Tuesday, 12 February 2013

Exaggeration considered harmful. Hiding the real story about cycling.

Click for the original copy
I just received a rather triumphant press release from Cambridge Cycling Campaign. It's a "good news story" about "5000 more bicycles" which I expect will be reported verbatim elsewhere. However as an ex resident of the city it instantly rang bells for me and I think it deserves a little analysis before blindly reproducing the claims.

The central claim is that there has been an increase of 37% of "employed people cycling to work everyday by residents of the City of Cambridge" between 2001 and 2011. There is also a claim that there are "now more people walking and cycling to work than people using buses, cars, taxis, or motorbikes combined", that the worst performing ward of Cambridge is Cherry Hinton with a "Percentage of residents cycling to work" of  23.3% while the best performing is Newnham with 40.6%. They also pointed out that "the rates of bicycle use may be under-representative of the actual bicycle use", suggesting that both the high rate of student cycling and multi-modal commutes both meant that these figures were underestimates. The reader is to assume that the actual cycling rate in the city is somewhat higher than these already high figures, based on nothing but a suggestion.

Cambridge and surroundings by bicycle. Darkest blue are
areas with between 20 and 30% of commutes by bike. The
light brown colour found beyond the boundary of the city
indicates 0-5% commutes by bike. Cambridge is an island
It all sounds amazing. But can be we believe these figures ? I think not. Cambridge certainly has the highest cycling modal share in the UK. That is supported by many published figures and Cambridge stands out as an island on the Guardian's visualisation of the census data. Cambridge almost certainly also has the highest cycling modal share of any city across the English speaking world (I've discussed some of the reasons why). I like the city a lot. We lived there for nearly 20 years in large part because of the cycling, which is somewhat ahead of the rest of the UK. However, this exaggeration is painting a false picture of how much cycling there actually is in the city.

I believe the figures are based on British Government 2011 census stats published a few days ago. These were summarised quite nice with an interactive map overlay on the Guardian website. Cambridge achieved the highest figure for cycling with 18.1% of people saying that cycling is the main way that they commute to work.

It is important to note the source of these figures. They come not from traffic counts or observations of actual behaviour of commuters but from a survey question which asked about "the main way people commute to work". Any such survey is open to Response Bias - a tendency for people to give the answer that they think the questioner wants to hear. Such surveys are also quite open to people's own interpretations of how they think they travel. i.e. "Cyclists" are likely to say that they cycle even if they do so only two days a week on average. There are many causes of error in survey results, and while with care we can compare one survey with another taken at the same place on the same date of a different year, any comparison with different dates, different places or with the result of an analysis of behaviour measured in a different manner is at the very least to make some dangerous assumptions.

The press release combined "walking and cycling" as that made the figures sound more exciting than they actually were by adding the 9.6% of people who self-reported as walking to work to the 18.1% who said that they cycle. Note also that the 3% of people who said they used the train to get to work, 0.2% who said they used a coach and 0.1% who say they use light rail have been excluded from the comparison group of "buses, cars, taxis and motorbikes". That's how it was made possible to claim that cycling and walking combined were "more popular in Cambridge than the internal combustion engine."

Cambridge and surroundings by car. Darkest blue are areas
of between 60 and 75% of commutes by car. A different
scale had to be used compared with that for bikes.

However, we must note that when we look at the raw stats on the Guardian site show that many places all the way across the UK, including just on the edge of Cambridge achieve a figure of between 45% and 60% of commutes by car without excluding anyone. The most extreme have between 60% and 75% by car, with one of these appearing just a few miles from Cambridge, and that's without anyone having to cook the books to make the numbers appear higher. Cycling remains a minority pursuit across the UK.


This press release was not the first example of exaggeration that I'd seen based on the same survey results. In fact, it took only a couple of days after their publication for the cycle of exaggeration to begin. A Cambridge politician excluded some of the groups used in the original report in order to report that "nearly 32% of those who travel to work in Cambridge get there by bicycle". This is presumably the reason for the peculiarly precise language used in the press release ("the number of employed people cycling to work everyday by residents of the City of Cambridge"). The more people we remove from the comparison, the more meaningless it becomes because each time we remove people we remove information. If we were to continue the process of removing groups of people who do not cycle, then we would end up reporting merely that 100% of cyclists ride bikes, which of course is a meaningless statement.

