What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

accident rates

iaw4

Well Known Member
I wanted to share a basic observation. I hope I have my numbers right, but I would be happy(ier) to be corrected.

About 10,000 Vans airplanes are flying these days. Looking at the 2015 and 2016 accident statistics at the AOPA data base, about 10 Vans airplanes have fatal accidents every year. That is, a 1-in-1000 death rate. For perspective,
this is about 10 times the death rate from opiods in the general population. Over 10 years of flying, a 1% chance of dying in one's airplane is nothing to sneeze it.

About half of Vans accidents are explainable as stupidity---some ******* (apologies) deciding to do aerobatics at 50' above altitude. Importantly, I can easily avoid such accidents myself. Surprisingly, only about 1 in 10 was IMC related. About 4-5 seem to have no clear cause(s). These are most worrying to me, because I do not know how to avoid them.

It would be great to see a better and less casual analysis of Vans serious accidents.
 
Wouldn't 1-in-1000 be a 0.1% death rate?

If you avoid the type of flying that adds a bit more risk such as low altitude aerobatics, formation flight, IMC, etc. and avoid risky behavior such as flathatting/barnstorming and flying outside of your skill limits your risk goes down significantly. Then you only need to worry about the risk associated with mechanical issues, which can be mitigated by following sound maintenance practices.

Another thought to add... aviation mishaps are usually discussed in mishaps or deaths per flight hours, so you really should compare other aircraft's records based on this criteria.
 
I think he got the 1% by saying at .1% a year over 10 years becomes 1%...

pretty sure that's not how statistics work though...
 
by type

What does the data show by type? Are they all the same or does one type lend itself to being less accident prone?

I wanted to share a basic observation. I hope I have my numbers right, but I would be happy(ier) to be corrected.

About 10,000 Vans airplanes are flying these days. Looking at the 2015 and 2016 accident statistics at the AOPA data base, about 10 Vans airplanes have fatal accidents every year. That is, a 1-in-1000 death rate. For perspective,
this is about 10 times the death rate from opiods in the general population. Over 10 years of flying, a 1% chance of dying in one's airplane is nothing to sneeze it.

About half of Vans accidents are explainable as stupidity---some ******* (apologies) deciding to do aerobatics at 50' above altitude. Importantly, I can easily avoid such accidents myself. Surprisingly, only about 1 in 10 was IMC related. About 4-5 seem to have no clear cause(s). These are most worrying to me, because I do not know how to avoid them.

It would be great to see a better and less casual analysis of Vans serious accidents.
 
I think he got the 1% by saying at .1% a year over 10 years becomes 1%...

pretty sure that's not how statistics work though...

Yes, that's not the absolutely correct answer. But it's very close. The correct answer is 0.9955% (to 4 digits) chance of dying over 10 years.
 
Yes, that's not the absolutely correct answer. But it's very close. The correct answer is 0.9955% (to 4 digits) chance of dying over 10 years.

So.. you're saying I have 2x chance of being in an accident this year then I did last year?

Meaning if you could fly for 100 years, you have 100% chance of being in an accident?

That's not how probability works. Each Flight/hour/year is it's own entity. Not added to the previous.
 
Ron Wantajja wrote a great series of articles in Kitplanes http dealt in detail with acccident statistics in the E-AB world back about five years ago. If you're a subscriber, you can search the archives for his name and find them. Fortunately, I've talked Ron in to updating his series with newer statistics and insights, so watch for those articles in the coming year!

Paul
 
So.. you're saying I have 2x chance of being in an accident this year then I did last year?

Meaning if you could fly for 100 years, you have 100% chance of being in an accident?

That's not how probability works. Each Flight/hour/year is it's own entity. Not added to the previous.

The exact formula is 1 - (1-ann-death-rate)^years . For small death rates and years, this is about ann-death-rate*years. Over 10 years, it is roughly (just under) 1%, indeed, but not exactly so. If you fly for 1,000 years, you have a 100% chance of being dead, because no one has managed to live for more than 120 years. If you fly for 100 years, it's probably pretty close to 100%, too. ;-). Incidentally, I fudged more by ignoring the change in the number of flying airplanes. And no one knows how many hours the Vans have flown---it would have been a much better number with this information.

my numbers came from https://www.aopa.org/asf/ntsb/search_ntsb.cfm . Choose: Make=Vans. Injury=FATAL; Build=Home Built.

