Re: Tom Knauff's newsletter

[posted and mailed]

tknauff@[EMAIL PROTECTED]
 wrote:
> Simply dividing the number of fatalities
> into the member****p gives a fatality rate that can be compared with
> other activities. The NTSB has web sites listing comparisons of
> fatality rates for many activities. Gliding is the worst of all.

I've done an analogous computation and your assertion above appears to
be surprisingly incorrect (or the source of my data or methodology is
highly corrupted - you decide.) By my calculations the metric you
mention actually shows aviation accidents of all types to be
approximately _over three times greater then gliding_. The single worst
accident rate in aviation appears to be helicopters, which is seven
times worse than gliding! 

My own conclusion is that gliding, by your suggested metric, is actually
safer than aviation in general! 

I'm afraid I can't reconcile my results with your assertion, since you
haven't provided specifics on your methodology or data. But for the
record (and critique) here's the methodology I used (numeric results are
at bullet point (7) for the impatient): 

(1) I used the NTSB accident query database as my data source for
accident counts: http://www.ntsb.gov/ntsb/query.asp

(2) I used the FAA Civil Airman Statistics web page to get an estimate
of the number of all active airmen certificates, those holding rotocraft
class, and those holding glider class (Tables 1, 7, and 8 at this site): 

http://www.faa.gov/data_statistics/aviation_data_statistics/civil_airmen_statistics/2007/


(3) In the NTSB web site I selected a large enough range that a single
bad or good year wouldn't cause unexpected outliers to dominate, but not
so large as to encompass long term trends. So I selected the 8 year
range of 2000-1-1 to 2007-12-31. 

(4) I elected to count only the number of fatal accident records, rather
than attempt to add up the fatalities in all the matching records. There
were simply too many accidents in the entire period to manually do the
arithmetic for this exercise (maybe some other time.) In any case, since
glider accidents rarely involve more than 1 fatality per accident, this
method actually should work against the gliders since other aviation
accidents, on average, involve more fatalities per accident. 

(5) As of 2007, the FAA estimates 590,349 active airmen certificates of
all cl*****. It estimates 30,853 pilots with rotocraft and 29,513 with
glider. Of those 29,513 with glider, 14,955 hold nothing but glider. 

(6) In the 8 year range, selecting only for "Fatal" accident severity,
then subselecting for "Category: All" I get 3407 accident records. For
"Category: Helicopter" I get 356 accident records. For "Category:
Glider" I get 46 accident records. 

(7) Using the above sources and method, my results are:

(A) For all aviation accidents:
   (3407 fatal accidents/8 years)/590,349 pilots =

   0.00072 fatal accidents/(pilot-year)

(R) For rotorcraft accidents:
   (356 fatal accidents/8 years)/30,853 =

   0.00144 fatal accidents/(pilot-year)

(G) For glider accidents:
   (46 fatal accidents/8 years)/29,513 =

   0.00020 fatal accidents/(pilot-year)

(G.1) If all the glider accidents were due only to those who could only
fly gliders: 
   (46 fatal accidents/8 years)/14,955 =

   0.00038 fatal accidents/(pilot-year)

While my computations seem to imply gliders are actually safer than
other methods of flying, the metric is of course highly suspect. 

Lastly, for the record, an Australian cross modal safety comparison
found that motorcycling was probably the least safe mode of trans****t: 

http://www.atsb.gov.au/publications/2005/pdf/Cross_modal_safety_comparisons.pdf


(Unfortunately gliders weren't broken out into their own category.)