Thursday, February 3, 2011

Cycling and Statistics

As someone with a background in the social and natural sciences, I was "raised" on statistics by the academic system. If we compare academia to religion, then making claims without statistical evidence is akin to taking the Lord's name in vain. But even beyond academia, we have an inherent faith in statistics as a culture. We respect numbers and charts, and we turn to them for comfort at times of uncertainty. Consider, for instance, this beautiful bar graph:

Now, some of us may have suspected that diamond frames tend to be ridden by men, whereas step-throughs and mixtes tend to be ridden by women, but only numbers and graphs have the power to lift us from the murky waters of speculation. We can now say that, in a recent poll conducted by Lovely Bicycle, of the 221 respondents who claimed to ride mainly diamond frame bikes for transportation, 76% were male. Of the 95 respondents who claimed to ride mainly step-through bikes for transportation, 80% were female. And of the 39 respondents who claimed to ride mainly mixte bikes for transportation, 66% were female. This numerical evidence we can wield like a mighty weapon the next time someone contradicts these tendencies.

Of course the one little problem with Statistics, is that it's mostly BS. In the words of comedian Vic Reeves, "88.2% of Statistics are made up on the spot" - which may very well be the case. But numbers need not be maliciously forged in order to misrepresent reality. There are multitudes of ways in which a study can be flawed or biased from the start, set up so as to elicit particular responses. Often this is done unintentionally, or at least unconsciously, by researchers eager to find evidence for their pet theories. Other scenarios can include how data is processed, or even how the final results are presented. Statistics are highly prone to human error and bias, which means that they are inherently subjective. This, combined with the fact that we respect them so much, makes our statistics-loving culture susceptible to misinformation.

[image via NHTSA]

The idea of statistics and misinformation brings me to what I really wanted to talk about here, and this is something I've been trying to make sense of for a while. I am puzzled by the use of safety statistics in bicycle advocacy, and I am hoping that someone could explain them to me. For example, many bicycle advocacy talks and internet presentations stress that it is safer to ride a bike than it is to travel in a car. In support of this, they use statistics such as this data from the National Highway Traffic Safety Administration (NHTSA), according to which there were about 30,000 motor vehicle traffic fatalities, and about 600 bicycle fatalities in the USA in 2009. These numbers are used by cycling advocates to point out how much safer it is to cycle than to drive. But unless I am missing something, the figures mean just the opposite.

Yes, the NHTSA numbers suggest that in 2009 there were 50 times more motor vehicle fatalities than there were bicycle fatalities in the US. But those numbers mean nothing until they are weighed against how many cars vs bicycles there are on the roads at large. If there were 50 times more cars on the road than bicycles, than the risks of fatal traffic accident would be equal for each mode of transportation. But I believe that in actuality, there are more like 1,000 more cars on the roads than bicycles... which means that the number of cycling fatalities is disproportionately high.

Obviously, I am not trying to prove that cycling is unsafe. But I do want to understand the reality of the situation. After all, if cycling advocates use statistics incorrectly, they open themselves up to some very harsh critique from unfriendly forces. Where could one go to obtain accurate statistics about the number of cars vs bicycles on the roads, and the number of traffic accidents for each?

46 comments:

  1. This numerical evidence we can wield like a mighty weapon the next time someone contradicts these tendencies.

    Yes, but only when arguing within the context of the LB readership. The demographic that participated in this poll may not be representative of, say, the pool of people stopped randomly in the street. The poll has pre-selected a demographic based on an active interest in reading a cycling blog.

    Sorry, just playing devil's advocate here, as someone whose career is also in performing statistical analyses and determining appropriate controls to make statistical data meaningful.

    That said, the figures from your poll fit squarely in my own estimation of the gender split on different types of bike frames.

    ReplyDelete
  2. You know, statistics don't lie, only the people using them.

    Maybe a better statistic for advocates to use would be how many bike/bike accidents resulted in death vs. automobile/automobile accidents. Or bike/pedestrian vs. automotbile/pedestrian. Or how many cyclists are responsible for a auto driver death. Somehow I think it would take a lot of lying using those stats to make it appear driving is safer.

