3 ms·

> The odds of that just being a coincidence are pretty damn low. To the statisticians of HN: I am genuinely curious. What are the odds of that? Do we model i

• tzs 8y ago
  In another thread, someone said that one accident per 10 million flights is normal, and that this new model 737 has had 150k flights. So, assuming an accident probability of 1/10000000 for each flight, and assuming that each flight is an independent event, then the probability of getting 2 or more accidents in 150k flights of normally functioning planes is 1/8978.23, if I've done the math correctly. The probability of 1 or more accidents in 150k flights is 1/67.1679. I have no idea how to interpret this. One of the challenges in statistics is asking the right questions, and I'm not sure that the questions I've answered are the right questions.
  • rjf72 8y ago
    Same answers on the math. Using a binomial I got 1/9009 for 2+, mostly same answer for 1+ in 150k. Probably rounding differences. Also agreed on asking the right questions. Could be a 1/9000 event given a 1/10million safety expectation, or it could simply be a much more probable event given that our assumption of 1/10million is incorrect. Ironically it could also be that the new plane is even more safe, but we've just experienced an e.g. 1 in 50,000 event. There's lies, damned lies, and statistics.
  • frutiger 8y ago
    Would it be worthwhile to compare to a daily risk that people take? e.g. what are the odds of getting hit by a car when crossing the road?
  • fulafel 8y ago
    A "one in 10k" coincident would qualify as mildly interesting at most. We run into coicidences all the time and I think it needs to be about 1 in 1000 to even register as a coincident.