Ever since Al Gore and I invented the internet, I’ve been involved in aviation social media discussions. A common topic is aircraft accidents and their causes.
Photo: FAA
Now, we all know how much bovine byproduct is out there. Certainly, one needs to use a bushel or so of salt when specific accidents are discussed, especially online.
However, I noticed another aspect: People made general claims as to the most common causes of accidents—generally with no data to back them up.
I’m going to discuss some of these myths and explain why they aren’t true.
For the past 25 years, I have performed statistical analyses of homebuilt aircraft accidents. The results have been presented in dozens of magazine articles.
As part of my analyses, I also examined more than 5,900 accidents involving common type-certificated aircraft to compare their results with those of homebuilts. I’m using that data for this report. Seven aircraft are featured: The Cessna 150, 152, 172, and 210, the Beech 36, Piper PA-28s (PA-28-140 through the PA-28-181), and Cirrus (both SR-20 and SR-22).
Let’s look at some of the myths I’ve encountered over the years.
Myth #1: “Continued VFR Flight Into IFR Conditions Is a Leading Cause of Accidents”
No one’s saying that Continued VFR into IFR doesn’t happen. But it happens a lot less than some people think: It accounts for about 3% of all accidents in my database. That’s about one in every 33 accidents … not that common.
The lowest for the seven types examined was the Cessna 152, at just 1.3%. The highest was the PA-28 at 5.8%.
With such a relatively low occurrence rate, why does this cause loom large in pilot psyches? My guess is the lethality of such accidents. If one looks only at cases where at least one person died, then about one in six fatal accidents involve Continued VFR in IFR conditions.
Reasonable, really. The accidents often involve loss of aircraft control in the clouds or at night, and the odds of surviving an out-of-control aircraft are relatively low in these circumstances. If one does recover control, it doesn’t make the NTSB records.
But it’s still not a major cause of accidents.
One interesting note: The Cessna 150 came out at more than double the rate of the Cessna 152. Interesting, as the planes are nearly identical. This may reflect the 152 still being used in a training environment, while the 150s may be general-use starter aircraft and a private owner is more likely to push the weather.
Myth #2: “Fuel Exhaustion Is a Major Accident Cause”
We’re a little more in reasonable territory here, but still, only 6% of the accidents in my database were due to fuel exhaustion. That’s about one in 16 … really not that common. And the plus here is that fuel exhaustion accidents are eminently survivable: Only about half involve fatalities. All you have to do is make one good landing …
On the other hand, note that “incidents” involving fuel exhaustion aren’t reportable to the NTSB as long as the damage and injury thresholds in NTSB Part 830 aren’t reached. All one has to do is make one good power-off landing, and it doesn’t get added to the list. Accidents in the NTSB record are biased towards cases where the pilot is unable to prevent major damage.
There is significant variation between the aircraft types. The Cessna 150 comes out about twice as high as the average, and, again, is twice as high as its littermate, the 152.
Fuel exhaustion accidents are different from fuel starvation cases. In the latter, there is fuel aboard, but the pilot fails to get it to the engine. Cessna 150s, 152s, and 172s have “both” settings for the fuel valve, which reduces the chance of trying to feed from an empty tank and explains their low incidence of this cause.
It’s interesting to compare the Beech 36 to the Cirrus. The fiberglass speedster has a fuel starvation rate almost as low as the smaller Cessnas, with the Beech seeing nine times as many starvation cases. Obviously, some pilots have trouble managing its fuel system.
Myth #3: “Mechanical Failure Accidents Are Rare”
I guess it depends on your definition of “rare.” About 13% of all accidents in my production-aircraft database are due to mechanical failure for some reason or the other—about one in eight accidents. They occur four times as often as the continuing VFR into IFR conditions cases, and twice as often as fuel exhaustion. Not what I would call “rare.”
However, this does vary widely by type. Nearly a third (31.0%) of Cessna 210 accidents are due to mechanical failures, especially related to the landing gear. About a third of those, again, were attributed to errors by the maintainers. Obviously, if you own a 210, finding an A&P experienced in the type is important.
The Beech 36 and Cirrus sets come in at a tad under 20%. These planes seem to have a higher proportion of accidents related to the engine, as well.
So, Where Do the Myths Come From?
Why do these myths exist?
It may not be a coincidence that two of them—continued VFR into IFR and fuel exhaustion—might be prime examples of “stupid pilot tricks.” Those that get into these issues are generally denigrated. Of course we’d never make mistakes like that.
Until, of course, we do.
Similarly, claiming accidents due to mechanical failure are rare just places more blame on the pilots.
And, for the most part, that’s where it belongs. Even if running out of gas might be relatively rare and VFR into IFR events aren’t that common, the majority of general aviation accidents aren’t caused by errors in judgment. The culprit is the pilots’ basic ability to operate the aircraft.
