Welcome to Netrider ... Connecting Riders!

Interested in talking motorbikes with a terrific community of riders?
Signup (it's quick and free) to join the discussions and access the full suite of tools and information that Netrider has to offer.

Report on risks of motorcycling

Discussion in 'Politics, Laws, Government & Insurance' started by Chollima, Nov 8, 2009.

  1. There's been a bit of discussion recently of risks for motorbikes vs cages, and evaluation of risks for riders. As numbers is my thing, I decided to find out for myself. For fatalities data I used the Australian Fatal Road Crash Database. To calculate fatality rates I used the (now annual) Australian Motor Vehicle Census, which contains registration numbers by vehicle type and state.

    A couple of quotes from my report are as follows:

    In 2008 in Australia, there were 4.65 driver fatalities for every 100,000 registered motor vehicles (excluding motorcycles), and there were 41.23 motorcycle rider fatalities for every 100,000 registered motorcycles.
    We conclude by emphasizing that the roads are becoming safer for both motorbike riders and motor vehicle drivers, and that motorbike riders have nine times the risk of death in a traffic accident than vehicle drivers.

    The report can currently be downloaded in pdf using the link on the following website:

    Although it is 10 pages it is virtually all graphs. Please use as you wish, as there is no copyright. I ask one thing: if you want to comment on the quotes above, please read the report first.

    I'll be interested to hear what you think.
  2. Unfortunately the average distance per registered bike is far less than those of cars, and when the TAC are assessing there numbers they are including recreational riding, which is commonly on unregistered bikes. So trying to argue the statistics is hard when there is no common grounds for measurement. (The TAC have seen to this)
  3. First up, well done. I've been playing with the figures myself for a while now, but I never went to the effort of writing it up. Bravo!

    Falcon-Lord has a point about the figures never quite being comparable. I don't know whether it is a deliberate strategy, but it definitely does make it difficult to present a clear case. So the TAC use their advertising budget to present whichever figure they think makes the best sound bite, and we all sit back and scratch our heads about where they came up wit the numbers.

    About this time last year they presented something that made it look as though male riders in their 40s were more at risk of collision than anybody else, but when I looked at the data it was totally bogus. They had broken down the age ranges in 4 - 5 year brackets, except for one huge bracket that covered more than 10 years. Of course it looked as though members of that age bracket had a higher risk - it was a bigger bracket! But it suited the message they wanted to put out at the time, which was that weekend riders who were just getting the bikes out of winter storage for the start of the good weather were at risk. Why let facts get in the way of a good statistic - particularly if you can twist it to say something you want to say.
  4. There's reality, and there's statistics. And there's politics. TAC are interested in the last two (I assume, as FL does, that TAC is the elephant in the room here).

    TAC is on a mission, and if it sees some numbers that will help it fulfill that mission, it will use them. It's assumed by some of us that discouraging motorcycle use is part of their strategy. If we we want to oppose that particular part of the strategy we need to use numbers just as strategically as they do. Chollima, you have donated some valuable building materials to that cause, for which we should thank you.
  5. Great effort

    Thanks for putting it all together. Good to see compulsory motorcycle training has brought the rider toll down.

    Shame about QLD and WA increasing of late - could it be due to the mining boom of buying bikes with big pay packet and not that many days off to enjoy it so packing in a heap of fun where they can often pay the ultimate price???
  6. You're right FL. I've now got a Version 2 up of my report that accounts for this. No extra graphs though. Did an analysis of fatalities per billion kilometers travelled by registered vehicles, but there is too much sampling variability in the distance estimates for it to be useful.

    And on risk of bike vs cage, I managed to get up to a factor of 34.6, but to do so requires you to make a statistical error. Quoting from my own report:

    The relative risk values given above also do not account for kilometers travelled by each vehicle type. Accounting for kilometers travelled increases the relative risk value since motorcycles typically travel less kilometers within a given time period than other motor vehicles. The 2007 SMVU gives two estimates for average use by vehicle type: including or excluding registered vehicles that did not travel at all during the 12 month reference period. For all registered vehicles (i.e. including vehicles that were not ridden or driven) motorcycles averaged 3,700 kms and other motor vehicles averaged 14,900 kilometers. For used registered vehicles (i.e. excluding vehicles that were not ridden or driven) motorcycles averaged 5,000 kms and other motor vehicles averaged 15,600 kms. This yields relative use figures of 4.03 and 3.12 respectively.

    The value 4.03 should not be used: the distributions of distances travelled for both motorcycles and other motor vehicles have spikes at zero corresponding to unused vehicles, and to summarize such distributions using a mean average is not appropriate. One should therefore use the value 3.12, which is lower as a consequence of there being relatively more unused motorcycles. Taking the worst case scenario for the motorcyclist, we can estimate the relative risk, accounting for kilometers travelled, as 8.58 \times 3.12 = 26.77. So accounting for kilometers travelled increases the relative risk from 8.58 to 26.77. This adjustment for distance travelled should be used with caution however, given the sampling variability of the SMVU data.

