• tburkhol@slrpnk.net
    link
    fedilink
    English
    arrow-up
    14
    arrow-down
    1
    ·
    3 days ago

    Standard deviation is meaningless for data sets, like wealth distribution, that are not normally distributed.

    • deadbeef79000@lemmy.nz
      link
      fedilink
      English
      arrow-up
      2
      ·
      2 days ago

      Wouldn’t it still tell you how concentrated the data are around the mean? Just not as reliably as for a gaussian distribution.

      • tburkhol@slrpnk.net
        link
        fedilink
        English
        arrow-up
        2
        ·
        1 day ago

        Things like wealth and income (and alcohol consumption) don’t even have useful mean. They’re monotonically decreasing distributions with most people scoring near 0 and a few people scoring extremely high. It’s more like an exponential distribution than a gaussian.

        In the US, mean household wealth is something like 3-4x the median, and if you plug the numbers into the standard deviation formula, you’ll get something close to the mean. Even greater. Maybe that informs a very sophisticated statistician audience, but if you tell the general public ‘standard deviation,’ they think “63% of people are within that of the mean,” and that is only true of Gaussian. For exponential distribution, mean±SD gets approximately all the samples.

    • jdr@lemmy.ml
      link
      fedilink
      English
      arrow-up
      2
      ·
      2 days ago

      The fuck are you talking about? It’s still a perfectly useful statistic.