This sort of thread would be remiss to not include the quote popularized by Twain about there being three types of falsehoods: "Lies, damn lies, and statistics"
Of course, in this case Twain is guilty of statistics. The group of people most likely to die (old, infirm, sick) are more likely to be in a bed than on a train. And the people in New York beds spend a significant fraction of the year there, while the people on the train are only there for only a few hours or days. If Twain made an honest comparison of deaths per hour across similar demographic groups, the bed would look a lot better than the train.
classic twain.. way ahead of his time..Its a pity we still interpret the statistical significance of events by their interestingness and not by the frequency and end result. Shark attacks get way more coverage in media than Deer accidents even though a deer is 300 times more likely to kill you than a shark (http://www.dalecarnegiecoaching.com/2009/06/which-do-you-fea...).
The location of one's death and the cause of one's death are not the same thing. It sure sounds nice...and there probably should be some limit on whether or not we 'care' about 'one-off' causes of death in the same way any sufficiently complex system is going to have undesirable outcomes. But that doesn't mean we should stop trying to avoid them: if I could save 26 lives by some simple means, I'd gladly do it.
Twain tweaks his wording to make a point. Of course you are more likely to die in bed then on a train, but dying on the train isn't the issue. Dying under the train is.
Amusingly enough, it's got its own Wikipedia page http://en.wikipedia.org/wiki/Lies,_damned_lies,_and_statisti...