• @Contramuffin
      link
      English
      56 days ago

      Not starting at zero is a common practice in science and data processing. The difference between bad and good data visualization is in relevance. Good data visualization starts an axis at non-zero numbers because the fluctuation is more relevant than the zero. Bad data visualization hides relevant data to present an alternate takeaway.

      Here, a change in birth rate of even 0.1 or 0.2 is a major societal change, and showing that change is more relevant than showing the zero (how would it even be possible that there were zero births in a year, anyways?)

    • LabPlotOP
      link
      fedilink
      6
      edit-2
      6 days ago

      @coucouf @europesays @[email protected] @dataisbeautiful

      Let us reply by quoting Howard Wainer. In his well-known paper “How to Display Data Badly” he wrote:

      “A second way to hide the data is in the scale. This corresponds to blowing up the scale (i.e., looking at the data from far away) so that any variation in the data is obscured by the magnitude of the scale. One can justify this practice by appealing to “honesty requires that we start the scale at zero,” or other sorts of sophistry.”

      • coucouf ⏚
        link
        fedilink
        16 days ago

        @[email protected] @europesays @[email protected] @dataisbeautiful very nicely framed but unconvincing. I can reverse the argument and call it sophistry. Data visualisation inherently conveys a message and you can’t avoid having to choose what you want to say. Besides most data has a confidence level so at some point by blindly applying this “principle” you’re just showing noise.

        • LabPlotOP
          link
          fedilink
          1
          edit-2
          5 days ago

          @coucouf @europesays @[email protected] @dataisbeautiful

          Thank you for your comment. For these types of charts describing variation in data, which also include upper and lower limits on the values that contain probable noise, not using 0 at the start on the y-axis makes sense, as it makes it easier to analyze this variation and detection of potential signals.

          We believe that Howard Wainer certainly would not recommend blindly applying this principle to all cases.