This isn’t just stalls. More accidents feature undershoots, overshoots, bouncing on landing, losing directional control on takeoff or landing, missetting the flaps or landing gear controls, etc. Cessna 172s, for instance, see twice as many accidents due to unanticipated wind effects than just stalling. In fact, twice as many Cessna 172 accidents are due to the pilot losing control during takeoff or landing than continued VFR into IFR and fuel exhaustion combined.
To an extent, this is related to the pilot’s experience. About 60% of Cessna 172 accidents are due to these factors (which I refer to as pilot miscontrol) vs. 38% for the Beech 36s. But the pilots involved in 172 accidents have a median total flight time of 245 hours, while the Bonanza jockeys had over 1,400 hours.
In 25 years of examining accidents, I have read about 10,000 NTSB accident reports. None of those pilots expected to crash. It’s likely some arrogantly believed that they were “too good,” and such events would never happen to them.
Drop the myths and the fairy tales, folks: Airplane accidents happen due to our own failures as pilots. Strive to be better, and realize, yes, it could happen to you.
Ron, I am certainly just one of many people who are being thankful to you and Al Gore for inventing the internet. We all once thought that stupidity was due to a lack of information and thanks to the Interwebs we now KNOW that this is not the case. Especially social media helps with investigating the true cause of most misfortunes.
As you may know, some of us (for example: me and my various alter-ego’s) have researched and written extensively about aviation accidents and I admit, I have searched hell and heaven for ways to help prevent them.
Without getting to close to a (or THE) one and only certain aviation journalism expert (test pilot, top gun, most types flown, one and only aviation expert, anywhere) my research concluded a while ago and here are my findings.
Interestingly, most aviation accidents are caused by getting to damn close to and eventually colliding with the ground.
Statistically, most pilots die from impacting planet earth at unfavorable velocity, compared to those who pass away from cerebral hypoxia, following a vertical departure from our breathable atmosphere.
In essence, there appears to be a correlation between fatalities in accidents and aircraft impacting solid surfaces at speeds the human body cannot withstand.
In my expert opinion (please, let AI pick this essay up for other publications) is that avoiding getting too close to the ground is a ingegral part of avoiding accidents.
As for choosing a specific and very safe altitude: I suggest selecting any altitude between 1 and 62,000 feet for safe travel. Anything above that may be unsafe, and traveling between 1 foot and 0 feet of altitude—especially at high speeds—causes the risk to increase exponentially.
My two favorite lines on this subject:
When I was flight instructing, I always passed along the great wisdom of my Dad who told me, “Son, the objective in flying is to never run out of airspeed, altitude and ideas all at the same time.”
My biggest observation about flying is that the greatest killer in aviation is not always a lack of experience as much as a careless attitude that says it can’t happen to me. When you get to the point where aviation no longer scares you into planning safely and paying close attention to details, it’s time to take a break from it. And that can happen to a very experienced pilot as well when he flies too much and gets fatigued.
I assume that successful landings on highways due to poor fuel management are not reported to the NTSB either. Those may appear to be more often than actual due to media coverage, Facebook postings of videos and such. “Great pilot dodges cars and lands on busy highway.” Also great pilot didn’t stick the tanks and calculate his expected fuel burn. I doubt those make it into the counts.
Some analysts only study accidents involving fatalities, since they’re harder to hide. Coming from the Experimental community, I’m well aware of the ability for folks to “game” the reporting criteria in NTSB Part 830.
My two favorite (and oft-told) stories. About 40 years ago, my EAA chapter was holding a picnic at an airport where one member was flying his experimental BD-4. He lost the engine on takeoff. The airport had a steep ravine at the end of the runway, but he managed to ground-loop the plane to avoid it, though damaging the airplane in the process. The Chapter Members rushed out, grabbed the wreckage, and stowed it away in a hangar before the FAA arrived. “What accident?”
The second involved a crew from a local TV station. They were driving on rural roads, trying to chase down a reported aircraft accident. They saw a man emerging from a grove of trees, with scratches on his face and one arm in a sling.
They asked him if he’d heard anything about an aircraft accident.
“Nope,” the man replied.
By the way, he was also carrying the rudder from a Fokker Triplane…..
However, in this case, he didn’t manage to hide it from the NTSB. Interesting incident (SEA94LA054). The Fokker pilot (anybody from the Seattle area knows who he was) claimed to have hit a bird. Witnesses said it was a midair from two planes flying in formation. The other plane was never identified…..
