Saturday 12 September 2015

It's An Accident Waiting To Happen!

They are the phrases so often seen in the local press; "something must be done before someone is killed" and "it's an accident waiting to happen"; and perhaps, they're a proxy for an underlying dissatisfaction with how our streets are arranged and managed.

Accident is not a word to be used when talking about our streets, simply because it has overtones of randomness, of an incident which simply happened and couldn't be prevented (although unless you have a TARDIS, you are not going to be able to prove prevention). The words to use are collision or crash, the second of which is perhaps getting a little more emotive, but often used emotively with good reason.

Those of us who design highway schemes routinely use casualty data to look at historic trends and to find patterns, but they really are the tip of the iceberg and so we also need to think about risk (plus looking as casualty data is another post in its own right). Road Safety Audit, when used properly, can highlight potential risks during the design and implementation of a scheme or we can simply consider collision risk in that similar types of features are likely to create conditions for similar collisions. So, we have data showing "things which have happened" and can use our experience to look at "things which might happen", but what about the space between, the "things which could have happened"?

Chris Boardman launches the event, talking about our inner chimp!
Fortunately (at least for people on bikes), there are people looking at just this. The Near Miss Project has just published a report on its first year and it makes for some interesting reading. I will leave you to read the detail, although I will pull some things out which I personally find interesting and useful as an engineer. Of course, I am also interested as I took part in the study which collected and analysed 4,000 incidents from a day diary kept by 1,500 volunteers. The day was selected in advance (so no picking the worst day of a week in retrospect), but yes, this type of thing does rely on honesty.

After the official launch which was kicked off by Chris Boardman (yes, that one) with the report highlights given by Dr. Rachel Aldred, those attending split off into group discussions on various topics including engineering and legal issues, plus there was lots of non-structured discussion too. Now, I am from the pro-infrastructure end of the cycling debate as you know (which doesn't mean building cycle tracks on every single road before someone takes issue) and it was interesting to listen to plenty of views which I don't agree with. Interesting, because the views often come as a result of people's experiences and how they cope with the conditions out on the streets. I have seen comments on Twitter suggesting that the report is there to sell segregation or that it dangerises cycling.

It has been said by some that the "near misses" recorded in the study were not real incidents. It may be that those articulating this point are doing so because they either don't recall being exposed to the same kind of thing or see it is happening to those "less experienced" cyclists. Additionally, these "non incidents" are only ever going to be anecdotal; they are only people's experiences and opinions plus, different people would view the same kind of things differently. Well, that is the point isn't it? There was also debate around the so called "safety in numbers" effect (which wasn't found by the research).

We are in relatively new territory here, because we are now asking the end user to provide feedback on the current situation. In the same way as casualty-analysis can show patterns, the use of this kind of data (and location was logged) will start to show patterns of where people feel most exposed.

So, back to the things which interested me especially. Despite this project having those reporting deciding for themselves what was scary or just annoying (with no thresholds), 94% of the reports classified near misses in 5 ways;

  • having the way blocked (38%)
  • a problematic (usually close) pass (29%)
  • a vehicle pulling out/ in (16%)
  • being driven at (6%)
  • left/ right hook (5%)
Having the way blocked is interesting as it is those situations where a cyclist needs to pull out to get round something such as a parked car. This type is issue wasn't considered as being particularly scary, but over time of course, it makes cycling hard work. The scary issues were the other four categories (56% together) although being driven at is not necessarily a driver being aggressive, it can be situations where there is a single lane available because of parked vehicles. The report muses overlaying the data with casualty records to look at other patterns which is a very good idea.

It was suggested that in areas with higher levels of cycling, there was no reduction in the rate of near misses (against the safety in numbers proposition). The project hopes to expand internationally in the future and it would be very interesting to look at near miss frequency in those high cycling places which coincidentally have well-developed and safe provision for cycling.

