What Kills Us?
We are all, I suppose, victims of our own obsessions and one of mine is the notion that among the many things we don't teach well in aviation is risk assessment. That's not risk management, mind you, which I view as an overused buzz word, but rather a realistic grasp of what's most likely to kill us when flying an airplane. That's another way of saying that many—if not most—pilots worry too much about the wrong things and not enough about the real killer risks.
I notice this when flying with other pilots or speaking to owners about equipment upgrades. Specifically, some pilots are absolutely terrified of midair collisions. While that's understandable, it's not necessarily justifiable. The graphic at right shows why. This is data I distilled from all of the general aviation fatal accidents in 2010. The numbers to the right of the circles represent the actual accident rate per 100,000 flight hours that these categories represent while the circles themselves graphically represent the relative size of the risk.
These categories were my own and intentionally painted with a broad brush so the risks would stand in sharper relief. There's definite bias in my analysis and I'll get to that in a moment. Also, there's obvious anomalous exposure inconsistency. On every flight, you're exposed to stalls, to loss of control, to running out of fuel and into trees and rocks. But icing and convective events represent a smaller universe because not as many pilots venture into such challenging conditions.
From my perspective, what the graphic shows is that pilots tend to have an inverted understanding of risk and what's most likely to kill them. So they spend big dollars on traffic systems to mitigate what's practically the smallest overall risk: a midair. There are thousands of traffic systems out there and although they're good, between-the-ears comfort providers, I doubt if they've moved the overall accident needle in the slightest. Midairs as a percentage of total accidents haven't changed much in the 20 years I've been following these numbers. Why? Big sky theory works. Big sky and not that many airplanes. See and avoid fills in the small gaps.
Sadly, upper strata of the accident heap hasn't changed much, either. Stall-related accidents are at or near the top of the accident results/causes year after year. I'm using the term results/causes deliberately because of the often inseparable juxtaposition of the two. If an engine fails and the pilot stalls the airplane trying to perform an emergency landing, what was the cause? The engine failure or the stall? Was the engine the cause, the stall the result? If we assume any pilot should be able to land an airplane survivably after an engine failure, then the stall looms larger as the fault in the system, even though an engine failure or fuel exhaustion may be been the gateway event for the stall. That's where the bias comes in. You might code these accidents a little differently than I would.
Not that it makes much difference, frankly. In that group of stall accidents are a couple that involved engine failure or fuel exhaustion, but the majority are accidents in which a pilot—while landing or taking off—simply stalled a perfectly functioning airplane. As much as I am at a loss to explain why this happens so often, I am equally baffled by how pilots can be so utterly unaware of the stall risk. It seems obvious that our stall awareness training isn't entirely broken because it effectively reaches the vast majority of pilots who can fly an airplane without stalling it. But for an unlucky slice of the pilot population, aerodynamic stalls, as a concept, seems to elude. Technology may or may not help. Cirrus invested heavily in a spin-resistant wing design, yet stalls and spins are a significant feature of the Cirrus accident record. If owners had a real sense of the stall risk, would they be more inclined to spend money on an angle-of-attack indicator first, a traffic system last? Or perhaps to seek out recurrent stall awareness training? These are questions worth considering.
My contribution to the discussion is this: I'd like to see the basic understanding of stalls de-linked from airspeed and more strongly associated with angle of attack. And not just with electronics indicators, but seat-of-the-pants, fly-the-wing pitch awareness. Reliance on glass gadgets has made many of us numb to reality of air flowing smoothly over a wing; increasingly, we understand only electronic abstractions, not real physical phenomena. I'd also like to see more instructors take their students through a range of aggressive stall awareness instead of just the pro forma basic stall series.
Then there's the data itself, which is incomplete if not suspect. That's why the second big risk circle is what I call unknown or other. Even two years after the fact, the NTSB often doesn't have enough data or doesn't publish enough to judge what really happened in some of these accidents. I further suspect some of the investigations, because of lack of resources, reach the wrong conclusions. Our sister publication, KITPLANES found this to be true in the NTSB's analysis of experimental aircraft accidents. Many of the airplanes had been miscoded as experimentals, but were in fact ultralights or certified aircraft. Groups like SAFE, which are pushing for an industry-wide reduction in the fatal accident rate, have similarly called for improved accident reporting accuracy. After all, how can you fix a problem if you can't identify it in the first place?
A word about CFITs, which are also a largish risk. The classic definition of CFIT is of an instrument or night accident where the pilot flies a normally functioning airplane into rocks or trees. Depressingly, it happens in daylight, too, with whatever object eventually becomes the bulls-eye in plain sight. Year after year, these inexplicable accidents appear in the database. Some are scud runs, some are VFR into IMC, some are undershoots or takeoff accidents. But all share one characteristic: the airplane was under control at impact and probably not stalled, suggesting the pilot had options he did not use.
On a scale of one to 10 as a risk taker, I'm probably an eight. But there are people out there who make me look like your white-gloved Aunt Jane. The very first fatal accident of 2010 occurred in January when a 77-year-old pilot on his way to the United Kingdom in a Cessna 172 crashed into Penobscot Bay after he encountered extreme turbulence, having departed into a forecast that included icing. He was a multi-thousand hour experienced ferry pilot, but still, I fail to grasp any sane risk/reward in flying the North Atlantic in winter in a Skyhawk.
Despite the dodgy data, I think it's accurate enough and sufficient enough to make broad conclusions about risk assessment, especially for pilots personally. And that's to say worry less about midairs, thunderstorms and icing and more about simple, basic control of the airplane. If you combine the stall accidents with what appear to be basic loss of control accidents—in IMC or not—you account for nearly half of all fatal accidents. That's a big number and plenty to chew on to worry about things that might kill you and less about things that won't.