I’ve just spotted that Transport for London’s new Draft Cycle Safety Action Plan attempts to pull the same trick that Norman Baker and Mike Penning tried to pull back in 2012.
That is, it makes a comparison between cycle safety in London and Amsterdam (along with other cities) on the basis of deaths per head of population, rather than deaths per total distance travelled by bike (or by total time spent travelling by bike).
Here’s the graph in question, from page 10 of the Plan –
Followed by the helpful explanation –
Internationally, in terms of cyclist fatalities per million population (Figure 2), London had fewer cyclist fatalities in 2012 than many other cities such as Amsterdam and New York.
So, looking at this graph, you might think that London (in yellow) is fantastically safe! Just look how much lower the number of fatalities there are, compared to Amsterdam, per capita. London had just 1.7 cycling fatalities in 2012 per million population, where Amsterdam had 6.5 – nearly four times higher.
But of course this is an entirely misleading comparison. It doesn’t take into account the fact that, across London, cycling only accounts for around 2% of all trips made, whereas in Amsterdam cycling accounts for nearly 40% of all trips made. There is much, much more cycling in Amsterdam per capita, so comparing cycling fatalities purely on a per capita basis is absurd. It’s like concluding it’s much safer to cycle in London than in Amsterdam if you have a Dutch name, because many more people with Dutch names are killed cycling in Amsterdam than in London.
This is the same logic that led Mike Penning to argue
I think the Netherlands may want to come and see us, to see how we are making sure that so few people are killed cycling
And (more recently) Denis McShane to suggest
@patmcfaddenmp @KenPenton Cycle deaths much higher in France than UK and truly awful in Netherlands
— Denis MacShane (@DenisMacShane) July 7, 2014
How much of this is down to stupidity or dishonesty is hard to tell. You would certainly think Transport for London and a Transport Under-Secretary (as Penning was, at the time) should know better.
The other thing that’s worth mentioning here – beyond the failure to use an appropriate rate – is that, in Amsterdam, children and the elderly (both more vulnerable groups, for different reasons) ride bikes in large numbers.
24% of all trips made by Dutch over-65s are cycled, while in London 95% of over-65s never cycle. If people that are, in general, more frail – and more likely to suffer death than a younger person in an equivalent incident – aren’t cycling at all, that will have a further skewing effect on casualty figures.
Thanks to the Road Danger Reduction Forum, who spotted this ‘measurement’ issue.
We haven’t got a chance whilst the lunatics are still running the asylum.
“How much of this is down to stupidity or dishonesty is hard to tell.”
With MacShane, we can be fairly sure it’s stupidity.
Jim, no I think you’ll find it’s dishonesty. He has been jailed for it after all.
The over-60 group adds disproportionately to the Dutch statistics. They make up about 60% of all Dutch cycling deaths but that says more about the health of elderly people than it does about road safety. If you compensate for age discrepancy I think you’ll end up with cycling being another factor two safer in the Netherlands for the age group that does most of the UK’s cycling, so I think it deserves more than just a footnote.
In addition to that, most of the over-60 group dies while riding a bike, not because they were riding a bike, and most accidents for this group do not involve a second vehicle. If your grandma dies while walking in the park, would you call it a ‘walking accident’?
Dishonesty, I’d say, for Baker, Penning, and TfL. This is the same crap that Robert Davis talks about in his book – stamp out walking and cycling then call it a “success” for “road safety” when people aren’t killed walking or cycling even though the danger has actually increased.
How would the number of motorist dealths per million population compare between the two cities?
Don’t know about cities, but from
“The fatality rate for car drivers or passengers is .. lower in Britain than in the Netherlands (in 2009, 1.6 per billion km compared to 2.1).”
Sorry, that’s rates per distance travelled.
I worked this out from some 2010 data:
In 2010 about 400 MV related deaths for NL population of 16 million
In 2010 about 1300 MV related deaths for UK population of 62 million
(ped and cyclist deaths not included)
So, a bit higher in the UK (note that the figures are all approximates).
