Login required to started new threads

Login required to post replies

Prev Next
Re: Can we quantify road bike risk? [MadTownTRI] [ In reply to ]
Quote | Reply
MadTownTRI wrote:
Dev said earlier to Dan in the thread that 1/1000 wasn't how the stats worked, though, so maybe he can clarify. To me, 1/1000 vs 1/40000 is enough to get my attention. It's basically someone in my section vs someone in the entire football stadium. Feels more proximate, though still very insignificant chance.

If you want to be more rigorous about it, I think this is correct. The odds of not dying are (39,999/40,000)^40, or .9990. Which is 1/1000, 0.1% chance of dying. Same result as adding the yearly chance. So like I said, for low odds like this you can just add them.

If the yearly odds of dying were 1/400, then it would be (399/400)^40, or .9047, 9.53% chance of dying vs 10% chance if you just did 40 x the yearly chance.
Quote Reply
Re: Can we quantify road bike risk? [Slowman] [ In reply to ]
Quote | Reply
Slowman wrote:
how? if you use general stats on bicycle fatalities, only 1 in 6 wore a helmet. does this sound like your cohort? so, what accident data are you referring to?


I don't believe that stat, but if you did, you could simply say that 5 out 6 fatalities do *not* have anything to do with our cohort. Which leaves a potential of a little over 100 fatalities per year in the US which do. And a large percentage of those get posted on ST.

It's tough to do what you are wanting, since no one is collecting the data at the basic level where it needs to happen. Which is why I just look at the agencies that publish WAGs on rider hours and miles along with the fatality stats. Even if you assume everyone is in your cohort, the odds are very low, like 1 per 5M hours. Actually lower than auto travel! IMO that should be good enough to satisfy anyone who has the perception that cycling is too dangerous, since no one gives a 2nd thought about riding in a car. And if you ride a lot your whole life it comes out to under 1% chance, out of a 100% chance you will die of something. You could go to a lot of effort to show that it is really a bit higher or lower for our cohort, but I don't know that anyone cares enough. Would be cool to know, but it's like a major research project and a lot of approximations will still need to be made.
Last edited by: rruff: Aug 30, 16 11:20
Quote Reply
Re: Can we quantify road bike risk? [Slowman] [ In reply to ]
Quote | Reply
Slowman wrote:
.

if you can help me figure out a way to broaden our cohort while not wrapping our arms around people on bicycles who are not really relevant to our style of riding, i'm all for it. open to suggestions.


Isn't it the same way that you'll find out about triathlete deaths? We read (or have read) about a cycling fatality, do some research online to find out if the cyclist was "like us" and let you know/add to database. I read today about a cyclist killed while cycling to work last week in the town where I grew up. Looks like he died from brain trauma. Happened early in the morning and could have been a hit and run. I found a couple of articles about the accident, found out via a search that he was registered for the current National Bike Challenge and had done a couple of 100+ mile days in the past two months. I doubt he has ever done a triathlon but I'm guessing he paid as much attention to safety as the average triathlete. I could see from photos that he had a mirror on his bike. It looks like he did most of his recreational riding on multiuse trails, so he might not have been wearing a helmet during his final ride. That's the only question I have at this point whether he was a cohort of mine, since I also commute to work via bike.
Last edited by: Mark Lemmon: Aug 30, 16 12:48
Quote Reply
Re: Can we quantify road bike risk? [cartsman] [ In reply to ]
Quote | Reply
Hello cartsman and All,

You bring up and interesting point .....

Cyclists that ride like triathletes (often not in a group) should be included in the database since their risk is not very different than a triathlete considering road position .... although a cyclist riding a road bike would seem to have more situational awareness (especially to the rear) and maneuverability than a triathlete in the aero bars .... and triathletes would be a less visible to a motorist by virtue of their position.

If we sent out 500 robot cycles that rode in the same fashion as triathletes they would be included also .... they could put in 24 hours days ..... reporting their position with near misses and collisions so we would soon have realtime streaming data.

