Saturday, December 1, 2007

Getting Real

Yesterday we got lined up in our final jump manifest formation. As in, the lines we'll be in when we load the planes and the order in which we'll jump out of them. So yeah, it's finally getting real that I'm going to be jumping out of a plane on Monday. I'm remarkably unperturbed by that thought, though the fact that I'm not worried now makes me concerned that I'll be more nervous than I now expect when I get on the plane. If that makes any sense at all. In any case, there are concerns, but mostly realistic, manageable ones. The course averages about five injuries (serious enough to prevent finishing subsequent jumps) per class of 360 or so, and that's in five jumps. So each jumper faces a 0.2% chance of significant injury per jump. Pretty minimal, and since most injuries are jumper induced, and I'm performing somewhat above average at the basic tasks, I'm confident I won't be one of those five.

3 comments:

Shane said...

As the resident stats nazi, I had to point out the errors in your math. If you assume that every jump is an independent event with no effect on any previous or subsequent jumps, then the probability of an injury on any given jump is .2793%. You have underestimated your risk by 28.4%!

If each event has independent probabilities, the probability of something happening on any of the events isn't found by adding the probabilities of each event, it's done by calculating the probability of 1-P(not getting hurt on a jump)^5 (for 5 jumps). Basic algebra will take you the rest of the way, since you're finding the P(getting hurt on a single jump), given the P(getting hurt on any of 5 jumps).

Jeez I'm a dork.

And I don't mean to further dampen your morale, but were you one of those guys who fell off that tower at the confidence course in basic? I know Harris did, but I can't remember who else it was.

Evan said...

If by "fell off the tower" you mean "was dropped off the tower by the incompetent monkeys I was forced to rely on" then yes. That was me. I did learn very well the lesson on exactly how far I can trust my fellow soldier. About as far as I can throw him, incidentally.

Thank you for correcting my math. I was in a hurry because my (stupidly expensive for an MWR 'service') internet time was running out, so I decided to just take the number and run. I should at least have rounded correctly, though.

... And I'm wondering, how does the fact that a given subject's injury prevents them from completing subsequent jumps (at which they would have had a presumably higher-than-average risk of injury, since injuries aren't entirely random) skew the numbers? This isn't the evening for this sort of pondering.

Also, Delta company broke a new Airborne school record, having 10 students dropped from the course for disciplinary reasons over a single 2-day weekend, ranging from various alcohol infractions (underage, DUI, or breaking the pre-jump prohibition) to having a pair of loaded .45s in the barracks, which they found during the Health & Welfare they did on Sunday night searching for the laptop and iPod that someone stole. We are awesome.

Shane said...

My brain hurts. I've been trying to intuitively figure this out without the help of Google and the (I'm not making this up; I actually brought this to Alaska) college Statistics textbook I have about 3m away in a box of books I no longer read.

It shouldn't change it too much, because if each event is perfectly independent (which it isn't), then the chance spreads itself evenly through all jumps. But one would expect that more injuries happen on the earlier jumps because the clumsy people who would hurt themselves will have already done so and eliminated themselves from the subsequent jumps. But if you make the first jump higher risk of injury, then you'll have to adjust the others to some sort of Bayesian "chance of injury, GIVEN THAT jumper was not injured on the first jump." But it doesn't matter because the average per jump risk still comes back to the original number. The only question is whether the average is a meaningful number.