I recently ordered the book "The Killing Zone" and am looking forward to reading it. Most of the reviews were positive, but one reader complained that the author made the mistake of using accident frequency counts, rather than accident rates. What is he talking about?

MAIN QUESTION: Are the statistics in “The Killing Zone” appropriate and scientifically correct?

It would seem to me that the natural statistics would be fatal accidents per flight hour. Does the book not use this measure?

I have read elsewhere that the highest accident rates are those involving pilots having between about 200 hours and 800 hours. Does that come from this book?

Interestingly enough, my personal experience with fatal accidents does not correspond with this. Every one of the 8 or so pilots I have known (indirectly in all cases) that have been killed in an accident were high time pilots (> 1500 hours).

  • 1
    $\begingroup$ You could just read the book and find out :-) It's a really good book, as far as I remember the numbers are based on accidents per hour and the highest accident rates are at 200-500hrs (or so). That 'window' is exactly the killing zone in the title. $\endgroup$ – Pondlife Feb 29 '16 at 17:10
  • 2
    $\begingroup$ If your questions are really, "What is he talking about?" and "Does that come from this book?", then I think you should read the book rather than posting here. $\endgroup$ – abelenky Feb 29 '16 at 17:21
  • 1
    $\begingroup$ My primary question is whether the statistics in the book are appropriate and scientifically correct. The secondary questions I can eventually answer by reading the book. I have bolded my primary question to make it clear. $\endgroup$ – Tyler Durden Feb 29 '16 at 17:29
  • 5
    $\begingroup$ If your question is really about the statistics in the books, I'd ask some statisticians about the scientific validity, not a bunch of pilots. $\endgroup$ – abelenky Feb 29 '16 at 17:32
  • 1
    $\begingroup$ The data can be statistically valid and "scientifically correct" whether the results are presented as absolute accident rates (number per year), rate per X-hours, or rate per-mile. Which metric is "appropriate" depends on exactly what you're trying to study (or alternatively how good you're trying to make yourself look: Accident rates per passenger-mile can look a lot better than per-flight-hour numbers, and per-flight-hour numbers can look a lot better than absolute occurrence rates...) $\endgroup$ – voretaq7 Mar 1 '16 at 7:02

I did a bit of reading and found the review questioning the book. That review was written by William R. Knecht.

Knecht has provided in that review a link to one of his papers. (Payment is required but the abstract is free to read anyway.)

Instead a bit of searching led me to this paper also by William Knecht which I believe covers similar material.

His argument requires a lot of specialist statistical knowledge to understand, but it is worth a read through, especially the Introduction and the Discussion sections.

There are huge numbers of low-hours pilots, and their accident counts swamp the true pattern of data (my interpretation). The high-hours pilots have some accidents too, but the data there is noisy due to the relatively low number of these pilots. (Poor data sources are mentioned too. See the paragraph beginning "We end with an appeal to the FAA and NTSB.") A more sophisticated analysis of this noisy data would seem to suggest that the "Killing zone" is much wider, reaching up to pilots with a few thousand hours, maybe more.

Disclaimer: we would probably need to be professional statisticians to understand every detail of the argument. Everything I've written above is my own interpretation, as a non-specialist.

  • 1
    $\begingroup$ I have access also to the first paper. I can confirm that it presents basically the same plots/data as the freely available one, made exception for the esposition style adjusted to the different audience. $\endgroup$ – Federico Mar 1 '16 at 10:15

Not the answer you're looking for? Browse other questions tagged or ask your own question.