I visited the Victoria Embankment after the launch. On busy routes
like this, the provision of high-quality infrastructure to keep people
who want to cycle away from vehicles and will provide good levels
of experienced safety.

If you don't want to cycle in places like this, then fine, but this is
designed for people who are scared of riding in heavy traffic and with
heavy vehicles such as this low loader.
The other point which interested me was that the faster people cycle, the less likely they are to report near misses. This conclusion was stripped of demographic issues and so was stated with good confidence. Cyclists how reach their destination at 8mph or less (over the whole journey) have three times more near misses per mile as those travelling at 12mph or faster. To put this into context (with an anecdote!) I had a look at my times for the Ride London 100 (well 86) from last year. For the first 17 miles, I averaged 16 miles per hour which was after a lot of training with an overall average speed of 11 miles per hour. I think my normal pootling speed is 9/10 miles per hour (I don't measure it) and so from the study, I would reckon that those at the 12mph end are most likely to be those who are fitter than me, cycle further, are used to mixing with traffic and are more numb to near misses because of the way in which they cycle as a way of coping with the conditions.

The good news is the study is going to continue and is looking for people to record their experiences towards the end of next month and you can sign up here. Beyond this, it is hoped to go international. What would be amazing is if in some way near misses could be reported, mapped and classified automatically and made open-source, just imagine the possibilities. This type of work is so valuable in understanding experienced safety because we can throw statistics at people about how safe cycling is in absolute injury terms, we can go on about the mode share for a particular area, but it is for nothing if people feel scared.


  1. Quote "The report muses overlaying the data with casualty records to look at other patterns which is a very good idea".
    TBH I think this is a vital part of the project if it is to mean anything practically to engineers, and not be written off as a collection of anecdotes. Having said that, conflict studies have been around since Spicer* in the late 60s/early 70s, and to my knowledge have never been used as a systematic tool to identify and remediate potentially dangerous sites by highway authorities. In fact in twenty years I've only ever done one. The merit of introducing this into the real world is that since it uses self-reporting it wouldn't require massive LHA resources for the data gathering.


    Andy R.

  2. I know where you're coming from re casualty data. It is a very important how this is measured because if it's just about reducing the numbers then you end up where the UK and Australia has i.e. with a low number of casualties for vulnerable road user modes (walking and cycling) achieved mainly by deterring these modes and forcing people into private/public motorised modes or worse, not travelling at all.

    It is interesting that just yesterday Dr Aldred tweeted that Sweden are reviewing their Vision Zero to incorporate health and other benefits ( I'm guessing this may mean accepting an increase in the raw casualty crash numbers for vulnerable modes but at a reduced casualty crash rate, either per trip or per km/mile, but with a long-term reduction in casualties from lifestyle disease caused by too much passivity.

    1. Indeed - there is much more going on behind "official" data - how do we capture people who are inactive because they don't feel able to use our streets through fear?

  3. I've read the paper, and was somewhat astonished by what it said. If I had one of those diaries, I think the most likely result would be that I forgot I was carrying it. I'm old and slow, but have ridden many miles. I don't get blocked, unless a traffic light turns red, presumably because I've learned to look ahead for the potential blocks. I've mostly learned to prevent, but look our for, unexpected right hooks. I ride far enough out from the kerb so that the cars behind me think rather than left hooking me. There's an art to that - if you block people you want to ride far enough out so they think they can't overtake because the road is too narrow, but not so far out that they think they can't overtake because you are deliberately blocking them. Bike lanes frighten me because of potential left hooks, and maybe right as well, and cycle tracks are, notoriously, worse.
    I think over the years I've ridden a bike about 100,000 miles, driven a car about 300,000, and travelled by plane for about 200,000.
    The diaries only recorded the first 10 incidents (in a fortnight) It might have been interesting to know more about the people at the extremes, both with many and those with few/no incidents
    The paper prompted me to look at the safety in numbers theory, supposedly a myth, maybe confirmed as a myth by a TRL study for TfL But I think its more complicated than that