Any statistical analysis can have nonsense results if you choose a ridiculously biased sample. The US census has various methods for establishing a sensible su

Any statistical analysis can have nonsense results if you choose a ridiculously biased sample. The US census has various methods for establishing a sensible subdivision of space for deriving statistics like this, including metropolitan areas, micropolitan areas, and combined statistical areas [1] The examples of Medina WA and penthouses in NYC don't fit any of those. 1. https://www.census.gov/programs-surveys/economic-census/guidance-geographies/levels.html#par_textimage_5 https://www.census.gov/programs-surveys/economic-census/guid...