We have had 26 fatal accidents, starting in 2015. 14 were in As (incl 12 and 10). 12 in tailwheels. Sorting by type: 2 in RV-12s, 2 in RV-10s, 3 in RV9As, 3 in RV-8s, 7 in RV-7s, 2 in RV-6s, 4 in RV-3/4s. [one was misclassified under Vans Company, rather than Vans.] If someone knows the number of flying aircrafts by type, we can see if some seemed less accident-prone than others.
 
man.. I need to go play the lottery every day. :)

What you're saying is every time I play the lottery, I have a higher chance of winning!! love this...

Lots of things can go wrong in many situations - flying being one. Every time you go up you have the same chance of not coming back as the previous time. Not less of a chance.
 
So.. you're saying I have 2x chance of being in an accident this year then I did last year?

Meaning if you could fly for 100 years, you have 100% chance of being in an accident?

That's not how probability works. Each Flight/hour/year is it's own entity. Not added to the previous.
This is where I introduce Monty Hall and derail this thread :D
 
man.. I need to go play the lottery every day. :)

What you're saying is every time I play the lottery, I have a higher chance of winning!! love this...

The rules of statistics involve strict logic, which sometimes seem odd. One rule is ?past results don?t matter?, since they are a certainty, having already happened. So if you decide to play the lottery next week, if you play twice, your odds of winning at least once are twice as high compared to the scenario where you plan to play just once. If you lose the first drawing, your odds on the second drawing are the same as everyone else?s who bought one ticket. If you win on the first drawing, your odds on the second drawing are also still the same as the other players! So, from a strictly statistical point of view, it doesn?t matter if you had zero, or ten, accidents last year; your odds next year are the same. But real people, not statistics, cause accidents. Real people who have ten accidents are highly likely to be risk takers, and thus more likely to have future accidents. You can only take statistics so far...
 
The rules of statistics involve strict logic, which sometimes seem odd. One rule is ?past results don?t matter?, since they are a certainty, having already happened. So if you decide to play the lottery next week, if you play twice, your odds of winning at least once are twice as high compared to the scenario where you plan to play just once. If you lose the first drawing, your odds on the second drawing are the same as everyone else?s who bought one ticket. If you win on the first drawing, your odds on the second drawing are also still the same as the other players! So, from a strictly statistical point of view, it doesn?t matter if you had zero, or ten, accidents last year; your odds next year are the same. But real people, not statistics, cause accidents. Real people who have ten accidents are highly likely to be risk takers, and thus more likely to have future accidents. You can only take statistics so far...
if you had ten accidents last year, your odds of having another this year are pretty high. LOL
 
...
my numbers came from https://www.aopa.org/asf/ntsb/search_ntsb.cfm . Choose: Make=Vans. Injury=FATAL; Build=Home Built.

We have had 26 fatal accidents, starting in 2015. 14 were in As (incl 12 and 10). 12 in tailwheels. Sorting by type: 2 in RV-12s, 2 in RV-10s, 3 in RV9As, 3 in RV-8s, 7 in RV-7s, 2 in RV-6s, 4 in RV-3/4s. [one was misclassified under Vans Company, rather than Vans.] If someone knows the number of flying aircrafts by type, we can see if some seemed less accident-prone than others.
Perfect, since no RV-9's were listed, I have zero % chance of killing myself this year.

Either that or I have 100% chance of killing myself in my -9 this year. I'm not sure which.

Here is another way to look at such things.
 
if you had ten accidents last year, your odds of having another this year are pretty high. LOL

Good point! LOL

As Insurance companies see it - the longer you go without an accident, the less likely you are of having one (or the more hours you have I guess)... So they lower your rate.

With the other poster's logic, the longer you go without an accident, the higher your insurance rates should go!!
 
