Human settlement densities get a lot of attention for two reasons. First, the data are usually easy to get; we know the areas and the populations of many places. Second, we rightly suspect that most people do their best work when they are near (but not too near) others who have worthy ideas and are doing good work.
Trouble is that most of the places for which we have data (countries, states, metro areas, even large cities) are too big. Human settlement densities vary considerably and averages are misleading.
Recently Sandy Ikeda and I considered densities for PUMAs (Public Use Micro-Sample Areas). Their average size is 100,000 and they approximate large neighborhoods.
Here are some of the fun facts:
For 2005 and for the top-25 PUMAs in terms of in-migrants with a Master's degree or higher, we found that these people migrated to parts of Manhattan as well as to areas such as Silicon Valley. These people were attracted to opportunities found in “low density” as well as in “high density” places. Four of these top 25 PUMAs were in Manhattan; four of the top 25 were in Silicon Valley; the other top-25 destinations were in West Los Angeles or suburban Washington DC, suburban Seattle, Austin or San Diego. The densest receiving area (in Manhattan) was thirty-eight times as dense as the most spread out (in Silicon Valley), yet each one succeeded in attracting many highly educated people. The areal sizes varied from below ten square km (Manhattan) to just over 300 square km(Washington DC suburbs) and one just over 250 square km (Silicon Valley).
There is no simple story even when the areas are relatively mall.