This low probability of an accident actually happening suggests that the condition probabilities presented in the article have huge standard errors. This may make the differences between sections of the plane insignificant. Then again, there aren't standard errors per se because we should have the population of all accidents rather than a sample. Interesting....

My understanding is that the per-minute risk of car travel and plane travel are roughly equivalent. Per mile, of course, planes are a heck of a lot safer. But if you wear your seatbelt in your car, you should be willing to sit at the back of a plane, right?