Drivers at fault in majority of cycling accidents

18 comments
By BikeRadar | Tuesday, Nov 23, 2010 11.45am

New research from Australia* has shed light on the causes of collisions and near misses involving cycle commuters. 13 adult cyclists in Melbourne were given helmet-mounted video cameras and asked to film 12 hours of commuting each over a four-week period.

In total, 127 hours and 38 minutes of usable footage was obtained. 54 'events' were captured on film – two collisions, six near-collisions (where rapid evasive action by the cyclist was needed) and 46 other incidents (where some collision avoidance was required).

The cameras also recorded the road position and behaviour of the cyclists – including head checks, reactions and manoeuvres. The aim was to identify risk factors for both cyclists and motorists.

In 88.9% of cases, the cyclist had been travelling in a safe/legal manner prior to the collision/near miss. Most happened at or near a junction (70.3%) and most were caused by sudden lane changes by the motorist, with sideswipe the most frequent cause (40.7%).

The motorist was judged at fault in the majority of events (87%), and 83.3% of drivers didn't realise the danger they had put the cyclist in – or at least didn't show any reaction. Riders who frequently looked over their shoulders to check for other traffic were the most successful at avoiding collisions.

While they both happened in cross traffic, the two collisions were very different. One was deemed to be the fault of a motorist, while in the other case the rider was cycling unsafely. In both cases the driver didn't see the cyclist.

Three of the six near-collisions involved trucks, at least five of the drivers didn't see the cyclist, and at least four of the incidents were deemed to be the motorist's fault. The vast majority of all types of incident (87%) happened where there was no traffic control, such as traffic lights or signage. 4x4 drivers were particularly likely not to see the cyclist (85.7%).

The authors of the study concluded that there was a need to improve driver awareness of cyclists and other road users, and to encourage motorists to use their indicators for longer. They also called for greater consistency in cycling facility design. In addition, they highlighted some things cyclists could do to improve their safety on the roads.

These include checking to the left more often (in countries where you ride/drive on the left, the tendency is to look right more often than left because vehicles in left-hand side roads should give way and are thus seen as a lesser threat), riding more defensively around cars and being particularly vigilant when it comes to drivers turning left at junctions, especially if they're in a large vehicle like a 4x4 or lorry.

* 'Naturalistic Cycling Study: Identifying risk factors for on-road commuter cyclists' by Marilyn Johnson, Judith Charlton, Jennifer Oxley and Stuart Newstead at Monash University Accident Research Centre in Melbourne (54th AAAM Annual Conference, Annals of Advances in Automotive Medicine, October 17-20, 2010)

You can follow BikeRadar on Twitter at twitter.com/bikeradar and on Facebook at facebook.com/BikeRadar.

You might also like...

User Comments

There are 18 comments on this post

Showing 1 - 18 of 18 comments

  • Whilst this is interesting, having a camera strapped to you recording your every move will in itself make you cycle in a safer manner.

    If the experiment were repeated with the camera attached to drivers then the cyclist would probably be seen to be responsible for more infringements than before.

    That said, it's best to ride as if every other road user is an idiot and make due allowance for them. Then you won't be in the position of being seriously injured but taking comfort in the fact that it wasn't your fault!

  • 1 city - 12 cyclists - 2 collisions. Statistical significance fail.

    Camera strapped to head. Blinding fail.

    etc etc

    Bad science.

  • Fletch, there's not enough information here to say if its bad science or not.

    It is probably appropriate to the population examined. If the participants had worn helmet cams etc before actual data collection, in my experience, they'd probably forget that they had them on. Also, it is possible that qualitative analysis was used on the quantitive data so "statistical significance" may not have been stated.

    A good study would be to compare data between a group trained in defensive riding and another group who was not :)

  • "Drivers at fault in majority of cycling accidents"

    This is the bad science.

    You are quite right about the study, I can't know its quality because I haven't read it. The article/headline above is Daily Mail worthy in its conclusion though.

  • "In 88.9% of cases, the cyclist had been travelling in a safe/legal manner prior to the collision/near miss."

    So 46 in 127hrs 38mins were the other vehicles fault.

    One every 2hr46mins.

    Doesn't seem too bad compared to London.

    Which makes me think the significance is also dependant on the number of cars/opportunities.

  • A flood starts with a trickle, If this piece of work is enough to get others to add to it, there could (eventually) be a real change in thinking... Oh hell lets just b*tch about... your right its not real science HHhhmmmmm

  • Most of the near misses i've encountered on the road were due to the driver not indicating or just not paying due attention to the road. Indicatorsmare there for a reason. Use them.

