1 in 100 GA Pilots Killed in an Airplane?

[201er]

Do we know how many pilots succumb to death in an airplane? With an average fatal accident rate of around 1 in 100,000, it comes down to how many hours a pilot spends in an airplane.

I’m guessing 1000 hours in a lifetime. 50 hours a year is pretty common in the GA community and I’m figuring 20 years of involvement. Does anyone have a better number?

1000 hours is 1% of 100,000. 1 in 100. The accident rate on a per mile or per flight basis isn’t too bad and a reasonable risk for infrequent passengers. But, what about the pilot? Are we really losing 1 in 100 friends, mooneyspacers, community members?

How many hours do Mooney pilots typically fly in a lifetime? How do our numbers compare to GA? Can we do better?

[Aerodon]

I used to think that the odds of dying in GA was about 1:10,000 of the participants and that it was similar to motorcycling, skiing, scuba diving etc.    But I think you are asking the correct question, if there are 500 pilots in a room, how many can expect to die in a year and how many will die in a 30 year flying career.

Initially your 1:100 sounds horrifying, but here is a website with some alternatives:

https://injuryfacts.nsc.org/all-injuries/preventable-death-overview/odds-of-dying/

First, the lifetime odds of all causes is 1:1.

Heart disease 1:6

Motor vehicles at 1:106

Motor cyclist 1:809.

So if your calculation of 1:100 is correct, I am happy with that as a risk compared to above.    I think your calculation is a bit flawed.  If there is a fatality rate of 0.75/100,000 hours, that will include passengers?  So if average fatal accident is 3 persons, then the pilot fatality rate is 0.25/100,000 hours.  If we all fly 50 hours a year (average), then there are 2000 pilots and .25 pilots die every year.  Or 1 every 4 years or 5 every 20 years.  5/2000 is 1 out of 400.  If your flying career is 40 years, then 10/2000 = 1 in 200.  

And then not to be morbid, I hate the saying 'he died what he enjoyed doing'.  But if I get 50 years of aviating and die 6 years earlier than I would have of heart disease, I would say thats less tragic than a 25 year old with 2 kids?

My mother's friend's son died on his first solo outing after getting his private licence at 18.  Thats tragic.

Aerodon

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[201er]

13 minutes ago, Aerodon said:

So if your calculation of 1:100 is correct, I am happy with that as a risk compared to above.    I think your calculation is a bit flawed.  If there is a fatality rate of 0.75/100,000 hours, that will include passengers?

It’s 0.75/100,000 fatal accidents, not fatalities. Although in some cases it could be the passenger dying and the pilot surviving, I would guess it’s much more frequent where the pilot is the sole occupant so a fatal accident would have to claim the pilot. I guess since a few fatal accidents claim passengers and not the pilot, maybe it’s 0.65/100,000 for pilots.

Also the 0.75 is on a good year. Over a decade it still averages out and rounds to 1/100,000.

[201er]

19 minutes ago, Aerodon said:

Initially your 1:100 sounds horrifying, but here is a website with some alternatives:

https://injuryfacts.nsc.org/all-injuries/preventable-death-overview/odds-of-dying/

First, the lifetime odds of all causes is 1:1.

Heart disease 1:6

Motor vehicles at 1:106

Motor cyclist 1:809.

So if your calculation of 1:100 is correct, I am happy with that as a risk compared to above.

Great, it turns out I’m more likely to kill myself than to kill myself in an airplane! I can sleep easier at night. 

Suicide

1 in 86

[jaylw314]

4 hours ago, 201er said:

Do we know how many pilots succumb to death in an airplane? With an average fatal accident rate of around 1 in 100,000, it comes down to how many hours a pilot spends in an airplane.

 

I’m guessing 1000 hours in a lifetime. 50 hours a year is pretty common in the GA community and I’m figuring 20 years of involvement. Does anyone have a better number?

1000 hours is 1% of 100,000. 1 in 100. The accident rate on a per mile or per flight basis isn’t too bad and a reasonable risk for infrequent passengers. But, what about the pilot? Are we really losing 1 in 100 friends, mooneyspacers, community members?