Now I have to note that it's rather self-serving for a politician to say this. Politicians have a long history of claiming to be working for cycling while in reality they spend far too little on cycling infrastructure. This is certainly true within Cambridge.

The press release goes on to say that "Much of this increase can be attributed to the hard work of councillors and officers at Cambridgeshire County Council and Cambridge City Council, for which we thank them." and that  "Cambridge has grown with virtually no additional motorcar traffic because of the money spent on cycling." when in fact, the investment in Cycling within Cambridge has for a very long time been a tiny fraction of that spent to benefit motorists. Indeed, it's not many years ago that the Cambridge Cycling Campaign itself pointed out that the expenditure on "cycleways, footpaths and safe routes to school combined" was just 0.6% of the local transport budget. When I wrote an introduction to an article printed in their newsletter I pointed out that local expenditure amounted to just "£1.45 per person per year".

The long running saga of, and
inadequate solution in Gilbert Road

is another example of Cambridge
Cycling Campaign acquiescence.
Unfortunately, Cambridge Cycling Campaign now seems to have joined with the council in their attempts to exaggerate the achievements of the city. It is a mystery to me why a campaign group which has watched remarkably little progress occur for nearly two decades should have decided to help the very politicians who have done so little for cyclists. To do this provides no basis for future campaigning. If everything's fine already, why even bother with campaigning ?

This is not the first time that false statements have been made about Cambridge. In the 1990s, one of the councils in Cambridge published a leaflet which described Cambridge as "the second best city in the world in which to cycle after Amsterdam" (or words along those lines). That wasn't true then, and this isn't true now.

Why exaggerate ?
Of course, Cambridge is far from the only place to exaggerate its achievements. Many other places do the same. I've pointed out some of Copenhagen's repeated exaggerations but also found Amsterdam exaggerating the truth. Portland made a similar claim to 1990s Cambridge when they said they were "only second to Amsterdam". Popular lists of "top cycling cities" regularly present figures which have no basis in fact, which are collected by sufficiently different means that they cannot be compared with one another, or the simply miss out many Dutch cities and this is often done in order to hype one place over another.

Of course it takes some effort to find good figures. I've found it difficult to get to the bottom even of many figures published here in the Netherlands. Emails which I've sent asking for the source of figures which apparently contradict each other have gone unanswered.

Is anyone actually interested in facts any more ? Is this exaggeration simply a reflection of people liking to be told that where-ever it is that they happen to live is good ? Does anyone want to actually learn from what works elsewhere, or are we all too busy claiming to be number one ?

Exaggeration considered harmful
A 20 km radius around Assen - the distance within which
it is normal for secondary school age students to cycle into
the city to school. Green spots show where school friends
of our children lived and from where they cycled to get to
school. The red line shows the route of a school triathlon
ridden by unaccompanied 12 year olds who had come to
school by bike and who afterwards rode back home.
Nothing out the ordinary in this country, but in the UK ?
The problem is that this exaggeration is harmful. It is harmful because it prevents progress from being made.

If all sets of figures have been "cooked" to some degree then we have absolutely no basis for comparison at all. If figures which are accurate from one place can only be compared with "cooked" figures from another then it makes it very difficult, if not impossible, to work out what works.

When figures are exaggerated, no new information is being created. All that happens is that the truth is swamped beneath piles of meaningless press releases. Hiding the facts does not help cyclists, and cycling campaigners do not help their own constituency by boosting their own cities purely for the sake of an impressive press release.

For cycling to grow we need to look at the best practice from all around the world, not encourage people to accept the low standards of what they already have. For progress, we have to continue to be critical. Even in the world's genuine top cycling city, campaigners are rightly not satisfied with what they are given if it doesn't result in an ever improving standard.