Humm

With the -6 being a scratch built and the -7 being prepunched among other things, I would have expected the -6 to have more occurrences over the -7 especially given the fact that there are almost twice more -6 completed than the -7. Interesting. (thinking this over, I admit that my own null hypothesis assumption is that the newer -7 designs have more consistent mfg processes over the old plans built -6 processes and thus more safer. However, the data seems to reject that assumption).

That being said, what the heck is going on with the -7? Are -7 pilots inherently more prone to die in their craft or is the design more prone to be "less forgiving", or is it a combination or interaction of the two?



The exact formula is 1 - (1-ann-death-rate)^years . For small death rates and years, this is about ann-death-rate*years. Over 10 years, it is roughly (just under) 1%, indeed, but not exactly so. If you fly for 1,000 years, you have a 100% chance of being dead, because no one has managed to live for more than 120 years. If you fly for 100 years, it's probably pretty close to 100%, too. ;-). Incidentally, I fudged more by ignoring the change in the number of flying airplanes. And no one knows how many hours the Vans have flown---it would have been a much better number with this information.

my numbers came from https://www.aopa.org/asf/ntsb/search_ntsb.cfm . Choose: Make=Vans. Injury=FATAL; Build=Home Built.

We have had 26 fatal accidents, starting in 2015. 14 were in As (incl 12 and 10). 12 in tailwheels. Sorting by type: 2 in RV-12s, 2 in RV-10s, 3 in RV9As, 3 in RV-8s, 7 in RV-7s, 2 in RV-6s, 4 in RV-3/4s. [one was misclassified under Vans Company, rather than Vans.] If someone knows the number of flying aircrafts by type, we can see if some seemed less accident-prone than others.
 
Interesting foot-note about the lottery; your odds don't improve a lot just because you buy a ticket!
 
I have two left hands, so I could never build an RV, but I am pretty good at statistical analysis (which I do for a living).

yes, what I quoted had many assumptions in them that are not likely to hold. nevertheless, they give useful summary information.

I had no idea how many RV-7s are flying, so maybe their accident rates is lower if most airplanes are 7s. Pilots and aircraft are heterogeneous, too. yes, conditional on pilot, the accident rate per hour may be similar (also a simplification; it is higher in periods with with bad weather, for example); but we do not have good measures of pilot heterogeneity. insurance companies pay a lot of money for data that helps them in modeling expected accident rates and payouts, given whatever information they can glimpse. they ask you for information like pilot hours, pilot age, etc., which all can be used to help. none of this info could flow into the aggregate summary statistics I quoted. with a lot more information, I could even calculate a customized personalized expected death rates for your next flight. but I don't have a lot more information.

the 1 in 1000 death rate per year may NOT apply to you. I sure hope it does not apply to me, or I would not fly. I hope it is no more than 1 in 10,000 for me IF I am a careful pilot---as careful as can be. but I admit that it worries me that I cannot judge my own probability, because I cannot judge how many of the 25 accidents in the last 3 years had no pilot error component in them.

advertising the so-named unconditional probability of death could help increase the caution in pilots. I was not aware of how bad it really was.

/iaw
 
Link to flying

Ivo,
Here is the link to how many are flying by type:

https://www.vansaircraft.com/public/first-flights.htm



I have two left hands, so I could never build an RV, but I am pretty good at statistical analysis (which I do for a living).

yes, what I quoted had many assumptions in them that are not likely to hold. nevertheless, they give useful summary information.

I had no idea how many RV-7s are flying, so maybe their accident rates is lower if most airplanes are 7s. Pilots and aircraft are heterogeneous, too. yes, conditional on pilot, the accident rate per hour may be similar (also a simplification; it is higher in periods with with bad weather, for example); but we do not have good measures of pilot heterogeneity. insurance companies pay a lot of money for data that helps them in modeling expected accident rates and payouts, given whatever information they can glimpse. they ask you for information like pilot hours, pilot age, etc., which all can be used to help. none of this info could flow into the aggregate summary statistics I quoted. with a lot more information, I could even calculate a customized personalized expected death rates for your next flight. but I don't have a lot more information.

the 1 in 1000 death rate per year may NOT apply to you. I sure hope it does not apply to me, or I would not fly. I hope it is no more than 1 in 10,000 for me IF I am a careful pilot---as careful as can be. but I admit that it worries me that I cannot judge my own probability, because I cannot judge how many of the 25 accidents in the last 3 years had no pilot error component in them.

advertising the so-named unconditional probability of death could help increase the caution in pilots. I was not aware of how bad it really was.

