Previous posts have described the record of Transport for London and the Greater London Authority under Mayors Livingstone and Johnson with regard to cycling. Whatever the verdict on this record is, there is one two-wheeler group that has done well in London since 2000 – motorcyclists. Motorcyclists have profited from virtually unhindered access to supposedly cycle-specific facilties such as Advanced Stop Lines and cycle gaps in road closures. Press attention is drawn to pedestrians killed in collisons with cyclists, but not the larger number in incidents involving motorcyclists. While cycling is persistently portrayed as hazardous, motorcycling – with far higher casualty rates – is not.
TfL’s pro-motorcycling agenda is shown up well in the saga of allowing motorcyclists into bus lanes. While the details may tend to bore all but the most hardened transport professional, this episode tells us a lot about how some road user groups can get their way, irrespective of the evidence supposedly required to justify legal changes. Time and again we can see in the history “road safety” how a safety benefit is consumed as a performance benefit. In this case it is even dubious whether any safety benefit for the measure taken has ever existed: we simply move to the performance benefit (of motorcyclists having extra road space) while using “road safety” as a justification.
Below Colin McKenzie summarises the latest stage in this story:
Transport for London recently published their report on the motorcycles in bus lanes trial, together with a press release to say that they will continue the trial, but with extra education and enforcement. What did the report say? Why the need for the education and enforcement?
First, a little explanation on statistics. Statistics are more reliable with bigger samples. The smaller the sample, the bigger a difference has to be to be real rather than random chance. The problem with trials of motorcycles in bus lanes is that up to now sample sizes for collisions have been too small for confident conclusions to be withdrawn. Statisticians do some arithmetic to work out the percentage chance that a difference is real rather than chance. A difference is generally accepted as real if there is a 95% or greater probability that it is. This is called a statistically significant difference.
This trial is the biggest UK trial yet of motorcycles in bus lanes, and produced two crucial statistically significant differences between the test and control sites. Control sites are similar bus lanes where motorcycles remained prohibited when they were allowed at the test sites. This minimises the effect of unknown factors on the results.
The report is mostly factual, though it does speculate from time to time. The press release puts a definite spin on the report’s findings to try to justify the decision to continue the trial.
The main points are:
There was a statistically significant increase in motorcycle collision rates at the test sites – far too big to be explained by increased numbers of motorcycles. At control sites there was a decrease in motorcycle collisions, which again cannot be explained by reduced numbers. Overall, the chance of the difference between the test and control sites being real is over 99%. The report, in attempting to explain this away, goes as far as to say that if there had been 4 more collisions at the control sites, the difference would not have been statistically significant. This is like saying there were no motorcyclist fatalities, but if there had been one there would have been some. The results are what they are.
There were reductions in motorcycle journey times, and increases in motorcycle speeds, with the average across all sites exceeding 30mph. This suggests motorcyclists are breaking the speed limit more. These figures are statistically significant too, because the sample size is big.
There was a good increase in cycling at both test and control sites of 10-15% between before and after surveys.
At test sites, there was a tripling of cycle collisions, big enough to be statistically significant even though numbers were low. At control sites, the increase in collisions was about the same as the increase in cycle traffic. Adjusting the tripling of cycle collisions at the test sites for the increase in cycling gives an increase in collisions per cyclist of 173% (versus a slight – insignificant – decrease at control sites). Numbers are low, but this difference is so big that it is statistically significant – the report estimates a 98% probablikty that it is real.
The increase in cycle collisions at test sites was not due to more collisions with motorcycles. This is hardly surprising – there was only one collision between a cycle and a motorcycle in the whole trial! But because of this lack of cycle/motorcycle collisions, the report claims that the increase in cycle collisions is not due to the trial. They suggest no other explanation.
This is simply unacceptable. The whole design of the experiment was to exclude other factors from the trial. Either the increase in cyclist collisions was due to allowing motorbikes into bus lanes, or the trial design is broken and all the results should be thrown away. They cannot have it both ways.
I can think of numerous ways in which the presence of motorcycles in bus lanes could increase cycle collisions. The most obvious is that cyclists are riding closer to the kerb (because of motorcycles passing fast and close), making them less visible to other road users.
At this point the press release departs from the facts completely, saying: “There was a fall in the cyclist collision rate across trial bus lanes and the control lanes, with a smaller decrease in the trial lanes”. The only way this statement can be construed as truthful is if it only refers to collisions between cycles and motorcycles. It completely ignores the overall increase in cycle collisions at trial sites, which alone should have been enough to get the trial stopped.
No statistically significant effect.
The research uses a definition of conflict which I fundamentally disagree with. Essentially it’s putting yourself in a road position where someone might have to brake to avoid you, with severity assessed according to how much braking actually happens. This means that overtaking a traffic queue safely on the outside is recorded as a conflict, even if nothing is coming the other way, because if there were traffic the other way it might have to brake.
To me, the most significant conflict in this trial is overtaking too fast and/or too close – there’s no point braking, but the cyclist is frightened and/or annoyed. This is not counted as a conflict in the report – they did not even count such incidents.
The report clearly shows that allowing motorcycles into bus lanes is increasing collision rates for both cyclists and motorcyclists, and is encouraging motorcyclists to break the law more. On this basis, the only defensible decision is to stop the trial immediately, to minimise further casualties.
Instead, the Mayor has chosen to gloss over the casualty increases and continue the trial. The proposed education and enforcement is an obvious attempt to reduce casualties to a more acceptable level.
This trial proves that allowing motorcycles into bus lanes increases road danger, both for motorcyclists and cyclists. It should be stopped forthwith and no further trials to be conducted. By continuing the trial, TfL is failing in its duty of care to the public. Some Councils in London have opposed having motorcycles in bus lanes on their roads, with others going along with the trial: the evidence points to them stopping motorcycles being allowed in bus lanes.
Colin McKenzie, June 2010