How many hours do Mooney pilots typically fly in a lifetime? How do our numbers compare to GA? Can we do better?

I can believe that, and I suspect the number of pilots that fail to survive is measurably higher than that.  There are fewer than 1000 pilots between all the pilots I know and all the pilots they know locally, yet the number of deaths I've come to hear about locally is more than just 10.

[takair]

These numbers are quite sobering and depressing.  Maybe doesn’t change the outcome too much, but do they double dip for heart disease or suicide causing a crash?  

[aviatoreb]

9 hours ago, 201er said:

Do we know how many pilots succumb to death in an airplane? With an average fatal accident rate of around 1 in 100,000, it comes down to how many hours a pilot spends in an airplane.

I’m guessing 1000 hours in a lifetime. 50 hours a year is pretty common in the GA community and I’m figuring 20 years of involvement. Does anyone have a better number?

1000 hours is 1% of 100,000. 1 in 100. The accident rate on a per mile or per flight basis isn’t too bad and a reasonable risk for infrequent passengers. But, what about the pilot? Are we really losing 1 in 100 friends, mooneyspacers, community members?

How many hours do Mooney pilots typically fly in a lifetime? How do our numbers compare to GA? Can we do better?

Your mathematics is not correct.

The data stated is a death rate of p=1 per 100,000 hrs, so take p=.00001

The probabilities don't add lilac that - Just to point out your error.  Suppose a person flies 200,000 hrs in their lifetime.  Then your calculation would say they have a 200% chance of dying.  Or more sobering - if they fly 99,999 hrs then since 100,000 is a 100% certainty of death, then they will die in the next hour.  There are a few pilots running around with 50,000 hrs.

The proper way of dealing with a specific (but dubious) assumption that p=0.00001 and they are independent and independently distributed (each hour has the same probability - they don't change over time - and also those probabilities  don't effect each other), is

P(death in n hours of flying)=1-(1-p)^n, which says 1-p is the probability you don't die in 1 hour.  (1-p)^n is you don't die in n hours.  1-(1-p)^n is you do die in n hours.  (but this isn't right either since a) clearly probabilities are changing over time, like maybe you lower your probability in time due to experience?  Or maybe you become more bold?  Or maybe you eventually become an old and crappy pilot who should hang it up?  ...ie challenging the independence and identical concept - AND what we really want is the probability that you might die in any one of the previous hours, so some kind of sum should be involve, but lets skip this since the complication is not important here - what I stated is the first thing I said, 1-(1-p)^n.

1-(1-.00001)^1000=0.009950215753607. which is very close to 0.01 but it is difference, and the difference grows with n.

1-(1-.00001)^10000=0.095163034385241 (so that is not 0.1)

1-(1-.00001)^100000=0.632122398231753 (so that is not 1=100%).

1-(1-.00001)^200000=0.864666070117244 (so that is 86% and not 200%).

But then I very much also doubt the independence.  Plus that 1.e-5 initial number is across the pilot population.  I very much doubt it is uniform - that said we do not know which kind we are individually.

That is a critique of the math only.  The message is correct.  Flying is dangerous.  However, we are a certain kind of people. I argue that many of us may well have taken up another dangerous activity if not flying - with comparable or worse stats.  E.g. sailing, canoeing (yes stats are worse than flying), bicycle riding, or couch sitting (I read once but lost the stats of those who sit on the couch more than a certain number of hours per day - like 8 - are horrific due to diseases of sloth).

So what is really relevant - is how do these stats for flying compare to something else we might be doing instead.

I ride bicycles. knock on wood.  Its worse than flying.

 

 

 

[aviatoreb]

5 hours ago, jaylw314 said:

I can believe that, and I suspect the number of pilots that fail to survive is measurably higher than that.  There are fewer than 1000 pilots between all the pilots I know and all the pilots they know locally, yet the number of deaths I've come to hear about locally is more than just 10.

I am going to guess the following.

That 10 were not from the 1000 you know (maybe some were?) but rather they were including the people that the people you knew knew and thus the information came to you?  That broader two step network can easily be 10 times 1000.  (made up number - but true mechanism).