  • [i]1 city - 12 cyclists - 2 collisions. Statistical significance fail.

    Camera strapped to head. Blinding fail.

    etc etc

    Bad science.[/i]

    Far from it. The sample size and sample period were clearly stated. Perfectly good science. Now if someone extrapolates wildly from this, it's user fail rather than science fail...

  • Not sure if the roads the people above travel on are on this planet, but as a cyclist of over 30yrs experience, I could easily say that at least 95% of deaths or injury are caused by motorists. There is an 'event' involving a motorist almost every ride you go on. Either they don't see you or they don't care, either way it's a sad symptom of modern life. Is it not obvious that a cyclist knows how vunerable they are and therefore would not knowingly endanger there own life? Motorists are safe in their tin boxes so don't have to worry about it.

    How many crashes are there between motorists? How many crashes are there between cyclists as a percentage?

    Cyclists cease to be human the moment they get on a bike, they become an obstacle. Motorists don't care because they don't have to. I ran over a cyclist? So I spend 3yrs doing community service, big deal.

    Open your eyes people, the government wants to reduce emissions? Get people on bikes? Make it safer and they will douche bags...

  • We as cyclists using the road, Commuter or otherwise, don't need these statistics "bad or good science" to tell us how much danger we are in every time we ride. We all know that most incidents are avoidable and caused by motorists driving in an unsafe and Illegal manner around us. If these drivers simply don't see us, how come they are so good at aiming their killing machines at us as they pass?

  • I was a participant in this study and can certify that my behavior did not change whilst I had the helmet cam on. To be honest it was so small and light I didn't even think about it being there.

    The cam also captured third party cyclist behavior, good and bad, and for the record helmets are compulsory in Australia, in-case anyone thinks that selecting helmeted riders can influence the results.

    Hopefully the findings can be put to good use and stimulate some discussion on cycle infrastructure and safety in general.

  • Re Science:

    This is a study attempted in the 'scientific style' - it is mostly behavioural related. It can never be repeated. The conditions are changing and the parameters are to complex. The study can not easily be done blind - although you could have one set of riders thinking they are recording. The analysis of video is always largely subjective - and even if repeated by many observers (cyclists and non cyclists for example) also can not be done blind.

    In science an experiment usually has a hypothesis that is tested under a very strict set of conditions that can be repeated and compared. Statistics are based on uncertainties of measurement and used to test the hypothesis. If a finding is made it can be tested by another group. Where subjectivity is involved a method of doing the investigation 'Double blind' should be used - ie the Investigators and the subjects don't know which group they are in.

    There is no strict definition of what science is -but frequently in medicine and behavioural studies experimenters use 'scientific style' with no deep understanding of the complexity, subjectivity, measure of uncertainties, in measurement and understanding of blind studies and the impact of 'self fulfilled prophecy' on the outcome of the 'study'

    This is really just observation and attempted in the 'scientific manner' - Early science was just simply that and formed the basis of most modern science that uses what is considered modern 'scientific method'.

    So.......The fact that people are investigating this has to be good - But we all know, as many have commented, that drivers mostly have a low awareness of cyclists and mostly cause accidents - you only have to go out and ride a bike for a few hours - or get knocked off - as a driver say's - sorry mate.

    We all know that an awareness campaign is required with continual reminders. But I am convinced that cycling is not promoted due to the economics of the automotive industry. It is just to important to the economy that everybody is enslaved to their car.

    Nobody is really going to explain how safe cycling really is and I refer to an early post that highlighted that despite approx 100 cyclist deaths on the road (UK) this compared to 3000 from everyday falls. All we are ever going to hear is about cycling accidents in a negative context as everybody then heads for their car.

    Let's see a real headline - cycling safer than being in your own home or doing DIY etc. No chance - just fear. I wish we could see more articles on the history and social history of cycling versus the car and put all this in context.

    Tom

  • Having been involved in a large scale Behavioural Management of Safety roll out with a big building contractor a while back, I came to understand that all "accidents" are avoidable if you really want to avoid them, and you will only get involved in an accident if you put yourself in the position to be able to be in one.

    So if you regularly ride up the inside of lorries at junctions, or go across red lights because there's "nothing coming", sooner or later you will be involved in an accident.

    As I said previously, if you're involved in a serious accident, being able to blame someone else doesn't compensate you for your life changing injuries.