    Thanks for a great blog!
    John

    ReplyDelete
  3. Yes, the NHTSA numbers suggest that in 2009 there were 50 times more motor vehicle fatalities than there were bicycle fatalities in the US. But those numbers mean nothing until they are weighed against how many cars vs bicycles there are on the roads at large.

    I concur. It's an impressive claim until you try to understand what it really means. Data is meaningless unless it is normalized to certain reference parameters. In the example of cyclists versus motorists, normalization would be number of bike-related injuries PER total number of cyclists. i.e., a ratio of parameter of interest over control parameter. But even then, determining those data is not always easy. For example, automobile-related injuries or accidents may be more systematically reported than cycling-related accidents. Therefore, even the available data may vary in its accuracy.

    ReplyDelete
  4. Unfortunately, there are no accurate, nation-wide statistics quantitating bike use. We would really like to know the total number of trips by bike, and the length of those trips in time and miles, to be able to do quantitative comparisons. But most cities do not collect precise data on bike trips.

    There are estimates of total motor vehicle miles traveled and motor vehicle trips. These are also somewhat imprecise, but at least every metro and state will release their estimates of this data.

    For bikes, we can either go by surveys that ask how often the respondents rode a bike in the past week, or we can rely on infrequent Census surveys of travel mode for commutes. Unfortunately, the surveys do not distinguish recreational biking, racing, mountain biking, commuting or other trips, and these sorts of surveys are not very reliable. The commuting survey is also weak, because it only asks "how did you get to work most often in the past week." So if you ride your bike only 1 or 2 times per week (or take transit 1 or 2 times a week) it is not counted. And multi-modal trips are not counted; you are only given one choice.

    The commuting statistics show bikes are about 1% of trips to work, but it is possible the number is higher if many people are riding bikes and then using transit for the majority of the trip.

    As far as the danger of bikes versus cars, should we compare them based on number of trips, based on amount of time used in each mode, or by number of miles? Per mile, air travel looks best, but by time planes are quite dangerous.

    Similarly, bikes look as safe as or safer than cars on a trip-by-trip basis or by amount of time spent on each activity, while cars look safer by miles traveled, based on the best guesses from the weaks statistics available.

    (This is based on Ken Kifer's website from a few years back: http://www.kenkifer.com/bikepages/health/risks.htm
    Tragically, Ken was murdered by a drunk driver while riding his bike in 2003... http://www.kenkifer.com/death.htm)

    ReplyDelete
  5. Good post- I've always noted that discrepancy when those statistics are quoted.
    Let's not even start helmet/ injury statistics.
    You know that 88% of kittens riding without helmets injured their tails while chasing yarn on bicycles :)

    ReplyDelete
  6. In your "The 'Lady's Bicycle': Descriptive, Offensive, or Merely Obsolete?" post you say:

    "And statistically, women are considerably more likely to wear skirts and are thus more likely to prefer a step-through transportation bike." without providing the evidence.

    Just saying! ;-)

    But anecdotally I'd have to conclude you're correct.

    ReplyDelete
  7. The big problem I have with all those safety statistics is that it never answers the question that actually matters to me: "How dangerous is it for *me* (and/or my passengers) to take *this* trip?"

    One thing that's almost always missing is the question of how different modes of transportation are counted. By distance? By time? By "trips"? Are we relying on telephone surveys, or some method of traffic counting? I hardly ever have confidence in the raw data, so I have even less confidence in the analysis.

    The other problem with safety studies is that they only concern themselves with injuries, not with the bigger picture, which includes general health consequences. Alas...

    ReplyDelete
  8. somervillain said...
    " 'This numerical evidence we can wield like a mighty weapon the next time someone contradicts these tendencies. '

    Yes, but only when arguing within the context of the LB readership."


    Come on. I was making fun of myself : )

    ReplyDelete
  9. Yeah, you're about to go down a rat hole. Consider the possibilities:

    1) Deaths per population (favors drivers)
    2) Deaths per hours of activity (favors cyclists)
    3) Death per distance (favors drivers)
    4) Deaths per day (favors cyclists)
    5) Deaths per sunspots (favors...someone)

    I gave it up. But I still read the studies. They're always interesting.