Whether the robots wore helmets or not would only be useful if it indicated a certain style of riding .... as a vehicular cyclist riding in the track of the motorist's left wheel or as a cyclist riding far to the right of the travel lane .... or some other style of riding .... and also indicated a style of parsing intersections.

It seems likely that cyclists wearing helmets are more generally safety conscious than than those that do not wear helmets.

Similarly ..... studies indicate female cyclists are more safety conscious than male cyclists .... female mode share increasing with improved infrastructure and less perceived risk .... canaries in the coal mine.

While getting information on triathlete cycling risk would be desirable .....

....... it would seem more to the point to determine what countermeasures influenced risk reduction the most ..... so we could adopt the safest style of riding ..... regardless of risk because we are probably going to ride anyway ....


I doubt many triathletes will adapt to a changed style of riding (indoor vs. outdoor) very quickly based on a study of calculated risk assessment .... as opposed to changing their outdoor style of riding based using counter measures that increase safety and reduce perceived risk.

The recent use of high powered lights to improve cyclist visibility is a good example of a type of countermeasure to increase cyclist safety and reduce perceived risk.

https://pvcycling.wordpress.com/...28/lighten-up-dummy/

Cheers, Neal

+1 mph Faster
Quote Reply
Re: Can we quantify road bike risk? [rruff] [ In reply to ]
Quote | Reply
rruff wrote:
MadTownTRI wrote:
Dev said earlier to Dan in the thread that 1/1000 wasn't how the stats worked, though, so maybe he can clarify. To me, 1/1000 vs 1/40000 is enough to get my attention. It's basically someone in my section vs someone in the entire football stadium. Feels more proximate, though still very insignificant chance.


If you want to be more rigorous about it, I think this is correct. The odds of not dying are (39,999/40,000)^40, or .9990. Which is 1/1000, 0.1% chance of dying. Same result as adding the yearly chance. So like I said, for low odds like this you can just add them.

If the yearly odds of dying were 1/400, then it would be (399/400)^40, or .9047, 9.53% chance of dying vs 10% chance if you just did 40 x the yearly chance.

That's your cumulative odds for a 40 year period, but you make the assumption that there is correlation year over year to run a cumulative probability scenario....but there is no year over year correlation thus there is no cumulative probability to be calculated.

In a given year it's still 1/40000. If you went out last year and did not die, your odds at the start of this year at 1/40000. Did not die after 39 years? Well starting this year, your odds are 1/40000. Why? Because the guy who started riding this year and you riding 39 years without dying experience the same odds/shit out on the road TODAY. These odds have nothing to do with what happened over the last 39 years. Do you see the difference.

Let's put it another way. How does the texting driver about to hit a rider know if it is me, riding in my 35th year without dying, slowman riding his 40th year without dying or my newbie tri friend that I just put on a bike for the first time this year? If anything, I bet the 40 year rider has better odds than the newbie, in the sense that his spider senses are battlefield sharpened, but the average odds are the same.....they have nothing to do with what happened last year or what happened to the other guy down the street or the other state. It's a national level yearly view.
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
devashish_paul wrote:
That's your cumulative odds for a 40 year period, but you make the assumption that there is correlation year over year to run a cumulative probability scenario....but there is no year over year correlation thus there is no cumulative probability to be calculated.

I am assuming that the odds are 1/40,000 each year. Given that, what are the odds over a 40 year period? That's what we want to know.
Quote Reply
Re: Can we quantify road bike risk? [rruff] [ In reply to ]
Quote | Reply
rruff wrote:
devashish_paul wrote:
That's your cumulative odds for a 40 year period, but you make the assumption that there is correlation year over year to run a cumulative probability scenario....but there is no year over year correlation thus there is no cumulative probability to be calculated.


I am assuming that the odds are 1/40,000 each year. Given that, what are the odds over a 40 year period? That's what we want to know.

Who cares about the probability over 40 years? It is meaningless at this moment in time THIS YEAR. It does not keep going up as time moves on. As I said, slowman needs to know the risk on THIS ride this year year. Its the same for him and the newbie going out on his first ride today. What slowman did for the last 39 does not increase his risk today more than the newbie.
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
devashish_paul wrote:
rruff wrote:
devashish_paul wrote:
That's your cumulative odds for a 40 year period, but you make the assumption that there is correlation year over year to run a cumulative probability scenario....but there is no year over year correlation thus there is no cumulative probability to be calculated.


