It’s often been claimed that pilots who fly 10hrs/year don’t crash more often (per mile or per hour) than pilots who fly 100hrs/year.
This is obviously counter-intuitive. I wonder whether it is due to risk compensation (low time pilots stick to “simple” flights on nice days, etc) or whether it is because currency is not the big thing which is often claimed?
From the point of view of a safety regulator, risk compensation would be frowned upon as a way to achieve a given level of safety (and obviously it doesn’t apply to commercial flights where you fly if the wx is above minima, or you get fired) but if it works, it works…
I would imagine there would be so many differences between the groups that all comparisons would be futile. Most of the 12 hour pilots will be in benign spamcans. The 100 hour pilots will be flying more demanding types in worse weather and doing things like aerobatics. In a way, not doing those things if you only fly 12 hours a year is risk compensation. I’m not sure what lessons to take from the observation.
I do not think the regulators would frown upon pilots being prudent. The consequences of a pilot embarking on a flight which is beyond his abilities will however be commented in the incident/accident report where pilot total and recent experience is always mentioned, and at length in the forums.
From the point of view of a safety regulator, risk compensation would be frowned upon as a way to achieve a given level of safety
What exactly do you mean by this?
I don’t think the claim is right. Paul Craig has identified an hour range, where most accidents are likely to happen. In his Book “The Killing Zone” he describes the pilots with 50 to 350 hours to be most represented in general aviation accidents.
In my experience low time pilots with confidence and high time pilots with complacency are the most dangerous people in the air. And Risk mitigation or risk compensation is what every instructor should aim at in new pilots, and what every seasoned IR pilot will tell you: Your personal limits are higher than your legal limits and they reduce slowly with experience.
The statistics in “the killing zone” are completely uselss because they do not compensate for the size and activity level of pilots with a certain number of hours. Of course low hour pilots have more accidents, there are a lot more of them.
Oh you can find something for every statistic that isn’t taken into account. The question is if it impacts the conclusions you draw for them, and for the killing zone, it seems quite reasonable to me, so far (not quite half way through, yet).
It’s true that there are lies, damn lies and statistics. On the other hand, carried out carefully and correctly, statistics can provide valuable insights. It’s true that perfection isn’t possible in statistics – but there’s still a gulf between the way a reasonable statistician approaches a problem – cognisant of all the weaknesses and assumptions in his argument – and the ignorant-and-unaware approach in ‘The Killing Zone’.
For a discussion of why the stats are so wrongheaded, the Amazon reviews are a good place to start:
My own review:
This author postulates the ‘killing zone’ – a period during which newly minted pilots are particularly vulnerable to making fatal misjudgements.
Whilst this ‘killing zone’ may or may not be real, the author fails to make a good case for it. For example, many student pilots take longer than the minimum hours to complete their pilot’s licence, so the observation that risk climbs towards 100 hours may simply reflect the fact that this is when many pilots are ‘let loose’ for the first time.
Likewise, the book fails to comment on the fact that most pilots give up flying without having logged vast numbers of hours. Most 300 hour pilots will never reach 1000 hours, whether they kill themselves or not. Another factor is that many pilots in training are working towards a commercial license, and will fly for around 300 hours in small piston aircraft before moving up to safer commercial aircraft. The observation that 300 hour pilots have more accidents than 1000 hour pilots may simply reflect the fact that there are that many more of them.
In a book from an academic publisher at an academic price, you would hope that the author to have provided a more robust argument to support his thesis. The discussions of the most common types of accident and how to avoid them are relatively good. However, if you read accident reports and safety articles in the magazines there is likely to be little new here. I learned far more from ‘That Worst Day’ than I ever did from this flawed book.
That of a federal employee, who has gone as far as writing a paper to debunk it.
Unfortunately, Craig repeatedly commits a rather serious statistical error in this book. He uses accident frequency counts, rather than accident rates, as the statistical basis for his conclusions about the range of the “killing zone.” Frequency counts are interesting, of course, but they don’t account for the number of pilots at each range of flight hours (which accounts for most of the effect he claims). Therefore, they say little about the risk that you yourself face as your flight experience increases. My concern is the nature of that zone, and that we use the right methodologies to explore the issue. You’ll have to forgive me for being geeky about this. It’s just that it’s part of what I do for a well-known agency having to do with aviation (which can’t be named, because I’m speaking here as a private citizen).