    We conclude by emphasizing that the roads are becoming safer for both motorbike riders and motor vehicle drivers. Motorbike riders have 9 times the risk of death in a traffic accident than vehicle drivers based on fatality rates using number of registered vehicles, but this figure increases to 27 times the risk when taking into account the distance travelled by each vehicle type.
  7. Great work.

    Can you put an Executive Summary at the start? If the conclusions could be summarised - that would be nice, and we could use it somewhere.
  8. Done. Version 3 up now. Open source publishing!
    Exec summary as follows (not WYSIWYG):

    This report presents a comparison of motorbike rider fatalities and motor vehicle driver fatalities, for the five most populated states and for Australia as a whole, using the number of fatalities per 100\,000 registered vehicles for the 20 year period 1989-2008. Figure \ref{regos} demonstrates the large increase in motorcycle use in recent times. Figures \ref{frdriver} and \ref{frrider} show that roads are becoming safer for all vehicle operators: fatality rates are decreasing. This decrease was particularly prominent in the period 1989-2000. In the period 2001-2008 fatality rates have also decreased, albeit at a slower rate, in New South Wales, Victoria and South Australia. There has been little progress in Western Australia and Queensland during the 2001-2008 period; the fatality rate for motorcyclists in Queensland shows a modest increase. We also show that, when taking into account the distance travelled by each vehicle type, motorcycle riders have 27 times the risk of fatality in a traffic accident than vehicle drivers.
  9. very interesting read.

    Now we just need a graph to show who concentrates more on the road. Riders or drivers?
    (i vote riders!! less distractions! eg. mobile phone, stereo, mates being idiots etc etc...)
  10. One of the things that I'm getting interested in is the difference between the risk profile here in Australia compared to other OECD countries - in particular the EU and UK. In general, we have a higher rate of injury accidents per 100,000 registered vehicles - which is not good.

    We also have a higher proportion of single vehicle accidents, which strikes me as odd. In the UK and Europe and the US it is usually somewhere around 70% multiple vehicle accidents (usually the other vehicle failing to give right of way to the bike) and around 30% single vehicle accidents. In Australia we have a little over 50% single vehicle accidents, and a little under 50% multiple vehicle accidents. Those accidents that do involve another vehicle seem to match the pattern seen in the EU and UK and USA - the majority of them are the fault of the other vehicle.

    But where does that difference come from? Why do more Australian riders die or injure themselves than their overseas counterparts? And why do more of them crash in single vehicle accidents? Is it something to do with our riders? Our licensing and training regime? Our road engineering and maintenance?

    My suspicion is a little of all three. I've seen motorcycle exams in the US, and in some states they are much, much harder than here in Australia. But then again, Vic has a criminally easy P test and yet we have the best accident figures in the country.

    What we don't have is a high population base to make maintenance of country roads economically viable. So I suspect that our roads are in worse condition than similar roads in Europe and the UK. But I don't know, and I haven't found any research to indicate why the discrepancies exist.
  11. do we know for certain that those accidents quoted as single vehicle accidents here, truly are? We hear stories from NSW and places where the other party who caused an accident isn't identified and the rider is then booked as having had a single vehile accident. In a fatality isn't it possible that the accident could be misclassified if there are no witnesses?

    as to a comparison between UK and here, there are couple of things that spring to mind: proportionally, I believe, there are more motorcycles in the UK. The way motor vehicle insurance works there means that it is impossible (literally) for a young person to get any type of insurance on anything like a performance car. This means that at 16-20 they graduate towards motorcycles in preference. The higher (I believe) numbers means that in general other motorists are more aware of them - they are certainly more tolerant - and as has been said, much safety advertising focusses on looking out for bikes.

    There is also a very real difference in the quality of roads. In general even major highways here are in a much poorer state of repair than in England. One of my friends is a highway engineer and it is not uncommon (although maybe not too common) for corners to be reengineered to be banked allowing faster, but safer, negotiation by motorcycles.

    Finally, there is a marked difference in the skills of car drivers between the UK and here. I am constantly shocked by things I see here. I don't know if it is this culture of driving a big V8 on cruise and steering with one wrist hooked over the top of the wheel or what? In many middle aged drivers they seem to be carrying on the way they used to twenty years ago when there was half the amount of traffic on the road, and making no allowances for things that have changed. IMO it is only the physical size of some of the roads and runoff areas that keep the road toll as LOW as it currently is.

    anyway, just some thoughts......
  12. some of the graphs have wildly varying cyclical swings (eg WA rider fatalities per 100,000 vehicles) I assume this is simply because the low number of registered vehicles means that a 1- or 2-fatality change from the previous year multiplies up into a 30- or 40-'fatality per 100,000' change. Is this the case? if so, how statistically valid are any of these numbers (extrapolating from a small base)
  13. And the worst weather for riding...so guessing fewer days/kms ridden
  14. Absolutely. But I think the same would hold true for other countries as well, don't you? In which case it would not explain the difference in the ration of single to multiple vehicle crashes.