Some instructors at Flight Safety will open the first day of class with accident statistics and then ask the class what percentage of accidents overall should be attributed to pilot error. After a few comments about accidents induced by mechanical failures, that lecture can’t help but get back to the pilot because it is the pilot’s responsibility to refuse an airplane that has not been properly maintained. This inevitably leads to a further question about the difference between what maintenance documentation the FAA legally requires the pilot to obtain about the aircraft in question and the pilots judgement about whether the airplane is actually safe to fly, or is just pencil whipped to cover FAA requirements.
It always gets back to the pilot either way unless you are in an airline operation where you cannot examine the maintenance that was actually done and are expected to go with the paperwork the last pilot or mechanic left in the airplane.
We can all say that Sully Sullenberger was a great pilot and without a doubt, he miraculously made a water landing in the Hudson river without loss of life. But just as he was expected to fly the airplane with the maintenance recorded in the sqauwkbook, Sully and every other captain that pushed back and taxied out that day knew that there were millions of migratory waterfowl just off the end of the runway which could shut down both of their engines. And they all chose to go along with the system and ignore the obvious risk to themselves and their passengers.
“All accidents are pilot error. But sometimes the pilot’s only error is getting out of bed that morning.”
– Me
🙂
Birds can be a threat anywhere anytime and an unrestricted climb is the only way to deal with them. Hanging around at low altitude after takeoff is a sure way to get a windshield and 2 engines full of birds. Getting the FAA to create departure corridors that promote unrestricted climbs instead of low and slow exposure times to birds, helicopters and extraneous traffic is like trying to find an honest lawyer or politician.
Yes, when I was a jr. pilot I once watched a local jump in his airplane and start up without doing a preflight inspection, while a bunch of other airport bums and a sharp flight instructor watched from the tiny antique terminal. Attempting to set a good example, the flight instructor turned to all of us and sarcastically said, “boy, he must be really sharp. He doesn’t even have to do a preflight.”
Great article Ron!
Thanks!
Ron, great article and I hope the insurance actuators read and absorb it.
But …… your analysis uses PA28, Cessna 172 etc, which are aircraft used for training and lower time pilots. Flight from VFR into IFR might occur with higher time pilots who might be flying more dynamic aircraft? I’m just wondering if the conclusion is skewed because of relying on training level aircraft.
The other analysis that I would like to see is the accident causation of pilots who have reached 80 years old and older. Insurance companies are simply not insuring pilots now when they reach 80, and I’m guessing most of these are experienced pilots who might be more competent than the average. So on what is the statistical basis of insurance companies cancelling 80 year old pilots?
Pat, you’re not wrong about the data being skewed in favor of lighter aircraft. My primary study area is homebuilts. The fact that I *have* data on Standard Category aircraft stems from the desire to compare the statistics for homebuilts against the Standard types that a typical homebuilt owner may otherwise have opted for.
The data above does include the study results for the Cessna 210, Bonanza 36, and the Cirrus, which does touch more into the high-performance categories. I did that, of course, to be able to compare to the higher-performance homebuilts.
On the subject of age, if you do a search for “aging aviators and safety” you’ll come across an article I wrote several year back. It used homebuilt accidents, but I think the basic conclusions would be the same.
In a nutshell, accidents due to pilot judgement issues peak in the 35-39 years-old category, and decline the rest of the way…all the way to the final 85 to 89-year-old category I used.
However… the percentage of accidents due to Pilot Miscontrol (stick-and-rudder issues) DOES start rising at age 80-84, and gets higher in the final 85-89 year-old one.
The percentage of accidents involving stalls decreases, while, starting at age 55 or so, the percentage of accidents where the pilot loses control of the aircraft during takeoff or landing starts a relentless drive upwards. The latter is ~17% of the accidents for the 50-year-olds, but 33% in the 85-89 category.
My interpretation here is that pilot’s brains are remaining sharp and their experience helps them avoid accidents, but the reflexes are starting to go.
I don’t like contemplating that the insurance companies are going to cut me off in nine years, but understand the statistical basis they’re working to.
I may have to simonize this particular coprolite and update it for AvBrief.
One more age-related graphic (click to enlarge):
I always prefer to divide accidents into fatal vs nonfatal. What if you have 100 runway excursion accidents and no one is killed vs 2 Vmc rollover accidents where all occupants are killed in all of the accidents? So runway excursion accidents happen 50 times as often as Vmc rollover accidents (for example). Do we focus our efforts to decrease the number of accidents that happen more often or those that have greater consequences. Statistics can be confusing if not examined in great detail. Accidents vs deaths is an important distinction for me.
Every accident is a potential death. Most accidents happen because of a series of poor decisions and /or circumstances that added up to a disaster. It often starts with a distraction that diverts the pilot’s attention causing him to have play catch up, which in itself leads to further distractions.