Far too much emphasis on "cycling cities"
The same 20 km radius around Cambridge showing how
even adult commuter cycling drops rapidly to below 5%.
 The red line shows the route that a 14 year old boy in a village
could not take to get to school
. Seemingly no-one thought it
reasonable to suggest that he cycled. I've cycled there myself
in the past and I wouldn't have suggested it either. Whatever
the quoted statistics say, there is an enormous difference in the
ease of cycling between Cambridge and Assen.
What the data from the Guardian website shows most vividly is just how much of an island Cambridge is. There is an amazingly sharp drop off in cycling as you reach the edge of the city. In most of the surrounding countryside the patterns of transport are the same as the rest of the country. i.e. Very low cycle usage, very high car usage.

The idea of "cycling cities" is itself harmful. There is no reason why we should only consider cities, nor why we should expect that people will only cycle in cities, nor why we should be so interested in counting only how much cycling there is in cities, nor how high the figures can be made to look for a subset of cyclists in a small area.

That is not how it works in the Netherlands. We don't have cycling cities, but cities, towns and villages in a cycling country. Cycling doesn't drop off at the boundaries of cities but is normal all the way out into the countryside and through all the villages until you reach the next city, where it's also normal. It's normal for secondary school children to commute relatively long distances through the countryside to reach the nearest city where their school is. It's normal for elderly people to go on rides together just because it's a pleasant thing to do. To make these things possible requires a lot more than boastful exaggeration. It requires reliably consistent quality of cycling infrastructure and investment of enough money to do a decent job of a comprehensive cycling network which goes everywhere - this ensuring a sufficient degree of subjective safety that cycling is accessible to everyone.

Now I could finish by quoting some spectacularly high figures for cycling in this area, but I won't. If I did, would you believe them ? I hope you would not without taking a close look first. Suffice to say that there is considerably more cycling here by a far wider demographic than exists anywhere in the UK, including in Cambridge. That's covered by the usual unspectacular figure quoted for the whole of the Netherlands: 27% of journeys in the whole country are made by bike, that's for all the people and all their journeys for all reasons.

Notes:

The picture above of Cambridge as an island may seem to fly in the face of frequently made claims that Cambridgeshire as a whole has a higher percentage of trips by bike than other counties. However, note that a quarter of the population of Cambridgeshire lives in Cambridge. The bubble in the city therefore makes the whole county appear to have a higher modal share for cycling than is actually the case outside the city of Cambridge itself.

It should also be pointed out that the claim made in the press release that cycling has risen by 37% between 2001 and 2011 is itself dubious and difficult to support based on previous claims from the Cambridge Cycling Campaign. The current figure from the census is 18.1%. However, Cambridge Cycling Campaign has claimed higher rates many times in the past. For example in an article on the 2001 census published in 2003, "In Cambridge in 2001, 25.91% of workers between the ages of 16 and 74 travelled to work by bicycle" (a remarkably "precise" figure !). Another page says that "According to the 2001 census 28% of all commuting trips within Cambridge are by bike". The 28% figure was also attributed to the RAC in 2005, while a 2007 article claimed just 25% "according to 2001 census data". None of these previously claimed figures for the rate of cycling to work in Cambridge is low enough to support the claim that there has been a 37% increase. Cycling campaign groups have often made attempts to show to cycling is "booming" in the UK, but the consistently fly in the face of the evidence


Do British people get the politicians that
they deserve ? Excerpt from "Watching
the English" by Kate Fox.
I don't want to single out the un-named Cambridge politician involved in cooking the figures above as he's no better or worse than the rest of them and I found him quite agreeable in person. However, I will point out that for several years I made a point of inviting all of them, including him, to accompany me on Study Tours in the Netherlands so that they could see for themselves how it is that Cambridge falls short of the standards of the Netherlands. Unfortunately, while many of them replied to let me know that they were "interested", every one of them sadly turned out not to have a gap in their busy agendas on any of the dates that I'd planned. Unfortunately, it also turned out that not one of them could make it on a date of their own choosing or even at my expense. Politicial commitment to cycling clearly only goes so far.

