• HarkMahlberg@kbin.earth
    link
    fedilink
    arrow-up
    7
    ·
    2 days ago

    Wouldn’t the total number affect accuracy of the rate? I think one chart in the article showed something like 700 proposals for low income areas and 100-200 proposals for high income areas. As N approaches zero, the rate of resistance or cancellation is a lot more sensitive to smaller numbers of events.

    • bryndos@fedia.io
      link
      fedilink
      arrow-up
      1
      ·
      56 minutes ago

      Looks like the total n c1,500 was for all known projects in the dataset. The quartiles were determined from within areas with known projects.

      So the Q4 range ( $133-250k median household), represents the 25% projects in the richest areas of areas that had projects in them. Hence n is 365/366 “projects” in each each group.

      If there are tracts with median hh incomes way above that, but no projects, then “resistance” rate is unknown or even undefined.