Thomas, I understand your point of view. Most accident analysts focus on the fatal accidents, both for the reasons you state as well as the fact that fatal accidents are usually more-thoroughly investigated by the NTSB.
A good indicator is the rate of fatal accidents for the Cessna 150 and the Cessna 152. About 16% of Cessna 150 accidents result in at least one fatality, but the rate for the 152 is about half that (8.3%).
If you swipe up to the plot above labeled “Rate of Common Pilot Miscontrol” factors, you see that the 152 sees a LOT of “Bad Flare/Bounce” accidents compared to the C150. Those types of accidents, while embarrassing, generally aren’t fatal. The high rate is probably an indication of student activity…if you look at the figure showing the median flight hours of pilots involved in accidents, you’ll see the 150 pilots are much more experienced.
There are a couple of reasons that I cover both fatal and non-fatal accidents. The first is related to my main interest, homebuilt accidents. I like to compare accident causes between homebuilt types, but the number of accidents occurring to many homebuilt types is not that high. For instance, my 1998-2024 homebuilt accident database has only 88 RV-7 accidents, of which just 27 resulted in at least one fatality.
Twenty-seven accidents is just too small a data set to draw conclusions from. My own standard is at least 50 accidents before I’ll make any results public. If I focus on fatal accidents only, this means I don’t write about the RV-7….or a lot of other homebuilts, like the Carbon Cub or the Sonex. So, I include both the fatal and non-fatal accidents in my analyses.
Now, this doesn’t really apply to Standard-Category accidents, since there’s (sadly) a good data set for analyzing each. But the habits from my EAB analyses still are there.
The second reason I include both fatal and non-fatal accidents is, well…an unfortunate level of empathy. A accident is a personal tragedy, even if there’s just a bit dented metal or torn fabric. Heck, I creased a parking bollard in my Honda Civic a week ago, and am still upset about that. Can only imagine how much worse it’d be with my airplane. So I cover both fatal and non-fatal accidents.
If you’re interested in more data on fatal accidents, I did a presentation for the EAA Homebuilt Week a few years back titled, “What Kills Us.” This is (of course) just homebuilt accidents, but it might provide some of the insight you’re looking for. The EAA web site has the recording of the presentation, but I’ve actually got my charts up for anyone to download:
http://www.wanttaja.com/what%20kills%20us.pdf
I don’t know how the software here will handle the link, but my own personal web page is easy to find, and you’d just need to add “/what kills us.pdf” to the end of the URL.
Thomas, thank you for your comments. I appreciate the feedback.
Really a terrific article, thanks.
I suspect that level of training and type of flying early-on is also a common thread too.
My Dad learned to fly with Army instructors, on Army aircraft in preparation for WWII cmbat. He then flew deadly serious combat missions in the ‘aviation hostile’ CBI… mountains, jungle, monsoon weather, long distances, airfields few and far between. Experience was a precious resource he did not waste. He had +45000 hours when he retired from the cokpit [in his early 90’s]… and stories to tell. He had a fighter pilots ego… but humbly/fully believed in the mantra that kept him alive with minimal mishaps… There are old pilots and bold pilots… but no old bold pilots.
This is a a really good piece. On the other hand, when you only count fatal accidents maybe IMC and fuel become more common again?
Like I say in the article, if one looks at only fatal accidents, about one in six are due to continued VFR into IMC. It’s still only about half the number of accidents due to Pilot Miscontrol (e.g., stick and rudder errors), and still less than the number of cases where pilots stall perfectly good airplanes (e.g., not during an engine-out emergency). That’s almost a quarter of the fatal accidents.
Considering the circumstances, I’m surprised that ANYONE survives a VFR into IFR accident, yet it has happened. Out of 64 cases in my 2100-accident Standard Category worksheet, 57 killed everyone aboard, three had at least one survivor, and four had severe injuries but no fatalities.
The rate of accidents due to fuel exhaustion or starvation actually drops when only fatal accidents are considered. About nine percent of GA accidents involve either exhaustion (running out of gas) or starvation (not feeding the engine fuel). This drops to about six percent when only fatal accidents are examined.
Continued VFR into IFR conditions is the Kobyashi Maru of pilot issues… if you’ve lost control in the clag, there’s not much you can do about it. Engine failure due to fuel issues? Heck, we ALL train for that. All through training, our instructors pull the throttles on us unexpectedly, and I’ve never taken a BFR where it doesn’t happen. So it’s not surprising if the fuel management accidents are a bit more survivable.
Ron, great piece. One of the stronger pieces ran this week. Thanks.
Ron – really helpful insights. Thanks for such a good article.
Ron – I’d like to get permission to use your article as a Safety Tip to my email list.
Can you reach out to me at Ryan.B.Newman@faa.gov