• catloaf@lemm.ee
    link
    fedilink
    English
    arrow-up
    4
    ·
    6 months ago

    For random events, that’s true. But we are able to poll people before the event to see how it will turn out. With a big enough sample size, you’re able to get pretty close to actual results. After all, the election itself is just one big poll, not a die roll.

    • Ghostalmedia@lemmy.world
      link
      fedilink
      English
      arrow-up
      5
      ·
      6 months ago

      I say this as someone who went to school for this stuff and does a lot of surveying and statistics daily.

      This doesn’t work when you’re comparing things that are going to be neck and neck. In order forecast with very high confidence, with something that is neck and neck, you need a huuuuge sample size and absolutely perfect surveying conditions.

      The reason polls have been a toss up lately isn’t because the polling is bad. The problem is that the big races were also ways going to be nail biters, and we’re looking at the odds that a race will be 1% one way or another.

      The good polls have been pretty damn close to the final vote percentages numbers. The problem is that the variance needed to swing a win right or left is absolutely minuscule. We’re often talking about percentages that are less than 2%, or less than 1%.