I direct much the same sentiment towards the Cambridge Cycling Campaign. They're a good bunch of people who I again largely know personally. However their aspirations remain far too low for what is possible in their city and they are too easily led by the local politicians. The Campaign has now been in existence for seventeen years. That's the same period of time as it took for the whole of the Netherlands to get from Stop de Kindermoord to a very good standard of infrastructure. Cambridge has not made the same transformation over the same period of time.

"Second to Amsterdam" is a remarkably lazy claim which demonstrates the lack of knowledge of those who would claim it. Amsterdam may be ahead of places outside of this country but it's certainly not "the best" place in the Netherlands. For a start, rates of school cycling are lower in Amsterdam than in other Dutch cities.


There's a Dutch saying "Meten is weten". Translated into English this means "Measuring is knowing". The principle is sound: unless you have reliable measurements, you don't really know anything.

Read more blog posts about Cambridge.

10 comments:

Jim said...

As far as I can see the Cambridge Cycling Campaign have presented the Census figures quite reasonably, and you are the one muddying the waters. The Guardian's figure of 18% cycling to work in Cambridge is misleading, since it includes people not in employment and those who work from home (who obviously don't 'travel to work') in the denominator. There is no place on the Census form for people not in employment to record their travel choices, so logically they cannot be included in the denominator for determining cycling's commuting modal share. When you exclude them and the small number who work from home you get a figure of 32% for cycling's share of commuting in Cambridge, or 30% if you insist on including those who work from home.

You are also wrong to suggest that the UK Census is equivalent to an opinion poll. It is a comprehensive national survey with a response rate of over 90%, and is a written questionnaire, so there is no 'questioner' to nudge people towards acceptable answers.

David Hembrow said...

Jim, you can't get "a figure of 32% for cycling's share of commuting in Cambridge" from these figures. The census was never intended to be able to tell you this. What you describe as a "comprehensive national survey" is actually the result of asking a single question: "How do you usually travel to work ?". The result of this question in 2011 may be comparable with that of the same question in different years, but that's it.

The question's form removes all nuance from answers. Respondents must pick one from a list.

Not everyone who cycles rides their bike to work every day. There are many reasons why. i.e. some people simply enjoy cycling three days a week out of five, some only cycle when its dry, some always drive on one day of the week in order to shop at a supermarket on the way home. Others drop off children by car on some days of the week and not others. There are many reasons. However all those people are counted as 100% cyclist by the census.

People who identify polically as "cyclists" will write down "Bicycle" whatever their normal pattern of commuting.

Neither of these issues affect other modes in the same way.

It is of significance that the census was on 27 March. the start of Spring. Many people take to cycling again in Spring. People who decide in winter that they must take more exercise make their start in Spring (they're also often give up before winter. It happens every year). This time of year will cause a boost to "Bicycle" as a response.

That 18.1% of correspondents from Cambridge answered "Bicycle" to the question on the form doesn't mean that 18.1% or 32% of commutes in Cambridge are by bike. It tells you only that 18.1% thought that was the best single answer to the question.

Removing people who work from home is not logical. Someone who works from home three days a week and commutes two day a week (I knew people who did this in Cambridge) can only enter "Work mainly at or from home". These answers have an equivalent value to the "Bicycle" responses and similar error. It doesn't tell you how many people never commute.

The third paragraph of the press release says "Human power is now more popular in Cambridge than the internal combustion engine." It's an example of trying to mislead through "clever" language.

While "walking and cycling" are presented together to represent "human power", the comparison is against a false opposite. i.e. motorized transport with internal combustion engines. What is written is factual, but this deliberately misleading "fact" gives an unrealistic impression of how many people cycle in Cambridge.

In short, all this had one purpose and that was to make the cycling percentage appear higher than it actually is.

