should be reduced as soon as possible, and then compliance achieved as quickly as possible by properly funded enforcement and/or engineering.
- There is no mention that a power/sample size calculation was done to ensure that the study was of sufficient size that small changes could be reliably detected, and yet this is usually thought to be critical According to Prof. Maher, a calculation was done (personal communication), and a copy of it is awaited.
- The reports fail to give a literature review of other reported studies, many of which have found reductions in casualties.
The unconstitutional refusal of many police force to enforce 20mph limitsThe reports deal with this poorly
- it is not just one or two rogue officers but a systematic abuse of power: 20mph speed limits are treated differently to other speed limits, and yet it is the most vulnerable road users who are, or should be, protected by 20mph limits. The police are acting like they used to act on domestic violence i.e. treating it as trivial, and like they used to act in failing to protect vulnerable teenage girls from sexual exploitation of teenage girls, which was widely known about for years
So the headlines ought to be "systematic abuse of police power undermining road safety".
Last updated: 29 Nov 2018