“I think Australia may want to come and see us, to see how we are making sure that so few people are killed by sharks”… TfL give yourselves a pat on the back!
Great comment, Neil. Captures the stupidity of the graph perfectly. And it made me laugh out load.
Just spotted my typo, meant “out loud”
Can this kind of nonsense not be officially withdrawn by the Office of National Statistics?
If another government body is misusing statistical data I’m sure they have this power. If the misuse is deliberate then they can and I’m sure have publicly censured those involved.
I would be interested to hear what the likes of Oliver Carsten of ITS in Leeds thinks about these uses of rates. He once gave a lecture all about the use of rates, the significance of exposure to risk (i.e. through increased amount of travel) and azbsolute figures.
And by similar logic, if nobody rides a bike, then there will be no cycling fatalities at all. A perfect cycling safety record!
That being said, for many years deaths per 100,000 population was also a common metric for traffic fatalities. We still have a long way to go before we have accurate measures of cycling (or pedestrian) distance travelled, and even then, is that the proper metric to use when comparing ‘relative safety’ across modes (ie x fatalities per 100,000 km travelled)? Is there even any point in making such comparisons?
Thanks for the acknowledgement Mark . Unfortunately the response deadline for the consultation to the CSAP draft is now gone, otherwise you could have flooded TfL with righteous indignation.
Ambrose White: I did a longer piece on the problem of using the official metric, and put forward a better alternatives here: http://rdrf.org.uk/2013/11/15/if-we-want-safer-roads-for-cycling-we-have-to-change-how-we-measure-road-safety/ in Local Transport Today taking into account and adding to Oliver Carsten’s points along the lines you mention.
This area is not just for numbers nerds. It shows how official “road safety” institutionally discriminates against having more people outside cars on the roads, particularly if they are elderly. It also asks questions about us stating what it is we actually want, and why.
Is there not data available then for deaths per distance traveled? As you say, this would be much more interesting statistic, and good to put a number on it rather than using cycle use as a proxy.
The best estimates I have seen are here.
Thanks, but “…are here” lights up on mouseover, but there’s no link with it. Can you post the URL? Thanks! 🙂
Apologies! Should be working now.
That’s brilliant, thanks!
I take the general point but if you’re going to complain about other people’s abuse of statistics then your own need to be squeaky clean. “24% of all trips made by Dutch over-65s are cycled, while in London 95% of over-65s never cycle.” is not comparing like with like. The 24% of all trips could be made by 2% of the over 65s for all we know.
That’s a fair point; it’s not a direct comparison, and I didn’t really intend it to be. Apologies if it came across that way.
On average, Dutch over-65s make around one trip by bike per day. Now of course – like the statistic that 24% of all trips by Dutch over-65s are cycled – this could be accounted for by a minority of elderly people cycling a great deal. That is very unlikely however.
I do know that only 1% of all trips made by British over-65s are cycled.
I’ll try to dig out statistics on the regularity with which Dutch elderly people cycle.
Click to access document000135.pdf
especially p10 for at least a partial answer to this question. (e.g. 94% of respondents cycle regularly (= at least once a week); 80% of over 75s never cycle! 20% do!!)
From David Spiegelhalter (Prof of understanding risk at Cambridge, writing in the journal “Significance”)
“In a recent European comparison
the risk per distance cycled in the UK,
22.4 deaths per million kilometres, was found to
be far higher than in the Netherlands, where the
figure is 12.4.”
If you want a typical current example of this kind of “road safety” drivel, look at http://www.transport-network.co.uk/Best-and-worst-councils-for-improving-road-safety/10735#.U9lexxF0ypo .
Nothing about what kind of road user is reported as being killed or seriously injured (“Serious Injury” itself depending on levels of reporting rates for the injured), nothing about if they hurt or killed themselves or were hurt or killed by others, nothing about the casualty rate per journey/time/distance travelled, and of course the assumption that “the Council” is responsible for the casualties in the first place.
You also need to take into account the cause of death, many of the deaths recorded on Dutch cycle paths are people over the age of 70 dying of heart attack or strokes (basically old age).