    If you want to stay safe, you need to get your head around the need to make allowance for other peoples poor judgement.

  • Great topic. I would be nice to see larger studies and in many different environments. I don't think any bicyclist would disagree that "drivers are at fault in a majority of cycling accidents". I am surprised that from this small sample size and test that the incidence of bike to car collision was so high.

    At anyrate, more and better studies should be done. In the meantime I have written a modest post on surviving while cycling among cars based on my years of successful riding. The post is: http://myworldfromabicycle.blogspot.com/2010/08/bike-safe-my-survival-tactics-for-road.html

    Incidentally, if anyone is interested in videoing their cycling I suggest they read my post of the best camera for the job: http://myworldfromabicycle.blogspot.com/2010/09/buy-best-high-definition-video-camera.html

    Regards,

    David Henderson

  • I'd been meaning to film my commute for some weeks.

    On Wednesday AM as it was sunny I mounted my decent HD video camera on my commuter bike for my 4 mile ride to Bristol Temple Meads Station.

    By freak chance, a total idiot of a taxi driver completely wasted me off my bike. He to my right in the 'straight on' filter lane and suddenly without warning, looking or indicating cut left sharply at a cross road junction. He totally cut me off and only by luck did I go over my bars and into him rather than under.

    It's pretty horrible viewing. I had 2 front lights on and a high vis jacket.

    Simple problem. HE DID NOT LOOK/ INDICATE.

    The moment I told him I had been filming he admitted fault. My camera luckily wasnt wrecked. Ended up in hospital though needing bandages on my knee and ankle.

    Hefty bill heading his way.

  • Okay 100's of riders and 500+ hours each with 1/2 of them with no camera's or something to make it 100% science based, but really time and money is a issue so we have to at times make do.

    2 Accidents are a blip based on 166 riding hours ( or it's CRAZY there ), but hey a good blip for our science!!

    Doesn't surprise me, I hate road riding and avoid it the best I can.

  • Found the above article of real interest but found some of the criticism of its findings rather pedantic and unfair. In my previous life I specialised in Quality Control, one strand of which was statistical analysis. Now it is a few years since I last dabbled but my belief is that this study falls in the area of the binomial distribution, i.e, a trial where only one of 2 result, good or bad, cyclist or motorists, is possible.The length of trial or even number of cyclists involved has no great bearing, only the examination of the incidents recorded. Now if we make the a priori assumption that both are equally likely to make a traffic error (the hypothesis) then we can test the sample findings against it. To shorten calculations I used the Normal approximation. My findings were that with 47 of the 54 incidents being the liability of the motorist then the results were highly significant at the 1% level. ie The results were not compatable with our assumption of equal liability to err and indicate that errors were much more likely from motorists. Hardly suprising wheh our assumption would expect motorist to be the cause of about 27 of the incidents.

    If we then calculate the 95% confidence interval, which is an attempt to predict driver error limits for all incidents occuring under the same conditions then the interval is 75% to 93% driver error.

    If you understand all this gobleydook then I'm not telling it right.

    The essence is this. I know that every time I go out on my bike I'm dicing with death. This study only confirms these perceptions. Does it not chime with you fears?The study team has set up this experiment, recorded the results, stated their sample size and objectively assessed the results. No extrapolation or conjecture, only facts as they found them. What's the grouse?

    As for repeating the experiment where is the problem ? Repeat it in as similar conditions as possible. I would not expect exactly similar results but I would be surprised if there were not the same trend. Wouldn't you? Lets not make the perfect the enemy of the good.

    As for blind trials, that one eludes me. How do you record the incidents if you dont wear a camera? Or are you referring to the blindfolds the drivers wear?

    Have I bored you sufficiently?

  • I think of this article as a data point... not enough for a general conclusion, but an eye-opening observation for sure.

    I am a pretty serious cyclist and former commuter who recently got into motorcycling. In order to start the licence process, I took a national safety course for motorcycles, where there were sessions on lane position, hand signalling, where to ride with respect to other cars, the importance of visibiltiy/reflective tape, and how to ride at low speeds. Not only has this really helped me get through a year of motorcycling safely (touch wood), but I noticed the habits really transferred to road cycling.

    A similar, widely-adopted program would be wonderful for cyclists. Even if the motorist is at fault, your actions can impact your visibility. Right of way doesn't matter a lot when you're in a body cast!

Post a Comment:

You need to login or register to post comments.