I am assuming that the odds are 1/40,000 each year. Given that, what are the odds over a 40 year period? That's what we want to know.


Who cares about the probability over 40 years? It is meaningless at this moment in time THIS YEAR. It does not keep going up as time moves on. As I said, slowman needs to know the risk on THIS ride this year year. Its the same for him and the newbie going out on his first ride today. What slowman did for the last 39 does not increase his risk today more than the newbie.

What are the odds that a single fair coin flip turns up heads?
What are the odds that 40 fair coin flips in a row turn up all heads?

What are the odds of not dying while riding a bike in a single year?
What are the odds of not dying while riding a bike 40 years in a row?

Bike riding in a given year, like a single fair coin flip, is an independent event, but one can still calculate the probability of a sequence of independent events. That is different than changing the odds from one year to the next based on each year's outcome.

----------------------------------
"Go yell at an M&M"
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
devashish_paul wrote:
Who cares about the probability over 40 years? It is meaningless at this moment in time THIS YEAR. It does not keep going up as time moves on.

Wut?! I want to know the odds that cycling will be the thing that kills me during my potential lifespan. The odds of me dying this year are a hell of a lot less than the odds of me dying during a 40 year period. Actually 40 times less. It seems like you are saying that the two are the same. Which is nuts.
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
Is there a difference in risk and odds?

Say I have a 1/400 risk of being bitten by a snark every time I swim in the ocean. It only makes sense that my risk on a particular day are only 1/400. But if I never get in the water my odds are zero. But, if I go every day my odds will go up. However, the risk is always the same?

Just trying to understand.
Quote Reply
Re: Can we quantify road bike risk? [nbaffaro] [ In reply to ]
Quote | Reply
nbaffaro wrote:
Is there a difference in risk and odds?

Say I have a 1/400 risk of being bitten by a snark every time I swim in the ocean. It only makes sense that my risk on a particular day are only 1/400. But if I never get in the water my odds are zero. But, if I go every day my odds will go up. However, the risk is always the same?

Just trying to understand.


How did your odds go up if you go every single day? They are 1/400 on any given day. You just expose yourself more times like playing Russian roulette which is 1/6 but you could spin the revolver and shoot at your head 10,000 times and actually live.

You could go one day and on that day it is 1/400....skip 398 days and it is 1/400, go the 398 days in between and it was 1/400 each day. You could go on of a million days and you did not increase the risk. It's 1/400 on the millionth day and even though you went out many multiples of 400, you magically did not get bitten. Some other poor suckers got bitten, but you did not, but one in 400 people are getting bitten on that trip to the ocean....you just got off the hook every time and every time you go to the ocean things get reset to zero so today it is 1/400 AGAIN, not 1/399, then 1/398, down to 1/5, 1/4,1/3, 1/2 to the last one where you number is up....that's the beauty....your number is never up. It gets reset on every outing.

You can take the risk to zero for sure. Never do the activity. But once you do it, it's 1/400 (or whatever that denominator is). If that was not the case, you absolutely would not want Gold card fliers on your airplane. Those guys flying 100K-250K miles per year, their presence MUST increase the chance of your plane crashing....and we know that is not the case!
Last edited by: devashish_paul: Aug 30, 16 18:05
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
let's take Russian roulette or rolling a die. Let say we want to roll a 1 with a 6 sided die. Roll one is 1/6, roll two is 1/6, roll three is 1/6 but as I continue to roll the die the chance that I will roll a 1 goes up. I'd bet a lot of money that you couldn't roll a die 100 times and not roll a 1. Is it possible sure. But, if you rolled 99 times and no one had appeared the chance of a one on 100th roll is still 1/6. I get that. The chance of rolling a specific number in a number of rolls is =1-(5/6)^n where n=number of rolls.
N=1 16%
N=2 30%
N=3 40%
N=4 50%
N=5 59%
N=6 65%
N=1000 99.99999999999999999999999999999999999999999999999999999999999999999999999 %(72 places)
N=10,000 100%

But, you seem to be saying that your exposure doesn't mean anything. That exposure has to mean something. I'm not saying it drastically effects the odds/chance/probability of something happening but it has to change it in some way.
Quote Reply
Re: Can we quantify road bike risk? [Slowman] [ In reply to ]
Quote | Reply
I think micro data can be particularly misleading. macro data is more meaningful I think.

