Sunday, October 12, 2014

(F)lying With Statistics -1

Another discovery arising from the previously reported packrat environment in which I live has been the source material for an article/essay I’ve been nursing for 29 years.
I now know it’s 29 years because I’ve found the yellowing clippings from The Los Angeles Times that started it all, and they’re dated September 1, 1985.  Finding them puts me in a position to make some corrections, which I’ll append to this post, but I’ll have to break the post into two parts; no one is going to read a 1200-word blog post, or not one of mine, anyway.  
“(F)lying With Statistics” is a damning expose of the commercial airlines’ shameless use of “passenger-miles” to reassure us of how safe it is to fly. (The “fractured-flyer miles” part of the article is my contribution.) With a bit of statistical counterpunching and an appeal to logic, I leave the airlines’ argument bleeding on the tarmac. I’ve refrained from going public with the material until now because it could cause panic amongst the traveling public – the next Y2K, maybe, and we know how that turned out. (If we don’t, we can go back to the July 27 post on this blog.)
Over its near-three-decades the piece has seen use as an essay, a tongue-in-cheek SEO article, and one installment of a “Private Eye” series. The blog seems the logical final resting place for it.

        Statistically, Flying Is the Safest Way to Travel

         -- but do you really want to be an air travel statistic?

I think it was Shelley Berman, in one of his comedy routines, who commented that the airlines could always prove that "flying is the safest way to fly."

It's the old "passenger-miles" ploy that does it.

The strategy is obvious, isn't it? With commercial airliners seating hundreds and distances in thousands, you multiply "passengers" by "miles" and one successful trip adds a couple of hundred thousand markers in the "win" column.

The argument goes, “The record of America’s scheduled airlines has averaged out to about one fatality per billion passenger-miles.” That statistic is then translated into a 1 in 10,000 chance of being killed. (Odds of being mutilated, disfigured, or maimed aren’t posted.)

So, are less successful flights reported in fractured-flyer miles? (To the airlines' advantage, the numbers would be smaller, since these people didn't get to finish the trip.)

No, one accident is one accident. Accident statistics -- the bad numbers -- are reported in terms of individual events. I've seen it argued by an air safety expert that except for the people involved, keeping score in terms of people killed is “a meaningless statistic. What counts is the number of individual accidents,” number of people on board being random.

Fair enough; but let's apply the same standard to the safety statistics -- the good numbers. If "people on board" is irrelevant in reporting accidents, why does it become significant when reporting on safety? The fair comparison to "number of individual accidents" is "number of individual flights without accidents." Tell me how many flights touched the ground only when and where they were supposed to. But that would bring the safety numbers down out of those reassuring billions.

Next post: Part 2, and I rest my case.