    Incidentally, you may have seen the New Yorker article a few weeks ago talking about the
    "decline effect," the number of research conclusions that don't prove out over time. It would kind of make you question the whole research apparatus entirely:

    http://www.newyorker.com/reporting/2010/12/13/101213fa_fact_lehrer

    ReplyDelete
  10. Velouria said.....
    "Where could one go to obtain accurate statistics about the number of cars vs bicycles on the roads, and the number of traffic accidents for each?"

    Boy, that's a tough one! To find numbers that a both accurate AND neutral will be tough to find. :((

    I struggle to understand the import of this topic in the first place. Whatever number is found a bit of digging will find another different number that will oppose it.

    In all honesty I doubt if any source will be more than statistical opinions since no one will know how many bicycles are on the road at any one moment in time.

    I also struggle to understand the reason for this topic when known life saving cycling equipment is neither worn nor discussed.

    ReplyDelete
  11. The bicycling advocates are in a dificult situation - they want to show that bicycling needs facilities to be safer but they don't want to show it as too unsafe and scare away perspective bicyclists.

    The actual numbers are hard to come by. We know the number of fatalities, because those are carefully recorded. Serious injuties are not as well known as there is some hit and miss in the record keeping. but they are probably good enough.

    But what is not know is the miles ridden, or number of trips, or horus riding (exposure) that led to the carefully recorded fatalities. So it is really hard to estimate the exact "rate" of fatalities to compare to automobile travel or other modes of transportation.


    I and others have done quite a bit work in this area. One anlysis by "Failure Associates" came up with a fatality risk per hour on a bicycle as 0.26 per million hours vs. 0.47 for automobile travel.
    Since automobiles travel, on average, about 2 to 3 times faster than bicycles, the risk per mile is roughtly the same.http://www.kenkifer.com/bikepages/health/risks.htm has a good summary. ken, ironically, was killed on his bicycle by an impaired driver.

    I've worked out the numbers myself and generally come out with a risk, per mile, of about double that for driving. However, that's the risk for the average bicyclist, most of whom have very little experience on a bicycle, don't know how they are suppose to behave on the roads, and includes high-risk groups like children and drunk bicyclists. The average bicyclist in this country is MUCH poorer trained than the average driver. Most bicyclsits I observe don't even have a clue about how to safely bicycle on the road. I would argue that the most minimal training, or a couple of years bicycling, will reduce the risk of serious injury or death in half or more, making it a good bit safer than motoring. A number of studies have shown that "club bicyclists" who are trained by cycling with their bicycling society, have a much lower injury and fatality rate than the average bicyclist.

    John Allen, at http://www.bikexprt.com/research/index.htm, has a lot of information about what the causes of bicycle accidents are and what the remediations might be. But in general, someone who takes, say, the Effective Cycling course is going to have a much lower accident rate than average.

    Facilities may make bicycling more plesant or attractive, but they do little to nothing for safety. But just a little training and/or experience goes a long way to make bicycling a relatively safe method of transportation.

    ReplyDelete
  12. cycler - I've read as much "for" and "against" literature on the kitten hard-hats as I've been able to lay my hands on. Some day I hope to write a calm and multi-faceted response about the whole issue. But at the moment, I don't think I am emotionally mature enough to tackle that. And so I try to avoid the topic here all together.

    ReplyDelete
  13. I totally assumed that those studies took into account relative numbers of motorists/cyclists/pedestrians/kittens.

    My favorite recent stat was that jaywalking is safer than crossing at the proper light because you are taking more responsibility and looking before you strut out into traffic. Holla back, chaos theory.

    ReplyDelete
  14. I was just teasing about the hats- that's why I imagined them on kittens, for the absurdity and humor :) Sorry- didn't mean to provoke controversy, was just commenting on how statistics get used and abused in the context of bicycle safety.

    I think statistics fail to have any significant meaning for so many people on so many issues. Ultimately we make decisions based on our own experiences and circumstances, and statistics supporting or conflicting with our decisions have little to do with how we feel about them.

    ReplyDelete
  15. Honestly, I would be surprised if stats show "transport" cycling is nearly as safe as "transport" driving.

