It’s an older article, but the point stands!

  • Pelicanen@sopuli.xyz
    link
    fedilink
    arrow-up
    4
    ·
    edit-2
    1 year ago

    if we divide each of those percentages by 50, we should come up with the odds of dying in a given vehicle per year,

    I’m being very nitpicky but this isn’t quite how it works, if you have a 90% chance of survival one year, you’d have 0.9^2 = 0.81= 81% chance of surviving two years in a row. With that in mind, the odds of dying should be relative to the fiftieth root of surviving fifty years, which gives:

    • Motorcycle: 0.478%
    • Car: 0.151%
    • Ferry: 0.065%
    • Amtrak: 0.0086%
    • Airplane: 0.0014%

    Without additional decimals it’s hard to see the change for the really small numbers but it doesn’t make much of a difference in reality.

      • Pelicanen@sopuli.xyz
        link
        fedilink
        arrow-up
        1
        ·
        1 year ago

        I have no idea about the statistics about motorcycle fatalities and for personal reasons I’d prefer not to get into them. I was just commenting on the way the statistics were calculated year-by-year with the assumption that the original statistics for fifty years were accurate. That being said, it’s possible that those statistics were not completely correctly calculated as well.

        • frostbiker@lemmy.ca
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          The fifty year statistics were also computed wrong, for the same reason you already explained. It doesn’t make much of a difference since the probabilities are so small anyway.