Unanswered questions about the DfT’s 20mph study
The debate about 20mph limits seems to be rumbling on in your pages but consultant Atkins and the DfT still haven’t answered the questions posed in my LTT letter of 1 March regarding the use of comparator areas for casualty reductions.
It may be useful to recap on the key issues that remain unclear: the report only looked at eight small case studies of residential roads in its casualty analysis, plus part of Brighton. The size of these areas varied considerably and whilst Winchester (Stanmore) had only four collisions per annum and Walsall (Rushall) just one collision on average for the five years beforehand, the Brighton (phase two) case study had 92 collisions per annum. In fact, on average Brighton comprised 46 per cent of the total dataset of collisions.
Hence, when the results were indexed and aggregated, the outcome was overly influenced by the Brighton figures. This should have sounded a warning bell to the authors that this was not a robust set of data to be used in such an evaluation.
There were other warnings that should have been heeded. In Winchester (Stanmore) the collisions reduced by 15 per cent after implementation, yet the casualties increased by 16 per cent.
In Nottingham (Bestwood) the collisions increased by nine per cent yet the casualties reduced by six per cent.
In Liverpool (area seven) the collisions reduced by seven per cent but the casualties by 20 per cent.
Any objective assessment of such contradictions and differences would lead one to believe that the numbers were just too small and events too random to be credible.
Then there was the sweeping comparison to background trends that were assessed using Stats19 figures on 30mph roads from across whole regions. Whilst we understand that these were looking at areas up to 100 times larger than the case studies, and so far more reliable in terms of size, we are not convinced that these reflected the same conditions as the roads in the case study areas. In fact, the same comparator area was used for both city centre and residential case studies. This is not logical.
With regard to the comparator areas, these were measured over a nine-year period (five years plus one year buffer before, three years afterwards). If there were engineering or other changes to those roads during this time this would have changed casualty rates and would need to be filtered out.
Were A and B roads filtered out of the comparator areas? A and B roads are not similar to 20mph residential roads.
Were 30mph roads that were subject to any engineering changes filtered out? Many could have been as a result of localised casualties and hence data for them would not be indicative of a background trend.
Were 30mph roads that were changed to 20mph filtered out? If not, this would have reduced the 30mph data pool and hence arbitrarily excluded casualties in the after period. Whilst any other case study areas in the comparator areas were excluded, what about other councils setting 20mph limits and radically reducing the 30mph data pool?
Our concern is that it is really too simplistic to compare casualties on a very specific set of roads in small case study areas to casualties across whole regions without considering these factors. We note that the North West comparator area showed a whopping 35 per cent reduction in casualties, which was attributed to road congestion and the closure of police station counters.
This should have been another warning bell to the authors regarding the credibility of what they were comparing. Surely in an age in which national road casualties are flatlining, if there is any reliability in this 35 per cent, then government road safety policy should now focus on increasing congestion and closing police station counters!
And let’s recognise that in the technical report itself the section on collisions and casualties in residential case study areas (10.2) repeats three times that the results are “unclear” on casualties, fatalities and casualty severity.
By all means take note of its findings on the extent of public support for 20mph limits, the importance of enforcement, and the need for national media engagement. However, this report fails to provide any credible analysis of the effect of 20mph limits on casualties in residential roads. The DfT needs to clarify that no inferences should be made from this report regarding casualty reduction.