    Until someone shows me that I'm wrong, I'll continue to think that it's *much* more dangerous in terms of accidental injury. Perhaps there are health benefits, discounting accidents, or benefits of convenience. Maybe it's faster and cheaper to commute by bike. Maybe it's unhealthy in the long term to commute in a car because you are breathing concentrated fumes and not exercising. I don't know, but if pure accidental injury risk is considered, my own guess is bikes are an order of magnitude more dangerous.

    The 600 vs 30k fatality stat sounds terrible if we assume 1 bike exposure unit to 1000 car units. My goodness, it's 20 times the fatality rate of cars, then. I don't know if it's quite _that_ dangerous, but I would guess it's up there.

    ReplyDelete
  16. The reality of the situation is that neither the raw numbers or any indexing to use is a predictor of your own personal risk exposure. Evaluating one's own behaviors is a better indicator of how safe you will be.

    The advocates should be forgiven their use of numbers in their rhetoric. Their interest is in persuasion, not science.

    ReplyDelete
  17. cycler - No worries. It's just important to me that people realise I have a reason for doing what I am doing, and my choices are not the result of ignorance. But okay, enough about that.

    MDI - "20 times more likely" does sound pretty bad. To get a more accurate number, we'd need to exclude fatalities that occur during competitive road racing or training, which I expect would bring it down considerably. But those kinds of statistics may be impossible to get.

    ReplyDelete
  18. "To get a more accurate number, we'd need to exclude fatalities that occur during competitive road racing or training, which I expect would bring it down considerably."

    From what I know, there aren't many deaths from racing, at least not professional racing. For instance, more people have died in support car crashes during the Tour de France than the four riders who have died. Excluding deaths during training probably would be more significant. However, in the States it would likely also substantially reduce the numbers of your "total cyclists" pool, leaving your "20 times" number either unchanged or worse.

    http://en.wikipedia.org/wiki/List_of_professional_cyclists_who_died_during_a_race

    ReplyDelete
  19. The biggest variable in all of this is the rider. So while broad statistics can be used to discuss and persuade (gasp!), I believe it's the skill, experience, and attentiveness of the particular rider, as well as their locale and route choices, that determine their individual safety. In other words, a careful, skilled rider who chooses her/his routes carefully is more likely to avoid harm than an unskilled rider who rides carelessly and chooses dangerous routes.

    Alan@EcoVelo

    ReplyDelete
  20. Peter - Absolutely. And as others have said, in order to get an accurate idea of our personal risk, we need to look for statistics that apply to our specific demographic as closely as possible. For example, how risky is cycling for a female, in Boston, riding for transportation, not using the sidewalk, and following traffic rules? That figure would probably be different if any one of those factors were to change.

    examinedspoke - I don't mean necessarily professional, so I guess I phrased it wrong. I mean more like, recreational road cycling vs transportational cycling.

    ReplyDelete
  21. Neighbortease--that jaywalking stat clearly was not compiled in my neighborhood. It seems I cannot drive (or bike) even two miles without a ped bolting out into the street without looking. Most peds seem to have absolutely no clue what is going on around them no matter where they are. (My own no. 1 safety rule is stay alert and pay attention to what's going on around you--because the other guy isn't.)

    ReplyDelete
  22. PS - Putting too much stock in statistics (at least when it comes to personal decisions about riding, not necessarily advocacy) assumes that we have no control over our personal safety while riding. Nothing could be further from the truth. Sure, sometimes skilled bicyclists are unavoidably injured or killed by negligent motorists, but we undoubtedly play a large role in avoiding accidents, even when the motorist is at fault.

    Alan

    ReplyDelete
  23. Mike Jenkins said...
    "The advocates should be forgiven their use of numbers in their rhetoric. Their interest is in persuasion, not science."


    I understand that point of view, but I don't agree with it. They are using science to persuade, which means that their persuasive power rests on their audience trusting the science.

    Now imagine the following headline in a major newspaper:

    "Cycling Advocacy Groups Lie to Public, Attempt to Promote Risky Cycling at Any Cost"

    The article would then go on to paint the advocacy groups as unethical and unscupulous manipulators, willing to sacrifice the public's safety just to further their crazy cycling cause. And what do you think that would do to the persuasive power of these groups, to the public's trust, to the funding allocated to anything cycling-related, and to cycling-friendly legislature?