I haven't seen for a while, but here in NZ the government publishes some data on cycling. As with any data collection, there will be errors but they collect data about cycling in the same-ish way they collect data about cars. I paraphrase below.

Number of registered cars on the road, distance travelled per year (note that in NZ we get a warrant of fitness check every year (used to be 6 months) to ensure our cars are safe. Odometer reading is one of the things that's checked, particularly for diesel vehicles that pay road user charges based on distance travelled). They get reasonably accurate number of cars on the road and a believable average distance per vehicle per year. They then use the reported deaths and injuries per year and hey presto, there's a number that represents your chance of death (or injury) per km travelled.

For bikes, they use bike sales per year, estimated bikes owned to get how many people have bikes. They have people with clipboards at various places around the city and country, counting bikes. Combined with online surveys re bike use, they get a reasonable idea about cycle use per year. Again, they look at reports of deaths and injuries each year and again, a risk per km is determined.

From memory, last I heard was about 8-10 years ago. While cycling has increased a lot in NZ since then, I suspect the rate of injury and death is something like the same. The last stats I saw showed that the accident and death rate was very close to identical for cars and bikes. Of course this is completely opposite to peoples perceptions (particularly the nervous mothers of precious children).

I remember showing the article to a friend. She had never ridden a bike as a child (she was legally blind) but thanks to an operation, saw well enough to take up cycling in her mid 40's. She eventually got good enough to ride the "Round Taupo" century ride with 12,000 other mad keen cyclists. She eventually gave up cycling because of her perception of the dangers. While she read the government statistics and agreed on their accuracy, she nodded and agreed right up to the time she said, "But it's still too dangerous to ride a bike". Even with clear evidence that cycling here was no more dangerous than driving her car, she still perceived the danger more on a bike. Insane in my book.

I think Dan, relying on anecdotal reporting, even from a fairly large group like ST, wouldn't be particularly valid.

TriDork

"Happiness is a myth. All you can hope for is to get laid once in a while, drunk once in a while and to eat chocolate every day"
Quote Reply
Re: Can we quantify road bike risk? [tridork] [ In reply to ]
Quote | Reply
"I think Dan, relying on anecdotal reporting, even from a fairly large group like ST, wouldn't be particularly valid."

i agree. and that's why we're not going to be doing that. and we never were going to be doing that. it was never contemplated.

Dan Empfield
aka Slowman
Quote Reply
Re: Can we quantify road bike risk? [nbaffaro] [ In reply to ]
Quote | Reply
This is what you guys are missing. We're not trying to calculate the mythical odds over a 40, 50 or 60 year cycling career of a cyclist. We're trying to figure out the risk today. In the vein, it's like rolling the dice today. It does not matter how many rolls happened over say the last 40 years. All we care about is if Dan does out to ride tomorrow, what are his odds tomorrow. There the same as your odds, my odds, the newbies odds, assuming we all did the same ride in the same place at the same time. The past history has zero impact on the current odds for tomorrow's ride. We're trying to get to the point of helping people decide, "Is my ride tomorrow or this weekend to train for USAT nationals or Kona safe, or do I take it to the trainer". We're not trying to help them decide the cumulative risk over 40 years. Just a single year. If you want to distill it down to the daily probability and then apply cumulative daily exposures, you'll actually find your odds get BETTER than the annual average since you are not riding every day. It's not going to get higher than the yearly average over an entire year, because by definition, almost everyone is riding less day in a year than there are days in a year.
Quote Reply
Re: Can we quantify road bike risk? [Slowman] [ In reply to ]
Quote | Reply
In that case Dan, getting hold of data that already exists, from each State, would be the first avenue to pursue. DMV?