    Yes, I agree with everything you said there. Except that the conclusion we should draw from that is that in the UK there should be fewer multiple vehicle collisions compared to single vehicle collisions because the car drivers are more alert for bikes. The figures show the opposite of this trend. In the UK there are more multiple vehicle accidents than single vehicle accidents. If they have better awareness of bikes - which I agree they probably do - then it must make our single vehicle accidents even more of an anomaly.

    Yeah, this is what I'm starting to think is the main suspect.

    I think this one is like the second point: if our car drivers are worse you would expect to see a higher proportion of multiple vehicle accidents compared to single vehicle accidents. We see the opposite.

    See why I think it is curious? We seem to be flying in the face of the rest of the developed world when it comes to motorcycle accidents. I think that is worth some investigation.
  15. According to the ABS figures for 2008:

    NSW 566 million km ridden
    VIC 479 million km ridden
    QLD 531 million km ridden
    SA 117 million km ridden
    WA 226 million km ridden
    TAS 40 million km ridden
    NT 22 million km ridden
    ACT 41 million km ridden

    I guess that is total though, not km per registered vehicle. I haven't found those figures yet.
  16. Nice work, very intersting read.

    I was surprised to see that the ratio of single vehicle accidents was higher for cars (51%) than motorbikes (41%). The last Stay Upright course I did said that most motorcycle accidents were single vehicle accidents on corners. But maybe the majority of these are not fatalities ?

    There is alot more variance in Rider Fatalities versus Driver Fatalities. What's your take on that ?

    The rider fatailty by age is a scary sight isn't it ! I'm glad I got through my twenties :) Why do you think that's the case ? I reckon it is mainly a function of speed and the fact that we tend to think were invincible when were young. Also experience I guess. The more k's you rack up the better your chance of survival.

    Why do motorcycles typically travel less kilometres within a given time period than cars ? Personally I do 300 K's a week on my motorbike and be lucky to do 30 K's in my car, however I suppose there are a lot of bikes that are just used for weekend warriors, but even those would rack up some k's.

    Finally, that last graph concerns me. That the relative risk of motorcycling is trending upwards. Hopefully it would be moving downwards
  17. Responses to comments:
    Disclaimer - I'm a numbers person and not a traffic person, so on the "why's" I have no idea.

    Not sure. Perhaps because there is more travel in urban/city areas in other countries, or perhaps because Australia has poorer motorways/freeways/highways compared to other countries?

    Yes, that's correct; it's due to natural variation (not cyclical) and is expected where low number of registered vehicles are involved, such as in smaller states or with bikes relative to cages. If you wish to extrapolate into the future, any statistical (time series) model would give you both estimates of future fatality rates and an assessment of how accurate or otherwise those estimates may be (a confidence interval).

    The actual historical values have no error; we know exactly the number of fatalities and exactly the number of registered vehicles (ignoring the issue of the timings of the Motor Vehicle Census, with is of minor importance). This is not the case with distances travelled, which are sample estimates: in layman's terms, the distance values are guesses, and not particularly accurate ones.

    Surprised me too, which was my reason for including it. As you say, if you have data on all reported accidents, things would probably be different.

    That's entirely to be expected; less bikes vs cages means less information and therefore more year-by-year variation. You should also see more variance in the smaller states. The smallest states TAS, NT and ACT were omitted since the year-by-year variation in fatalities per registered vehicles is so high that plots by year are useless. Unfortunately this variation can be abused by those who wish to misuse statistics; see example in my next post.

    I think you may be misinterpreting the plots; Figure 3 gives fatality numbers, and this is not a measure of risk; it may just reflect usage. For example, there are more fatalities of riders in their twenties, but this may simply be because there are alot more riders in their twenties than any other age group. It would be great to do similar plots for fatality rates, but I couldn't find the data that would enable this.

    Not sure on that. I guess there must be many bikes stuck in garages and hardly ever used, which is a bit of a shame.

    It seems to be going up a little, but you can see why by looking at the fatality rates for Australia in the bottom plot of Figures 5 and 6: riding bikes is becoming safer, but driving cages is becoming even more safer, hence the slight rise in the relative risk over the last 20 years. I would guess that this is due to additional safety features of the cages themselves (air bags, crumple zones etc), as bikes cannot benefit from this technology.
  18. I'm putting my teaching hat on now.
    The following is a quote from a media release by the Victorian minister for roads and ports on the motorcycle levy dated end of August 2007.

    If you are into numbers, you should recognize this type of statistic as potentially being from a data dredging exercise. It's a stat that compares a base year (2001) to the current one (or current 12 month period). There is potential for somebody to calculate the stat for lots of different base years, and report only the one that looks the best.

    I have a question for you: why did Victoria use the base year 2001 for this stat, even though they didn't start collecting the bike levy until October 2002?
    Clue: Look at the line for Victoria in the bottom plot of Figure 1 in the report!