    One could argue, that cycling advocacy groups who make use of faulty statistics to persuade are shooting themselves in the foot and setting themselves up for a backlash.

    ReplyDelete
  24. I have to agree with Merlin. No matter what the statistics may say, they can never compensate for an individuals skill or competency level & therefore you have to consider how you will be affected by the manner in which you ride & the local road conditions.

    There are many bad drivers on the road & there are many bad cyclists too, while you can't control what the drivers are doing, you do have control over how you ride.

    I personally prefer to think of the British Medical Association study which found that you are better off cycling (with or without a helmet) by a factor of 20:1 than sitting on the couch eating pizza & nachos (ok, I made that last bit up but you get my drift).

    ReplyDelete
  25. "I mean more like, recreational road cycling vs transportational cycling."

    Yeah...hmmm...I think it's tough to separate them for statistical purposes. Here in L.A., the bicycle mode share would probably fall around seventy-percent (I'm making up numbers) if you excluded recreational cyclists. They're all we have, basically, as the number of honest-to-goodness commuters is pretty small. Then there's the case of my college roommate and his old training partner, both of whom used to ride to work in full racing gear... until they were hit by a truck, and his partner was killed. Training or commuting?

    ReplyDelete
  26. Merlin/ Gary - I agree. I am a terrible, neurotic driver in the city, but a surprisingly calm and collected cyclist, which skews my risk factors for each considerably. And yes, a sedetary lifestyle carries its own risks, which are not taken into consideration at all in safety statistics.

    ReplyDelete
  27. Nobody who reads this here will die as a result of a cycling related accident.

    Guaranteed.

    ;)

    ReplyDelete
  28. In the words of comedian Vic Reeves, "88.2% of Statistics are made up on the spot

    In the words of satirist Mark Twain: There are three kinds of lies: lies, damned lies, and statistics"

    ReplyDelete
  29. That's similar to the assertion in Freakonomics (or the sequel) that walking drunk is much more dangerous than driving drunk, when you take the fatal "drunk walking" incidents and weigh them against how many people walk drunk versus drive. SIGNIFICANTLY higher proportion. But the intersting thing is how obviously a drunk walker stands almost NO chance of killing anyone else by walking drunk, while a drunk driver is much more likely to hurt or kill another person. The same could be said of bikers in general - we're more at risk to ourselves than drivers, but much less of a risk to others if we make a mistake or an accident occurs.
    ~crissy

    ReplyDelete
  30. Nice topic, your are right; most statistics are BS(is that Bachelor of Science?) When non cyclists admonish me for engaging "crazy people out there", I quote the CPSC whose study said that comparing mile ridden versus driven, you are 10 times more likely to be injured or killed in a car than a bike. That shuts people up.
    Truth is, I like reality. In 39 years on the street as a cyclist I have been in three accidents. The most dangerous did not involve a car, I fell down, hit my head on the curb and have not idea how it happened, but I was ok. The others were non-injury low speed accidents you can read about here:
    http://simplecycle-marc.blogspot.com/2011/01/safety-comedy-of-erroneous-thinking.html
    In the same length of time, I have had twice as many accidents in cars, no injuries but much more expense for car repair, insurance rates etc.etc. Pick your poison or your BS, I'll choose the bike whenever I can.

    Marc

    ReplyDelete
  31. Isn't that why one is supposed to measure transportation safety by vehicle, estimated average hours traveled and estimated average miles traveled my modal user, not by total fatalities per modal type measured against one another? That's how all the studies I've read comparing the safety of the to two modes were analyzed. I can't say I've ever seen anyone compare total cycling and motorist fatalities to make the argument that cycles are safer than automobiles.

    ReplyDelete
  32. The simple fact is that good, unbiased, current, crash statistics involving cyclists don't exist. The only thing that is relatively indisputable is that safely riding cyclists are much safer than those that ride like idiots, but the counts reflect both types indiscriminantly. From the FAR data, the worst ones are drunk males, riding in the dark against traffic, and without lights. The FARs didn't say if they were wearing a helmet and/or dark clothing.

