"Dude" writes:

though there isn't much data on the crash in Spain) were CFIT. Hard
to blame these on the plane per se.


"per se"?

Trying to separate out the "plane is a death trap" argument from the "plane
attracts idiots" argument.


Accusing those of us who think the statistics are relevant of hyperbole will
not save any lives, nor win the argument. The fatalities per 100,000 flight
hours stat is a very valid and fair stat.

And like all statistics it says only what it says, and drawing
conclusions from a statistic (particularly a single one) is very
risky. You have to ask a series of questions: What does the statistic
actually measure? Is the measurement statistically significant? Are
similar statistics comparable, and what do the comparisons mean? What
does the statistic have to say as a predictor for an individual (which
is really what people are concerned with)?

For example, if there's never been a fatality in a DA40, does that
mean that a Cirrus is infinitely more dangerous? If it turns out that
lots of idiots by Cirrus aircraft, does that mean that if you decide
to buy one then you as an individual are more likely to become an
idiot?

The statistics (assuming that they pass significance tests) really
tell you only that something is going on, but they can't tell you
what. This is a red flag to go and actually examine the accident
records and try to make an honest evaluation and decide for yourself
what they mean to you.

Once again, you can't take out the "stupidity factor" from one
manufacturer's stats, and not the others.

I fully agree, and herein lies the heart of the issue. Since there is
no independent "idiots per 100,000 pilots" statistic measured, it's
very difficult to quantitatively describe it. Furthermore, an
anecdotal analysis of GA accidents gives the impression that the
"stupidity factor" overwhelms all other contributors, which implies
that the planes, per se (there's that phrase again) are not a major
part of the problem. So unless the plane causes you to become an
idiot, as an individual thinking of purchasing one the statistics say
almost nothing about how much risk the plane itself poses to you in
particular.

I believe they are over a million fleet hours, and I am told that is
generally considered the time at which the numbers become valid.

This would imply an average of somewhere around 700 hours per
aircraft, which is way off the mark, considering that probably close
to half of the fleet was delivered in the last year. I suspect that
the fleet time is at most probably half of that, but of course we're just
making up numbers since this is never actually measured.

Another would be that everyone of us is likely to decide that we are not one
of those idiots. In fact, the ones that are dead likely thought that.

"Aviation, where all of the children are above average." ;-)

Either the design is safe or it is not.

C'mon, this is patently and obviously not true, unless your measure of
"safe" is "nothing bad ever happens" in which case all designs are
unsafe. Short of pieces coming off, it's almost impossible to measure
safety, except in very specific ways (things like impact tests, though
those are not necessarily predictors of anything useful either) or very
general ways (statistics, which don't tell you much.)

There is almost no practical
way to prove the cause without changing the results. Therefore, the design
is bad until it is found to be performing more safely. If Cirrus implements
a change, and then gets different results, then we can talk again. (the
parachute fix seems to have helped).

I think you're overreaching logically here. Perhaps I can restate it:
if there is a statistically significant difference in (accidents,
fatalities, choose your measure) then there is likely to be some
factor or factors that could be changed to reduce it. It's not just
"design" or training or even marketing; it's also things like low
time-in-type and mission profile (long XCs may be inherently more
risky due to unfamiliar terrain, multiple weather systems, etc.)

I don't see that a design change of the physical airplane will keep
people from doing stupid stuff (I suppose TAWS might reduce CFIT
accidents, but people would scream "crutch," which I agree with to a
point, though I suspect it will be standard in the avionics before
long.)

Cirrus has implemented changes to the training program, and COPA has
provided a number of resources, including recurrent training and
critical decisionmaking seminars, and a number of the insurance
underwriters are raising requirements for time and training. Whether
these changes will reduce the accident rate (or have already) will
take another chunk of time to determine. There are a few data points,
however; the rate of landing accidents (prop strikes, etc.) seem to
have dropped since they got rid of the original training organization
and started stressing speed and landing attitude control more. The
situation is not static by any means.

If the problem is indeed personality, perhaps they are selling the planes to
the wrong people. I would not necessarily disagree that this is the case
except to point out that they are not changing their sales practices and
other than looking at experience levels what are you going to do anyway.

I don't think either of us have any facts as to whether or not Cirrus
is changing their sales practices. And as you note, it's also not
clear how a sales rep is supposed to determine whether a customer is a
"wrong person" or not; they're not psychologists or mind readers, so
short of someone's experience level (or at least how they represent
it, as they're not getting a background investigation) there's not
much to go on, and it's unclear that overall experience levels
correlate with the accidents in any case.

Cirrus could get some good PR by simply dropping the SRV idea, and requiring
a high level of hours to buy their SR20 and SR22. I don't see this
happening, so I guess we will have a bunch more Thurman Munson Jr.'s.

It's unclear that this would actually help. One could make a case
that an SR20 or SRV would be an excellent aircraft in which to take
primary and instrument training, assuming that the pilot understands
that the process will take longer than it would in a 152. Typical
trainers are more forgiving, but after the first ten hours I'd argue
that this is a detriment, as it allows all kinds of bad habits (like
the 50' AGL roundout) to develop. Teaching speed discipline on
landing, and getting early and thorough exposure to the avionics,
could well make them better pilots. There are a fair number of pilots
who bought an SR20 to learn in, and so far as I know, none of them
have come to a tragic end. As such, their statistics look excellent,
for what that's worth.

The SR22 is arguably too much of a handful as a primary trainer,
though a few people have done it. The insurance people are the main
gatekeepers in this case. The number of low-total-time pilots flying
SR22s is probably quite small.