/iaw
 
Code:
RV-3/4      1,711	4	0.23%
RV-6/6A     2,631	2	0.08%
RV-7/7A     1,671	7	0.42%
RV-8/8A     1,448	3	0.21%
RV-9/9A     1,076	3	0.28%
RV-10       ,877	2	0.23%
RV-12       ,577	2	0.35%

these are 3-year accident numbers divided by the last count of airplanes flying. to the extent that relatively more of the latter models have come on line later (and fewer of the earlier), you may want to tilt it up in the latter years and down in the earlier years.

but just looking at the numbers, the only difference that seems meaningful is between 6's and 7's, and because these are essentially the same airplane (and if anything the 7 should be a little more consistent and safer), the rates are all reasonably similar. think a little under 1% per year. if I had to guess as to why the 7s look worse, it would be because the buyer profile could be a little different, maybe younger and more aggressive?! who knows.

it would make the analysis more interestingly if we would know the NTSB verdict on each of the 25 accidents, too.

/iaw
 
Last edited:
With the -6 being a scratch built and the -7 being prepunched among other things, I would have expected the -6 to have more occurrences over the -7 especially given the fact that there are almost twice more -6 completed than the -7. Interesting. (thinking this over, I admit that my own null hypothesis assumption is that the newer -7 designs have more consistent mfg processes over the old plans built -6 processes and thus more safer. However, the data seems to reject that assumption).

The core airframe (and its construction quality) rarely have much influence in accident causes.
It is the systems (primarily engine and fuel) that have the primary influence. Mistakes during installation or bad design decisions of systems have an equal potential on both models to cause an accident.
 
Bathtub curve - ?

Good point! LOL

As Insurance companies see it - the longer you go without an accident, the less likely you are of having one (or the more hours you have I guess)... So they lower your rate.

With the other poster's logic, the longer you go without an accident, the higher your insurance rates should go!!

In electronics we see a 'bathtub curve' of failures in the components, with 'infant mortality' in the first 90 days, and then a relatively long productive life, ending in a slowly increasing mortality in the latter portion of the MTBF (Mean Time Between Failure). This results in weeding out the failure prone in the earliest period and is the basis for attractive longer warranty offerings to the surviving systems with relatively incidental costs to the manufacturer.

Maybe the bulk of the -6's were built years ago and the failure prone have faded from the stats? Or maybe the owners/pilot communities of the earlier models are now the most experienced segment of the sample? (Old pilots, not bold pilots?).

Trying to quantify life based on an inadequate number of known factors has always lead me to conclude it's not about the formula, it's mostly about the quality of the data.
 
The core airframe (and its construction quality) rarely have much influence in accident causes.
It is the systems (primarily engine and fuel) that have the primary influence. Mistakes during installation or bad design decisions of systems have an equal potential on both models to cause an accident.

Maybe true, for most, but what about all those eye witnesses that describe parts of the aircraft falling off separately from the aircraft before the crash?
 
Maybe true, for most, but what about all those eye witnesses that describe parts of the aircraft falling off separately from the aircraft before the crash?

What aircraft are you referring to? Parts "fall off" of aircraft for for reasons other than improper installation or other air frame faults . Several causes are: 1. Over-stress (exceeding G limits or load limits). 2. Severe turbulence encountered above Vno. 3. Flying into a thunderstorm. 4. Exceeding Vne and encountering flutter...

Skylor
RV-8
 
What aircraft are you referring to? Parts "fall off" of aircraft for for reasons other than improper installation or other air frame faults . Several causes are: 1. Over-stress (exceeding G limits or load limits). 2. Severe turbulence encountered above Vno. 3. Flying into a thunderstorm. 4. Exceeding Vne and encountering flutter...