    Seriously, in my LCI class, a major point was that a cyclist is about equally likely to crash regardless of the distance ridden. Experience and training help. Of course the statistics to support that point are quite old, but at least it has the feel that it might be true.

    ReplyDelete
  33. Apart from all possible flaws that a statistical study might have, one thing I always ask myself when reading statistical studies about safety is: will these chances be the same for me as for every other member of the sampled population?
    For instance, in case of air travel yes, but regarding driving or cycling I don't think so. In this case I have some control over my safety. Lets say a study indicates 1 fatality/4 milion hours.
    Sample: 100 cyclists
    90 were riding recklessly and had 1 fatality/200.000 hours.
    10 were obeying the rules of the road and had 1 fatality/8.2 milion hours.
    A similar example could be given regarding the type of streets where the accidents happen.
    I think that when it comes to safety we have to trust our instincts more than any statistics.

    ReplyDelete
  34. Peter:

    "Facilities may make bicycling more plesant or attractive, but they do little to nothing for safety. But just a little training and/or experience goes a long way to make bicycling a relatively safe method of transportation."

    I'm not sure what was meant by "facilities", but assuming you are referring to cycle paths and the like, I think it is worth reading David Hembrow's essay on the three types of safety: http://hembrow.blogspot.com/2008/09/three-types-of-safety.html

    Once the safety discussion begins to include infrastructure, it becomes worthwhile to compare different cities, rather than just different modes of transport.

    ReplyDelete
  35. MDI:

    Honestly, I would be surprised if stats show "transport" cycling is nearly as safe as "transport" driving."

    My first reaction was to strongly disagree, assuming you don't ride drunk, against traffic, on sidewalks, in the door zone, or as a bike ninja. Hospital studies might help there, but I can't cite any.

    However, it occurs to me that we've mainly been discussing variations in how the bicycle is operated here, neglecting the various ways that people use their cars. Some significant percentage of accidents in which automobile passengers are killed occur on roads where cyclists (almost) never go: divided highways. To properly answer the question, "Am I more likely to die on the way to the grocery store if I drive my car or ride my bike?", we also have to exclude highway fatalities. Nobody ever falls asleep on the pedals.

    ReplyDelete
  36. Alan:

    In other words, a careful, skilled rider who chooses her/his routes carefully is more likely to avoid harm than an unskilled rider who rides carelessly and chooses dangerous routes.

    I agree wholeheartedly with the idea that the skill of a cyclist and the care he or she takes is a huge part of the accident avoidance equation. I'd like to point out, however, that you are neglecting (at least) one group in that statement -- a group that I see frequently: unskilled cyclists who are attempting to ride carefully. They can often be found on sidewalks, frequently against traffic. They do this because they feel safer there than they do on the road. They're not being reckless; they really think it's safer.

    ReplyDelete
  37. The suggested fact that most, or many car fatalities occur on high-speed highways only makes cars appear safer for transportation than bicycles. I don't need to be on the highway to get to work by car or by bike. The only time I ever need to be on the highway, I am making some sort of long distance trip, which I wouldn't do by bike.

    So, Merlin, if we exclude highway deaths from the 30k mentioned above, oh dear, then "transportational" use of bikes may seem even more so dangerous.

    ReplyDelete
  38. Adam & Mark - Actually, accidents per miles traveled is supposed to be a misleading way of measuring the rates, but I completely forget why that is. It made sense when I read about why, so I'll try to find it.

    ReplyDelete
  39. Another thing to keep in mind is that there is a huge asymmetry in the causes of deaths between the two vehicular modes.

    Drivers kill drivers. Drivers kill cyclists. Generally speaking however, cyclists do not kill cyclists and cyclists do not kill drivers.

    So when we describe cycling as dangerous, we are saying that it is dangerous for the cyclist; it is not, generally speaking, dangerous to others. Driving, on the other hand, is dangerous for both reasons - for the driver and for everyone else on the road.

    Of course this observation isn't very helpful when it comes to assessing one's own personal risk with respect to transportation choice. But I think it is a far too often overlooked point when it comes to consideration of transportation policy.