[201er]

13 minutes ago, aviatoreb said:

Your mathematics is not correct.

So skipping all the who-ha, roughly where does this put us? 1 in 50? 1 in 1000? Out of a large gathering of ga pilots, say Oshkosh, roughly how many are going to die “doing what they love” (flying, not canoeing) at some point?

Edited January 9, 2021 by 201er

[201er]

12 minutes ago, aviatoreb said:

I am going to guess the following.

That 10 were not from the 1000 you know (maybe some were?) but rather they were including the people that the people you knew knew and thus the information came to you?  That broader two step network can easily be 10 times 1000.  (made up number - but true mechanism).

At a safety presentation by John and Martha King at Air Venture, they asked how many in the audience personally knew someone who died piloting an airplane and a sobering chunk raised their hand. Then they asked how many knew someone who knew someone who died in an airplane crash and everyone raised their hand. What was missing was how many of the attendees in that tent will have someone raising a hand for at some point and that’s what I’m asking about here.

[aviatoreb]

5 minutes ago, 201er said:

So skipping all the who-ha, roughly where does this put us? 1 in 50? 1 in 1000? Out of a large gathering of ga pilots, say Oshkosh, roughly how many are going to die “doing what they love” (flying, not canoeing) at some point?

It puts us in the true and sobering position that there is a legit and significant risk to flying and it should be treated that way.

I was just describing the details of the math were quite different from what you said.  The math you asserted would describe that 95% and 95% would make aa 190% chance of occurence, but rather under i.i.d. assumption, 1-(1-.95)^2=0.997500000000000-> 99.75%.

Also it must be contrasted to other activities we might do instead to get a true idea of risk to self.

I don't have the actual data to make a proper assessment.  But if I could ask it, I would not want to work with the risk per hour, but risk per population.  I would want a population study of death rates per pilot across years of activity.  E.g. give me the distribution of deaths for 30 year active pilots.  31 year active pilots.  32 year active pilots.  I could do something with that.  I am certain that the insurance industry is doing exactly that.  Surely the ntsb has someone also doing that.  In any case, I am sure the numbers are sobering.

 

[aviatoreb]

16 minutes ago, 201er said:

At a safety presentation by John and Martha King at Air Venture, they asked how many in the audience personally knew someone who died piloting an airplane and a sobering chunk raised their hand. Then they asked how many knew someone who knew someone who died in an airplane crash and everyone raised their hand. What was missing was how many of the attendees in that tent will have someone raising a hand for at some point and that’s what I’m asking about here.

This is very interesting and exactly on topic of what I was saying.  The problem is to making something empirical out of it, would take a controlled experiment.  I don't know if those people were telling us if they knew someone who knew someone who died.  Or if actually they also were saying they knew someone who knew someone who knew someone since psychologically this seems the same.

Why is this important the distinction I am making?

If you ask people at your home airport, then you likely know a lot of the same people in your same circle (this is called assortatiivity coefficient in network science).

But go to Oshkosh and people come from all likes.  I bet assortativity in the crowd is very low.  Ask them if they know someone who knows someone that *&&*& and I bet many will answer that a yes inadvertently even if it is 3 steps instead of 2?  In any case, I claim that by the time you go past 2 steps, you are actually picking up a large fraction of the entire pilot population.  So for sure I would expect most hands to go up.

This is not as informative as it seems even if it is psychologically a strong thing.

By the way, who has heard of this phenomenon - the six degrees of separation as it is called.  https://en.wikipedia.org/wiki/Six_degrees_of_separation. it is actually about the diameter of a network, and the idea of minimal spanning trees.  I know someone who knows someone who knows someone who met the queen of England.  

Its also very relevant right now in the moment of a pandemic!  How many interactions does it take for the UK variant to go from UK to my small town?

I know one person directly who died in aviation.  He was a 15,000 hr pilot.  We all know a person who died, who was a very energetic and terrific kid, Patrick I feel I knew him too even though I never met him.

Every year a few crashes are posted here that were Mooneys.  How many Mooneys are there? 10,000 or something in the usa?