The figures have then been presented, just as you have, in ways which do not follow logically from the data. GIGO. You cannot calculate the true cycle "commuting" modal share from these figures. I would in any case suggest that counting only commuters is itself a way of cooking the books because this excludes the demographic groups most difficult to attract to cycling.

Response bias doesn't require a human questioner, just for respondents to perhaps think they "ought" to behave in a particular manner. e.g. they may never do so but perhaps they ought to cycle to work.

Jim Moore said...

The ice must be melting in Assen as the big bear has awakened! ;-)

Adelaide in South Australia also excels at exaggeration. The State Government's cycling target is couched in similar woolly numbers: (See http://saplan.org.au/targets/2-cycling). What this target doesn't make explicit is that the cycling mode share is about 1% and thus the target by 2020 is only 2%. Despite this, because Adelaide hosts the Tour Down Under (basically the first serious training run of the European cycling season) many South Australians, including Transport Ministers, proudly boast that Adelaide is the cycling capital of Australia. Adelaide is hosting the ECF's Velo City in 2014 - I expect the propaganda will be ramped up enormously. I hope the Euro visitors to the conference don't hold back on their criticisms of our almost non-existing cycling infrastructure for everyday trips (and of course our stupid mandatory helmet law).

Cambridge appears to be quite similar to Davis in California, another university town. Much propaganda but the stats tell a completely different story. I read a recent paper where the mode share in Davis has dropped from about 30% in the 90's to about 18% today. The cause of this is partly because they have failed to update their cycling infrastructure over this time, unlike the Netherlands which is continually improving theirs.

Thanks for this post David, it is much appreciated as usual.

Jim said...

David,

Thanks, if I ever need to know how to rubbish a survey without any real evidence to back me up, I'll know where to come. You don't have any evidence that the Census overstates cycle commuting but you sure do have a lot of theories! Of course, there are equally plausible-sounding arguments I could make to suggest that the Census understates cycling, but in the absence of any proof either way the most reasonable thing to do is to believe that when people write on a Census form that they usually cycle to work they're not just making it up, even if for some reason you would rather they were.

I honestly don't understand why you're so keen to convince people the Census results are exaggerated. It's almost as if you're actually disappointed by the thought that in some small parts of England quite a few people cycle to work. Your point that cycling is much more widespread in the Netherlands is a very important one, and the accompanying epic of straw-grasping only detracted from it.

Jim

3rdWorldCyclinginGB said...

Would you not, however, agree that the census data are proportional to modal share in the sense that Cambridge has one of the highest percentages of cycling modal share in the UK, and that most areas have an order of magnitude less? Most areas have such a low percentage that no amount of exaggeration will make them look good.

Massaging the stats aside, it doesn't matter what anyone clains about modal share for Cambridge and a few other "successes", if the rest of the country is uncyclable by any reasonable European standard. But in any case, it's pretty clear that the UK is going to continue to be a world-wide laughing stock in terms of cycling provision for the forseeable future when the so-called cycling minister shows such a marked lack of ambition(1); Scottish politicians openly deride attempts to promote cycling infrastructure(2); and even the relatively high modal share of Cambridge doesn't appear to make the anti-cyclist sentiment and activity any less vituperative (as documented by such as the cambridgecyclist and cottenhamcyclist blogs).

(1) http://www.guardian.co.uk/environment/bike-blog/2013/jan/30/cycling-uk-netherlands

(2) http://www.magnatom.net/2013/02/balance-and-vision.html

David Hembrow said...

Jim: I've heard it before. If you can't argue against the points that someone makes, attack the person instead. I could just as easily turn that on its head and say that you're the man to go to should anyone want to "prove" that 2+2=5.

The census has to be seen as what it is. Once every ten years, the British government asks the population a series of questions in order to gather rough information about the population. The stated reasons for the census over time has varied from working out "tax purposes", "to ascertain the number of men able to fight in the Napoleonic wars" to the "calculation of resource allocation for regions".

Never has there been a census which had a primary purpose of gathering data about the modal share for cycling.

In 2011, there were 43 personal questions, the 41st of which is the single question about transport, "How do you usually travel to work?".

