(Rich Stowell) wrote
Your experience vis-a-vis how NTSB accident numbers (reported) stack
up compared to the actual total number of accidents (reported +
unreported) is not at all unusual. I too know of numerous unreported
accidents. But this isn't peculiar to aviation -- it probably happens
in a lot of different settings and more often than we realize. That's
what makes estimating and/or extrapolating accident information so
difficult. Perhaps if we always stated a number plus-or-minus some
estimate of the error that'd be satisfactory?

Well, it would be satisfactory if you had a valid way of estimating
the error. However, this still requires data. It's really very
unfortunate that data collection in GA is a bad joke. It's a lot
better at the Part 135/121 levels. For example, in those operations
every engine failure is reported (at least for turbine engines). A
skydiver friend of mine once looked up engine failure rates on the
PT-6 engines that power Twin Otters. The numbers looked artificailly
low to him. He started making phone calls to friends. It's not easy
data to get, but his best estimate is that Twin Otters in Part 91
service have an engine failure rate about 300 times higher than in
Part 91 service. He admits that the real number could be as low as 50
and as high as 1000, because of uncertainties in how operating hours
are estimated as well as uncertainties in which inciddents were
unique.

It's very frustrating as an instructor not to be able to give your
student solid estimates on which risks are truly significant, but the
alternative (reaching firm conclusions from nonexistent numbers) is
far worse.

"Relatively few" is the 1 accident per 331 (or X) total hazardous
encounters.

Well, that's great - now define hazardous encounters. With engine
failure, it's easy. The engine stops making power (or stops making
sufficient power to maintain level flight) and you have an encounter.
With off-field landings it's easy - when you commit to an off-field
landing, you have an encounter. Stall-spins are harder. Does the
hazard encounter start when autorotation begins? When the stall
breaks? When the buffet starts? When the stall horn goes off? See
the problem?

As a flight instructor charged with the task of educating pilots and
(hopefully) offering them guidance in terms of how often to practice
certain procedures/maneuvers on their own, what frequency do you
recommend in this regard, and on what is that recommendation based?

Whatever is required to maintain proficiency. If during your
recurrent training cycle you handle the engine failure competently
(meaning accomplish the task smoothly and with the successful outcome
never seriously in doubt) then you are maintaining proficiency.
Otherwise you are not.

Specific numbers are very much a function of the airplane and pilot
proficiency, and one size does not fit all.

I think we'd agree that the number of such practice approaches is
somewhere between 0 percent of the time (airline-type flying) and 100
percent of the time (gliders). Also, and though it hasn't been stated
explicitly, I'm talking in terms of the "average," "typical," "normal"
pilot flying the typical light airplane on a typical flight. That
said, I do believe that an average pilot who performs 10 gliding
approaches out of the 20 approaches annually will be likely to react
appropriately to an engine failure.

I think so too, assuming that what you mean by react appropriately is
point the plane someplace reasonable and maintain proper flying speed
to the flare. I doubt he will select the optimal location or make a
great landing, but those things are probably not crucial to his
survival.

Now, because this pilot may lack
the breadth of experience of your atypical 300-landings-per-year pilot
who practices gliding 10 percent of the time, the less-experienced
pilot (but who is more representative of the norm) may not be as
precise overall, yet the fundamental skill set needed to cope should
be there nonetheless.

No argument.

In fact, typical pilots under duress will invariably only be able to
perform as well as their most basic skill set allows.

Define most basic skill set. Keep in mind that for some, this will
include night partial panel flying. For others, it may be
substantially more limited.

And those skills
that are the most practiced, the most familiar, the most "natural" to
the pilot are the ones that will largely determine the outcome. Again,
this is based on my anecdotal experience instructing 1,000's of pilots
while they are placed under duress during emergency maneuver training
-- typical pilots from across the country who are representative
"products" of our national flight training system.

I'm not convinced that's true. I suspect that the pilots who
voluntarily get emergency maneuvers training are the same pilots who
doubt their ability to handle emergencies. Such doubts are usually
justified.

I don't believe a stall-spin involves a typical pilot at all.

The numbers and the anecdotal experience of professional
spin/aerobatic flight instructors are totally at odds with your
belief.

So how many inadvertent stall-spins do they get to see under normal
conditions?

The typical pilot is trained by the typical flight instructor,
who himself/herself has a marginal understanding of, and marginal
practical experience with, anything related to stalls and spins and
therefore, is incapable of adequately providing stall/spin awareness
training to their students.

That's probably true, but on the other hand most modern airplanes have
to be pretty severely mishandled to cause an inadvertent spin. The
same is not true of gliders and aerobatic airplanes, but the people
flying those tend to be trained by a completely different sort of
instructor.

And pilots with fewer than either 500 hours total time, or 100
hours in type, are more likely to encounter an inadvertent stall/spin
than to have a genuine engine failure.

Why do you keep going back to the patently unprovable? All we know is
that they are more likely to have an accident caused by stall-spin
rather than engine failure. This tells us nothing about the
likelihood of encountering either hazard.

Now let's consider something else. A pilot who flies 200 hours a year
is 10 times more likely to have an engine failure than one who flies
20 hours a year, since engine failure is not under his control. Are
you seriously suggesting that a pilot who flies 200 hours a year is 10
times more likely to inadvertently spin than one who flies 20 hours a
year? I would argue that he is LESS likely to inadvertently spin,
since the higher level of proficiency that is a nearly inevitable
result of flying a lot and often will make him less likely to miss the
rather obvious clues.

Consider the following 1987 stats as well: the U.S. boasted 699,653
active pilots who collectively logged an estimated 47.9 million flight
hours. Amortized, pilots averaged 68 hours each that year
(unfortunately, this average had decreased to less than 50 hours per
pilot per year during the 1990's).

However, only half of those are private pilots. Tens of thousands are
professional pilots who fly hundreds of hours a year. Thus the
reality is that a large number (maybe the majority) of these 'active'
private pilots are actually flying less than 20 hours a year. IMO it
is not possible for such a pilot to be proficient at all. In fact,
I'm going to take back what I said - if the average pilot is indeed
flying 20 hours a year, then I will readily admit that inadvertent
stall-spin accidents are going to happen to the average pilot. I
guess my point is that they ought not to happen to the proficient
pilot. Instead of looking at total hours, we should really be looking
at hours flown a year, and even then there are other factors since not
all hours are created equal.

Somewhere, somehow the discussion shifted from "total accidents" to
"fatal accidents only." The 1:30:300 is all hazardous encounters
leading to all accidents, not total accidents vs. fatal accidents vs.
some-injury accidents vs. non-injury accidents. Regarding mid-airs,
the question would be, "for each mid-air, how many times are airplanes
coming close enough to each other to be considered a hazardous
encounter (especially when pilots in both airplanes have their heads
buried in the cockpit on a clear VFR day)?" Maybe 331 times as many as
the total number of mid-airs that resulted in accidents (whether those
on board were injured, killed, or not)???

Sure. I guarantee you that there is SOME definiton of "close enough
to each other to be considered a hazardous encounter" that will make
the numbers work. By the same token, there is SOME definiton of
"stall-spin hazard" that will make the numbers work. This is
meaningless.

Michael