Doing any new data collection is out of the question. The trick would be normalizing the information so valid comparisons from state to state could be made. I'd also suggest that having a baseline to compare with, would be good. Comparing death and injury rates with cars, would be the obvious choice Also, adding in seat belt and non-seatbelt information in cars would be good as well (if it exists).

Is cycling safer than driving in a car or more dangerous? (the higher speed of motoring may make driving more dangerous). Is cycling more dangerous that wearing a seatbelt but less dangerous that not wearing a seatbelt.

I remember years ago, when undergoing asbestos training, that the stats on the screen looked pretty scary and we were all mortified that anyone would go anywhere near asbestos based on the cancer rates. Then the instructor pointed out that even unprotected asbestos workers got cancer at lower rates than smokers, did we all realize it was actually pretty safe to deal with asbestos. Similar to comments I made in a previous response, perception is not a very good predictor of fact. But I guess that's why you're trying to collect the cycling data in the first place.

TriDork

"Happiness is a myth. All you can hope for is to get laid once in a while, drunk once in a while and to eat chocolate every day"
Quote Reply
Re: Can we quantify road bike risk? [devashish_paul] [ In reply to ]
Quote | Reply
agreed that each day is a new day and the odds/chance are not affected by the past or the future.

But, how can you not agree that the more you increase your exposure to risk the more likely it will happen.

Just as you said someone could pull the trigger 10,000 times in a game of life & death and never once die. Each of those 10,000 pulls of the trigger have the same 1/6 chance. But, when you look at it over time the chances of hitting the bullet approach 100%

The more you expose yourself to the odds the better the probability of them happening. Whether it's airplanes, Sharks, lottery, or dice. That is a fact you cannot refute.

How this data is compiled and used has no bearing on that.
Quote Reply
Re: Can we quantify road bike risk? [copperman] [ In reply to ]
Quote | Reply
copperman wrote:
windschatten wrote:
As long as this is not normalized to miles ridden, or time spent on a specific setting (urban/suburban/country) and road layout (w/o bike lane etc.pp), these data are not going to mean anything.

I actually encourage everybody to make their own assessment/quantification, by looking where they ride and when, take into account where chances are higher to get hit (or where people had been hit/killed).

Personal risk assessment is way more meaningful than calculating a generalized risk.

There are no guarantees in statistics, ever.


Slowman wrote:
"What is the metric you're trying to work toward? "

what is my chance, your chance, of getting killed while riding a bike. per year, per 5 years. per decade. if you ride for 20 years, 30, 40, 50. your risk over your lifetime, depending on what you consider a lifetime of riding, is that your chances of dying while riding your bike on the road is 1 in 8. or in 80. or 800. or 8000. or 80,000.

that's it.



Statistics are a much much much much much better way to evaluate risk than to "look around when you ride".

Source - worked on wall Street.

Can confirm.
Source: work at insurance company
Quote Reply
Re: Can we quantify road bike risk? [Slowman] [ In reply to ]
Quote | Reply
- collect statewide crash data and mine for cycling related accidents. Some states do not file a report if no injury occurred, so these may trend toward more sever accidents. Also assume that all bike/car accidents are under reported.
- collect census data of population by block or block group or place (some scale that is small enough to discern between high and low population density).
- many municipalities collect transportation mode share data via survey. You can use this to estimate the number of cycling trips made per year as well as the # of those that are recreational v. commute trips.
- Strava Metro data may be of use here as well since their data collection trends towards recreational trips. This can be used to gain a sense of miles travelled per year by location.
- aggregate the data by zip code, or some geographic reference, so that crash data can be compared against local population density.
- cross tab the data to figure out what the likelihood of an accident is based on location, population, bike mode share, accidents per bike trips.

There are probably a bunch of other steps that I am omitting. And as others have posted, this data will give you results that are intensely local in nature.

Jon Ryder
RPM Coaching
Quote Reply

Prev Next