Skylor
RV-8
so the question is why that happens at a higher percentage on the 7 vs 6. In addition to a larger number of 6s in circulation, they have probably logged more hours on average too.
 
Maybe true, for most, but what about all those eye witnesses that describe parts of the aircraft falling off separately from the aircraft before the crash?

I am not sure what your question is?

Are you implying that as long as all of the structure on a kit plane was built exactly per the plans, that there should never be any situation or condition that could cause something to break?

That would mean, it was indestructible?
 
I am not sure what your question is?

Are you implying that as long as all of the structure on a kit plane was built exactly per the plans, that there should never be any situation or condition that could cause something to break?

That would mean, it was indestructible?
The question is, for 2 planes that are very similar (6 and 7), rate per unit flying is 5x. rate per hours flown is probably even larger (assuming the average 6 has accumulated more hours than the average 7). Explanation may have nothing to do with the actual plane - curious what would explain it.
 
The question is, for 2 planes that are very similar (6 and 7), rate per unit flying is 5x. rate per hours flown is probably even larger (assuming the average 6 has accumulated more hours than the average 7). Explanation may have nothing to do with the actual plane - curious what would explain it.

I understand that is the question some are asking.

That is not the question that he asked.....
 
Since the accident statistics originally posted were from a recent one year span, one could assume that the number of RV6 "first flights" (if any) were nowhere near the number of RV7 "first flights" during this period. This is the most failure prone period of an experimental aircraft's existence.
Already mentioned was a parallel factor, that most RV6 pilots probably have more time in type than the typical RV7 pilot.
Another factor that could be related; are more RV7s flown in IMC than the typical RV6 (3,4,8?). We'd need to know the determined cause of the accidents before anything meaningful could be derived from the stats.
I'd be interested to see THAT data, for sure!
 
I stand corrected, I was pointing out your statement on post 22 specifically the word rarely the cause, but it is those few that get my attention more than the out of gas situation, we know what caused those planes to come down.
 
Last edited:
uhmmm...I think you guys are over-interpreting the statistics. even over the 3 years, we are talking about only a handful of deaths when we want to distinguish between the 6 and the 7. one more or one fewer deadly crash is mostly noise.

I have been looking at the NTSB data base, but they only have 7 fatal accidents, 15 when I look for vans and rv, out of which about half are foreign. they do not contain the synopses (which I have read before as something like "primary cause was alcohol, secondary cause was pilot error" or the like), and which I could use to classify accidents. so, learning what the fatal mistakes were for our Vans pilot friends remains a dark mystery. it would be more productive to direct energy towards classifying causes than it is to parse 6's vs 7's.
 
uhmmm...I think you guys are over-interpreting the statistics. even over the 3 years, we are talking about only a handful of deaths when we want to distinguish between the 6 and the 7. one more or one fewer deadly crash is mostly noise.

I have been looking at the NTSB data base, but they only have 7 fatal accidents, 15 when I look for vans and rv, out of which about half are foreign. they do not contain the synopses (which I have read before as something like "primary cause was alcohol, secondary cause was pilot error" or the like), and which I could use to classify accidents. so, learning what the fatal mistakes were for our Vans pilot friends remains a dark mystery. it would be more productive to direct energy towards classifying causes than it is to parse 6's vs 7's.

I agree with you that causes are way more important than numbers - and the only way to get at those causes is to read final reports. There are lots and lots of final reports you can read - its not necessary to restrict yourself to the last three years, you can go back as far as you want because it is really, really (really!) hard for someone to come up with a new way to crash an airplane. People keep making the same mistakes over and over.

While it is true that the actual number of accidents in E-AB has gone down over the past few years, the actual CAUSES of those accidents hasn't changed much.

Sit down and enjoy the reading though - I am betting you'll learn a great deal about how to keep yourself from falling into many of the same traps.
 
Sit down and enjoy the reading though - I am betting you'll learn a great deal about how to keep yourself from falling into many of the same traps.

