• scarabine@lemmynsfw.com
    link
    fedilink
    English
    arrow-up
    6
    ·
    24 days ago

    Consider this: given two options,

    • 10 polls with no bias, weighted by only past performance
    • 20 polls, half biased, weighted by an arbitrary secondary metric to remove bias (which only trends the data towards an average)

    Which do you suppose would be more reliable?

    • sandwichsaregood@lemmy.world
      link
      fedilink
      arrow-up
      1
      ·
      24 days ago

      This is the paradox of polling, there’s no way to tell. Either could be better, both could be wrong, both could be right. There is no such thing as a poll with no bias, because the only way to take a tryly unbiased poll is to know the outcome of the race a priori.