I have known personally several people who died in car crashes.  Some who have died in snowmobile accidents, hunting accidents, bicycle accidents, homicidal shootings, sailing, and other strange ways if we expand this to people I know who know.  Or people I know who know who know.

[toto]

6 minutes ago, aviatoreb said:

In any case, I am sure the numbers are sobering.

 

Given the apparently disproportionate number of fatal accidents resulting from CFIT or continued VFR into IMC, I suspect that the nature of a pilot's flying activities would be super relevant to an actuary.  A low-experience pilot who has high personal minimums and a flexible schedule seems far less likely to be involved in a fatal accident, simply because s/he avoids the scenarios that produce fatalities.  If you're just flying a local tour with severe clear and no wind, you might not be gaining much experience, but you're also avoiding a few of the most popular ways to die in an airplane.

The chart showing 100,000 hours of CAVU local tours vs 100,000 hours of hard IMC in a single-engine piston would look like a chart comparing Scrabble to BASE jumping.

As carusoam would say.. pilot armchair thoughts only, not a statistician ..

[MooneyMitch]

Such a wonderful inspirational thread we’re having! 

One thing for sure, we all know everyone on MS is 100% guaranteed of dying of something........

Me? I think I’ll take a little ride on my motorcycle down to the airport and take a little flight in my Mooney.......it’s a risky day to be alive . 

Oh, yeah, don’t let your wife read this thread ! 

[aviatoreb]

12 minutes ago, toto said:

Given the apparently disproportionate number of fatal accidents resulting from CFIT or continued VFR into IMC, I suspect that the nature of a pilot's flying activities would be super relevant to an actuary.  A low-experience pilot who has high personal minimums and a flexible schedule seems far less likely to be involved in a fatal accident, simply because s/he avoids the scenarios that produce fatalities.  If you're just flying a local tour with severe clear and no wind, you might not be gaining much experience, but you're also avoiding a few of the most popular ways to die in an airplane.

The chart showing 100,000 hours of CAVU local tours vs 100,000 hours of hard IMC in a single-engine piston would look like a chart comparing Scrabble to BASE jumping.

As carusoam would say.. pilot armchair thoughts only, not a statistician ..

I agree completely and what I was trying to allude to that probability is not identically distributed but changes over time and it is different for different people.

I presume that the insurance industry at least has a lot of conditionals when describing conditional probabilities.  Risk of accident given the person has a record of traffic violations.  Risk of accident given a person has a previous accident.  Risk of accident given a person has a super high performance experimental but no ifr cert.  written P(accident | super high performance experimental, no ifr cert).

We do our best to lower our own risk.  But psychologically we think we are all above average.

In Lake Wobegone, all the children are above average.  (Garrison Keillor).

One way to behave- there is only one hour that matters.  It is the next hour.  In about 2.5 hours I have a flight planned, 45 minutes, KPTD->KIUA. Wish me luck!  I am already doing something to hopefully improve my chances.  I am doing it today since it is gorgeous today but earlier in the week when I was thinking about it, it was not so nice in various ways - icing mostly.  And I will fly higher than the most fuel efficient route because I like altitude for a backup plan.  Knock on wood friends and wish me luck.

Actually, and perhaps irrationally, I am balancing another risk (as are we all) but that scares me more (again perhaps irrationally).  There is a pandemic going on and it really creeps me out.  What is the risk of catching and dying of covid during my activity today - dropping off my plane at an avionics shop?  Driving home.  Probably very low.  But anyway that is running through my head too (no politics please - just revealing my inner thinking regarding risk management). 

[aviatoreb]

There is a phrase I like, which is "artificial precision" - 67.389% of all statistics are made up.

Which is fun - but they are used a lot, and false and erroneous computations when making real world decisions.  By regular folks, and also people in real positions of importance, at heads of companies, engineering positions, political positions.  I am all for data driven decision making - obviously - consider what I do for a living - but bad data and approximate computations can be used carefully and usefully.  Sometimes bad data and bad computations are used dangerously to make really bad decisions.  I am just saying that in general terms and given specific and uncomfortable positions I have been pushed toward professionally over the years.

