The safety record of the airline industry is well advertised. They are extremely proud that their death rate is ten or twenty times less[1] than the rate for automobile travel. Recently, though, Andrew Weir published a book[2] which claims that automobile travel is actually ten times safer than airline travel!
How is this possible? Are they using the same data? Is someone lying?
Yes, they are using the same data, and no, no one is lying. As usual, they are playing with statistics. I am going to present an over- simplified set of statistics, and illustrate how we can draw such completely different conclusions.
Let's assume that 100 million airline passengers have taken a trip of 1000 miles, and 100 of those passengers have died in a plane accident. On a per 100,000,000 passenger-mile basis (the usual basis), that is a death rate of .1 (one tenth). For automobiles, let's use an average trip length of ten miles, and a death rate of 1 (one). That is, for every 100,000,000 automobile passenger-miles, one death will occur. That's not too far from actuality, and the round numbers will make our calculations easier.
You can see why the airlines prefer to calculate death rates on a passenger-mile basis. It makes their industry look like it is ten times (1 divided by .1) safer than automobile travel. What Andrew Weir did was to calculate the death rate on a per trip basis. For the airlines, this means .1 death per 100 million miles, or 100 thousand trips. That's one death in a million trips. For automobiles, this means 1 death per 100 million miles, but that's 10 million trips, for a death rate of only one tenth per million trips. Suddenly, the death rate for automobiles is ten times better than that for airlines!
Which do you choose? You have to believe both of them-- neither one is a lie, they're using the same data. Which one do you actually use?
It makes sense to use the airline industry figures. For a given trip, of say, one thousand miles, the death rate is ten times greater for an automobile trip over the same one thousand miles. Andrew Weir's statistic is essentially comparing an airline trip of a thousand miles with an automobile trip of only ten miles. If you're going to make a trip, the odds of dying while on the trip are much less if you take the airline.
But that's not the end of the story. Airline travel is typically at speeds ten times greater than the average automobile travel. On that thousand mile trip, you'd spend ten times as much time in the car as you would in the airplane--so, on a per hour basis, the death rate for automobiles and airlines is the same! The only reason you're ten times more likely to die in the car on that thousand mile trip is because you spend ten times more hours in the car.
Still, why would you subject your precious body to that death rate for twenty hours, when you can get away with only two hours by taking a plane? The simple answer to that is, that per hour death rate just happens to be the same as the rest of your life!
Actually, if you do the calculations, it comes out to maybe even half. The overall death rate in the United States is greater than the death rate, per hour, for travel in cars and planes. But the travel death rate just reflects accidents--it doesn't for instance include heart attacks while traveling--and, sure enough, heart attacks can kill as many airline passengers as plane crashes.
That pretty much explains why an ordinary well-adjusted human takes car and plane travel in stride, without too much fretting about the danger. As long as you take reasonable precautions, there is no greatly increased danger. In fact, those "reasonable precautions" were probably devised just so we could maintain a level amount of risk in our lives. Statistically speaking, of course.
[1] American Public Transportation Association
Fatality Rates by Mode of Travel, 1995-1997 Average
http://www.apta.com/stats/safety/natsafe.htm
[2] The Tombstone Imperative--The Truth About Air
Safety, by Andrew Weir, Simon & Schuster, 1999
http://www.newscientist.com/ns/19990807/flightinto.html