    I feel that in this arena too much emphasis is placed on the cyclist to take responsibility for their own safety (after all, they are the one engaging in the *dangerous* activity) when the emphasis really needs to be placed on the reduction of the dangerousness of cars - both to cyclists and to other road users. By trying instead to remedy the "dangerousness" of cycling (e.g. where I am from there is a compulsory helmet law), policy makers miss the point.

    ReplyDelete
  40. Velouria, miles traveled is misleading when you are comparing cycling, walking, driving, train and air travel because of the huge differences in the miles traversed while the time spent traveling are more similar. Five hours in an airplane will get you 3,000 miles, while five hours walking will get you 17 miles. A similar discrepancy will exist if you compare cycling to driving. Hours traveled or per trip traveled will both result in more useful data, theoretically, if you have any faith in the federally estimated hours traveled per modal share per person.

    ReplyDelete
  41. Most people prefer "exposure" to calculate risks - i.e., time, as it is best for comparing the risk of divergent activities that are necessarily distant based. It allows you to compare the risk of, say, cooking to bicycling or sky diving or going down stairs. But I prefer miles because I'm a transportational bicyclist trying to get from Point A to Point B by bicycle and would like to compare the risk between bike and car.

    As I said earlier, the actual risk per mile (or hour if you prefer) is not accurately known because no one is recording actual miles ( or time) ridden by bicyclists the way they do with cars ( via survey of odometers, etc. ). Most analysis relies on surveys.

    The range of risk ( per mile) from this approach varies from about half the risk of driving to 6 times the risk! But ignoring the outliers, most groups come up with a risk that is 2 times that of driving.

    You probably know the risk of driving or motorcycling falls dramatically with a year or two of experience. the same is true of bicyclists. The problem is, the vast majority of people only ride a bicycle a handfull of times a year and NEVER get any real experience. But anybody who has ridden for transportation for even a year is an "expert" compared to the average bicyclist and will have a significantly reduced accident rate.

    So I'm almost positive that any transportational cyclist has a fatality rate per mile roughtly the same, if not less than, that of drivers.

    John Forester, the controversial "vehicular cyclist" proponent, estimates from bicycle club records and other sources that the total risk to a bicyclist in one year stays relatively constant regardeless of the distance ridden due to this experience factor.

    So a bicyclist who rides 6000 miles a year has the same probability of getting seriously injuried or killed on a bike as your weekend recreational bicyclist who rides 400 miles a year.

    ReplyDelete
  42. I would recommend micromorts as a way of making this type of risk comparison

    http://understandinguncertainty.org/micromorts

    As mentioned before this doesn't factor in the health benefits of active travel generally estimated to outweigh any additional risk by 20:1 for cycling

    ReplyDelete
  43. MDI:

    So, Merlin, if we exclude highway deaths from the 30k mentioned above, oh dear, then "transportational" use of bikes may seem even more so dangerous.

    Yes, this is exactly the point that I intended to make. Apologies if I implied the reverse.

    The safety question, as others have hinted at above, is actually two separate questions -- a personal one, and a societal one. The selfish, personal question is, "How do I minimize risk to myself, and maximize my own health?". The societal question that governments should be asking is, "How do we minimize injury and death, and maximize health?". These are different questions, with potentially very different answers. Arguably, society benefits as the percentage of trips by car decreases, but the individuals forgoing their cars in favor of a (presumably) riskier mode of transport may not.

    ReplyDelete
  44. I was reading Mark Twain's autobiography today and he quoted Disreali as saying:

    "There are three kinds of lies; lies, damned lies and statistics."

    ReplyDelete
  45. MDI,

    " we exclude highway deaths from the 30k mentioned above, oh dear, then "transportational" use of bikes may seem even more so dangerous."

    Nah, not if we are comparing miles, or hours. Divided, controlled access highways (Freeways, etc) have extremely low accident rates (though the fatality rate is only "pretty low"). Rural highways are more dangerous, due to lots of single-vehicle accidents, but still have many fewer accidents per mile than city streets and arterials.

    Excluding highways might make cars look more dangerous per hour, and certainly they would look much more dangerous per mile.

    ReplyDelete
  46. http://en.wikipedia.org/wiki/Air_safety#Misinformation_and_lack_of_information

    See the heading "Statistics"

    ReplyDelete