Statistically, rates aren’t interchangeable with frequencies. Rates subtract the effect of how many individuals are present in each “bin” of a frequency distribution (in this case, the y-axis, where the x-axis would be flight hours). In fact, it appears that about 70% of the “zone” may be an artifact, and can be explained just by the fact that the frequency distribution of NON-accident pilots looks nearly identical to the distribution of accident pilots. See my paper http://www.sciencedirect.com/science/article/pii/S0001457513003242 regarding this. Or, see the free government technical report at https://www.faa.gov/data_research/research/med_humanfacs/oamtechreports/2010s/media/201503.pdf .
Bottom line: The kind of analysis we use on data like these is very tricky, is all I’m saying.
There is a lot more discussion following this post which is worth reading:
Does it matter that the stats are bunk? Yes. Part of me was just upset at being fleeced for a fairly large sum of money for a book I don’t have much respect for. But looking at Robert Knecht’s paper, it suggests the book could promote a false sense of security amongst pilots coming up to the 300 hour mark, as the risky period may last a lot longer than this. Secondly, I’m not sure that low-hours pilots are likely to be quite as low-risk as he believes. On the US Amazon.com site, some of the reviews also pick holes in his mental models of how aircraft fly.
Interesting, thanks. The book was a gift, so I didn’t research anythin on it (yet). (But I doubt it was even close to academic book prices. Don’t make me look it up :-) )
Anyway, sure, Craigs method of determining his killing zone is a bit blunt, but as I understood it wasn’t meant to be a precise rule, less a scientific evaluation of total flight hours. I thought it to be a mere peculiarity in counting accidents and he was using this to write a book to educate new pilots on airmenship and system knowledge, until they reach experience and knowledge that puts them out of reach of Dunning Kruger (more or less). It is written quite entertaining and I never would have guessed it was meant as a scientific literature, since he doesn’t back up his claims in any scientific way. Thus I am stunned to learn that anyone took his claims that serious. Especially, while the variance of personalities and utilisation within the general aviation is as huge as it gets – a fact that is stated by Craig right at the beginning and supposedly the reason why he chose to write from accident to accident. I would still recommend the book for what it is: a guide how to think about accident reports, catch up on common mistakes pilots make and get knowledgeable about your own risk of flying.
And Peter is right. Risk of being involved in an accident has very much to do with the situations, the pilot allows himself to get into. And this is a function of the kind of flying the pilot was trained to do and the kind of learning he will keep up for his aviation career. Put me with my knowledge solo in a hand flown IFR approach to minimums at night, and my risk will be massively elevated. I can learn to fly all this, but I haven’t. Put Peter (on the basis that all his experience is in a TB20) solo in a Pitts 12 and his risk will be much more elevated than in the former described scenario. This is why I don’t do the former, and Peter doesn’t do the latter. But you have to be able to know your limits.
I think risk compensation is a huge factor in low time pilots. (Almost) anybody who can get a PPL is not stupid and people are generally smart enough to play safe.
Unfortunately that (doing Bembridge etc every time) also brings the risk of giving up, via boredom and exhaustion of the pool of prospective passengers, and maybe that’s another factor in the accident rates being evenly spread. It’s a bit like one of the reasons why a tracker fund works pretty well is because the companies that do badly drop out of the index automatically. In aviation, many of the pilots who might get killed (due to various reasons) tend to drop out of flying due to boredom, pretty soon.
In 2012 I did a presentation on European VFR flying and one of the things I said was that people should make the go/no-go decision on “technical data” rather than on how they “feel”, because the latter is just a route to giving up. I don’t think everybody agreed
The more worrying thing is that there is also a subset of pilots who just couldn’t care less. I think most of them eventually get killed.
As regards the Pitts, I think my risk factor would be 100% before getting airborne