[201er]

37 minutes ago, aviatoreb said:

This is very interesting and exactly on topic of what I was saying.  The problem is to making something empirical out of it, would take a controlled experiment.  I don't know if those people were telling us if they knew someone who knew someone who died.  Or if actually they also were saying they knew someone who knew someone who knew someone since psychologically this seems the same.

Ok, let's see where our community stands.

 

[MooneyMitch]

4 minutes ago, aviatoreb said:

I agree completely and what I was trying to allude to that probability is not identically distributed but changes over time and it is different for different people.

I presume that the insurance industry at least has a lot of conditionals when describing conditional probabilities.  Risk of accident given the person has a record of traffic violations.  Risk of accident given a person has a previous accident.  Risk of accident given a person has a super high performance experimental but no ifr cert.  written P(accident | super high performance experimental, no ifr cert).

We do our best to lower our own risk.  But psychologically we think we are all above average.

In Lake Wobegone, all the children are above average.  (Garrison Keillor).

One way to behave- there is only one hour that matters.  It is the next hour.  In about 2.5 hours I have a flight planned, 45 minutes, KPTD->KIUA. Wish me luck!  I am already doing something to hopefully improve my chances.  I am doing it today since it is gorgeous today but earlier in the week when I was thinking about it, it was not so nice in various ways - icing mostly.  And I will fly higher than the most fuel efficient route because I like altitude for a backup plan.  Knock on wood friends and wish me luck.

Actually, and perhaps irrationally, I am balancing another risk (as are we all) but that scares me more (again perhaps irrationally).  There is a pandemic going on and it really creeps me out.  What is the risk of catching and dying of covid during my activity today - dropping off my plane at an avionics shop?  Driving home.  Probably very low.  But anyway that is running through my head too (no politics please - just revealing my inner thinking regarding risk management). 

I like this !! Have a great day living everyone !

[aviatoreb]

Just now, MooneyMitch said:

I like this !! Have a great day living everyone !

Thank you!  I am dropping off for install of a pair of GI275's about which I am very excited!

[toto]

Just now, aviatoreb said:

Thank you!  I am dropping off for install of a pair of GI275's about which I am very excited!

Btw, looks like this thread put you over 10,000 posts.  Cheers

[aviatoreb]

3 minutes ago, 201er said:

Ok, let's see where our community stands.

 

Thank you for the thoughtful threads 201er.   This is a topic I have thought a lot about.

Actually there was a moment about 15 years ago when a person I knew, knew a person who died in a bicycle accident.  I have been an avid bicycle rider for about 35 years.  It really gave me pause since it was a horrific accident, including paralysis, survival and a six month period of suffering before he slipped away.  It really gave me pause and I decided as a math professor I know how to do this - and I dug in and started looking for as much publicly available data as I could find on the topic of risks and modality of accidents.  Shortly thereafter I also dug into all the other kinds of activities I do.  All this stuff rolls through my head sometimes.

It is an important topic, and super we are talking about it and by considering it in detail I do believe we can each make ourselves individually even a little bit better.

One point - these statistics are population statistics.  But we are individuals.  We want to deviate from the population in a good way - some deviate from the population statistics in a bad way.  Be like Garrison Keillor says, all the children are above average.

[aviatoreb]

1 minute ago, toto said:

Btw, looks like this thread put you over 10,000 posts.  Cheers

Wohoo!!  I had not realized!  Wow - and What a good and proper thread to have earned me that position.  A thread sort of a bit about math - as related to aviation risk.  Something I think a lot about.

[201er]

1 minute ago, aviatoreb said:

Wohoo!!  I had not realized!  Wow - and What a good and proper thread to have earned me that position.  A thread sort of a bit about math - as related to aviation risk.  Something I think a lot about.

Is it time to put the old horse out to pasture after 10,000 posts?

[aviatoreb]

15 minutes ago, 201er said:

Is it time to put the old horse out to pasture after 10,000 posts?

Oh yeah - horses are dangerous too - especially old ones!!

[MooneyMitch]

20 minutes ago, aviatoreb said:

Oh yeah - horses are dangerous too - especially old ones!!

Yep, anything that large and powerful, with a brain the size of a peanut frightens me !!