In 2001 the same transport question was asked in a near identical manner (though on a different date, the end of April instead of end of March and with the possible answers supplied in a slightly different order).

I'm not accusing the census of bias either for or against cycling. That would be ridiculous as the census clearly wasn't designed to find out about cycling at all.

To gather accurate information about the cycling habits of the nation requires a lot more than asking a single generic question about commuting transport modes just once every ten years.

The questions asked would have to allow people to state how often they cycle, not just to choose between 100% cyclist or 100% some other mode.

The straw-grasping happens when people take this limited information and extrapolate from it. Go and read the press release for yourself and tell me that this was not done.

The press release doesn't just ask people to believe that the answer to this single question is definitive, or even just to believe that 18.1% of respondents who chose "Bicycle" can be assumed to be 32% of commuters. It goes much further than that, stating that these figures represent an increase of actual cycling of 37% since 2001 (a far larger percentage increase than you get by comparing the 32% claims in the city now with the 28% claimed for the 2001 census in 2003), that there are "5000 more bicycles" in Cambridge, that the "hard work of councillors and officers" in the city have lead to the increase, that this has happened "because of the money spent on cycling", and additional notes 4, 5 and 6 suggest that the reader should do their own further extrapolation based on no evidence at all.

The press release contents do not logically follow from the census data.

David Hembrow said...

3rdWorldCyclingGB: I not only agree that the data suggests that Cambridge has one of the highest percentages of cycling modal share in the UK, but I wrote that in the blog post that you've just read. Indeed, I lived there myself because there was more cycling in Cambridge than elsewhere in the UK.

Further, when you get past the first part of the blog post about the exaggeration, take a look at the two maps, showing Cambridge as an island of cycling, and how the surrounding countryside starting even right next to the city has almost no cyclists at all.

I experienced anti-cyclist sentiment in Cambridge myself during the 20 years that I lived there. Not pleasant at all, and it was rather distressing to find that the police were as guilty of this as anyone else.

Also see the maps showing the 20 km radius from Assen and Cambridge. That a 14 year old child was unable to cycle 11 km to get to school in the city which claims the highest cycling rate in the country is a good example of how "the rest of the country is uncyclable by any reasonable European standard" starting right on the doorstep of the city. Indeed, while quite a few secondary school age children do cycle in Cambridge, their cycling is also depressed relative to Dutch levels and this is a function of the low subjective safety even within the city, relative to what would be considered normal in any Dutch town.

Cycling in Cambridge has a social division between the demographics who cycle and those who do not. This divide cannot be spanned by those in the group who already cycle telling everyone else how wonderful it is.

It amuses me that the boosters of Cambridge continue to claim great things apparently quite blind to the problems that exist. We don't see stories in the newspaper over here about families who are considering giving up cycling because of danger.

andreacasalotti said...

I was in Cambridge last Saturday, cycling for four hours; obviously an insufficient time to have a full picture, however my impressions:
1. cycling facilities in the centre OK, but certainly below Dutch standards
2. cycling facilities in the suburbs poor and drivers behaviour as inconsiderate as in the rest of the country.

Overall it feels that decent facilities have been built only where it was easy to do, or where they made a (politically useful) statement; but overall the bicycle is not privileged; on the contrary, if you live two km. from the centre the infrastructure tells you "You need a car".

I agree with David: exaggeration is harmful because it gives politicians an excuse to take credit for the increase in cycling, when in actual fact they have done insufficient work.

andreacasalotti said...

Probably the reason I felt that Cambridge motorists were inconsiderate is explained here:
http://cottenhamcyclist.blogspot.co.uk/2013/02/the-failure-of-cambridgeshire-police.html

Photo Blogger said...

A sustainable transport association from Austria called "Verkehrs Club Österreich" compared cycling levels in European cities, putting the modal share for cycling in Cambridge at 27% (similar to the university towns of Freiburg in Baden-Württemberg (Germany) and Ferrara in Emilia-Romagna (Italy): http://www.camcycle.org.uk/newsletters/109/article14.html