One of the things I did 20 years ago when the interweb was new to me was to read the synopsis of every fatal accident in the online NTSB files. There were probably 20 years worth available at that time.

As you wrote (and I clipped), there aren't many new ways to crash an airplane, but a careful study will turn up a few gems, like a Cessna Skmaster that came apart on a high altitude photo mission. Turned out that the shop which refilled the O2 tanks filled the tanks from the shop air compressor. So the pilot lost consciousness at altitude, the airplane went out of control and structurally failed (self disintegrated) on the way down.

The pilot died.

The really interesting part was that his passenger survived both the oxygen starvation and the crash.

https://www.flyingmag.com/safety/accident-investigations/hazard-oz
 
Good info

These discussions are important because while many of us have participated in them dozens of times, there are always people who are new to aviation safety who can benefit from them. Also, as has been proven many times, humans need regular reminders and are prone to complacency.

It might be useful to create and maintain a personal checklist of things to do to avoid becoming shamed in an accident database.

Something like:

  1. Fly the aircraft
  2. Keep enough fuel to get safely to the ground
  3. Stay in weather that you and your aircraft can handle
  4. Maintain the aircraft

I know if I do these fundamental things, I will have significantly increased my chances of avoiding an accident.

Volumes have been written developing each of these items. I'm sure you have read many of them. I continue to read them just to reinforce the message to myself.
 
Code:
RV-3/4      1,711	4	0.23%
RV-6/6A     2,631	2	0.08%
RV-7/7A     1,671	7	0.42%
RV-8/8A     1,448	3	0.21%
RV-9/9A     1,076	3	0.28%
RV-10       ,877	2	0.23%
RV-12       ,577	2	0.35%

these are 3-year accident numbers divided by the last count of airplanes flying. to the extent that relatively more of the latter models have come on line later (and fewer of the earlier), you may want to tilt it up in the latter years and down in the earlier years.

but just looking at the numbers, the only difference that seems meaningful is between 6's and 7's, and because these are essentially the same airplane (and if anything the 7 should be a little more consistent and safer), the rates are all reasonably similar. think a little under 1% per year. if I had to guess as to why the 7s look worse, it would be because the buyer profile could be a little different, maybe younger and more aggressive?! who knows.

it would make the analysis more interestingly if we would know the NTSB verdict on each of the 25 accidents, too.

/iaw

Is the apparent difference between 6s and 7s really statistically significant? The actual numbers are small. If I flip a coin I might get heads 7 times on the first 10 flips, and only twice on the next 10 flips. This doesn't reflect different probabilities, its just "the law of small numbers" at work.
 
RV Accident Stats

Code:
RV-3/4      1,711	4	0.23%
RV-6/6A     2,631	2	0.08%
RV-7/7A     1,671	7	0.42%
RV-8/8A     1,448	3	0.21%
RV-9/9A     1,076	3	0.28%
RV-10       ,877	2	0.23%
RV-12       ,577	2	0.35%

these are 3-year accident numbers divided by the last count of airplanes flying. to the extent that relatively more of the latter models have come on line later (and fewer of the earlier), you may want to tilt it up in the latter years and down in the earlier years.

but just looking at the numbers, the only difference that seems meaningful is between 6's and 7's, and because these are essentially the same airplane (and if anything the 7 should be a little more consistent and safer), the rates are all reasonably similar. think a little under 1% per year. if I had to guess as to why the 7s look worse, it would be because the buyer profile could be a little different, maybe younger and more aggressive?! who knows.

it would make the analysis more interestingly if we would know the NTSB verdict on each of the 25 accidents, too.

/iaw

So this must be just "fatal" accidents over the three year period? From about 2000-2005 I pulled all the RV accidents listed in the NSTB monthly summaries, and each year the total number of RV accidents was consistently between 20 and 30 aircraft (no RV-12 or 14's then). If this convers all accidents, then as a community we must be flying more safely than in the past...

Dou Lomheim
RV-3A Sold
RV-9A Mazda 13B / FWF
 
Back
Top