Aviation Accident Statistics - Innumeracy Strikes Again!

[CraginS]

Can we PLEASE figure out how to educate the population in general, and the press in particular, on the basics of simple statistics and statistical analysis???

Today I read the following example of how to mis-use statistics and mislead by the unspoken, implied (and improper) correlation:
"According to the New York Times’ coverage on a recent study appearing in the December issue of “Aviation, Space, and Environmental Medicine”, over 1,000 people are hospitalized after sustaining aviation-related injuries annually. Reportedly, only 10.6 percent of these individuals were passengers aboard commercial airplanes, with 32 percent traveling in private airplanes, 11 percent injured while hang gliding, while another 28.9 percent of the flight-related injuries affected parachutists."

This is from the article,
Virginia Aviation Accident - Two injured after single-engine plane crash
December 26, 2009 (NewYorkInjuryNews.com - Injury News)
New Source: JusticeNewsFlash.com
http://tinyurl.com/yhsbzeh for
http://www.newyorkinjurynews.com/2009/1 ... 61888.html

Ask the average reader of this article what that paragraph tells them, and you are likely to hear something like, "Private planes are the most dangerous way to fly." Delve deeper into that answer, and you will learn that the reader will rank types of aviation safety by category (most safe to least) as commercial airplanes (10.6%), hang gliding (11%)*, parachuting (28.9%), and finally private airplanes (32%).

* Gee, it looks like hang gliding and airline travel tie for best safety! (NOT!!)

It is too bad that, based only on the numbers in the article, there is absolutely no validity to either of the described conclusions. In fact, the numbers as provided are incomplete; they don't even add up to 100%.

There are several potentially meaningful comparisons we might make to understand the relative safety of different forms of aviation. All of these comparisons depend on looking at rates of occurrence, not absolute numbers of occurrence. The news article only offered (incomplete) information on absolute numbers of people hospitalized.
Rates of occurrence that might provide us with meaningful conclusions on relative safety include nine or more different statistics for each aviation category:

1) Participation rate: Number of people hospitalized divided by the total number of people who took part in the activity.

2) Activity rate by count: Number of people hospitalized divided by the number of person-flights taken by all participants. (One flight involving three people, such as pilot and two passengers, adds three to the person-flight count.)

3) Activity rate by time: Number of people hospitalized divided by number of person-hours of flight time. (A one hour flight of three people adds three person hours to the count.)

4-6) Injury severity rates using each of the first three activity rates above (number of people, number of person-flights, and number of person-hours, but dividing each of those numbers into the number of person-days spent in the hospital.

7-9) Mortality rates using each of the first three activity rates above (number of people. number of person-flights, and number of person-hours, but dividing each of those numbers into the number of deaths.

I am sure the numerate (cf: literate) in the forum could describe several other meaningful statistics useful for comparing the level of safety of various forms of flight.

My point is that the news article included incomplete, extraneous, and misleading statistics in the report, which add nothing to the real value of the report. In fact, by reporting these numbers as they have, they set up the reader population to misinterpret and misunderstand the situation. Such reporting, is, in my opinion, beyond incompetent, and, in fact, malfeasant.

Holding a jaded view of the press in general, I see no value in calling the editors and publishers on this error. I expect they would reply simply that they report the facts, and leave conclusions to the readers ("We report. You decide."). Further, the numbers they cited from the New York Times, a tertiary report
http://www.nytimes.com/2009/12/22/health/22stat.html
based on a secondary source article in “Aviation, Space, and Environmental Medicine,” which, in turn, reports on the actual primary source study, are probably accurate.
(Article abstract is at http://tinyurl.com/yb4f8le for
http://www.ingentaconnect.com/content/a ... 2/art00001)

The fact that they are meaningless is inconsequential to the reporter and editor in this case.

I try to remember the maxim, "Never attribute to malice that which can be explained by incompetence." Applying that principal in this case, I do not believe that the reporter and editor were trying to mislead anyone. I think, simply, that they are too ignorant to understand the implications of what they reported. They would have been better off not mentioning the accident numbers at all. However, this was a short snippet of an article. I bet the editor told the reporter or the re-write desk to go grab a few aviation accident numbers and throw them into the article to beef it up.

Going back to my real rant here - can we please have reporters and editors educated well enough that they do not make such egregious statistical errors and inferences as we see here?

Oh, and can we please ask all hang glider pilots to avoid using these numbers to try to claim that hang gliding is as safe as flying in commercial airplane? Such claims will only embarrass us all.


I'm cross-posting this commentary on both the CHGPA Forum and the OZ Report forum.
http://ozreport.com/forum/viewtopic.php?p=79358#79358