In The Death and Life of Great American Cities, Jane Jacobs describes the “intricate ballet”
necessary for the health of urban spaces: a constant cross-use between spaces and neighborhoods.
This “ballet” is today measurable with unprecedented granularity.
Using GPS locations from cell phone data provided by LiveRamp through Carto,
this project measures interactions between neighborhoods in Chicago.
It makes the case that the form of integration in the urban space is consequential,
both because it represents individuals’ daily trajectories — their actual lives —
and because it reflects communities’ social structures and capacity for control.
In the map below, I juxtapose two characterizations of neighborhood-level integration (left and center),
with an indicator of “social disadvantage” (basically education and income; right).
The clustering coefficient encodes the degree to which
destinations visited by residents of a neigborhood are also visited by the residents of those destinations.
In other words, if I visit neighborhoods A and B, do residents of A visit B?
In sociology, this characteristic is related to the concept of “social closure.”
A related measure, the nearest-neighbor interaction fraction,
is the fraction of mobility by residents within their immediate vicinity:
the smallest region around them containing 40,000 people.
This is constructed to reflect Jacobs's cross-use,
and it is closely related to the nearest neighbors weight matrix familiar in geographic analysis.
There is a clear relationship between these three variables:
mobility parallels in social disadvantage.
The suburbs and wealthier North Side of Chicago have high clustering and interactions, and lower disadvantage.
The reverse is true on the South and West Sides.
Along Lake Michigan on the South Side,
the privileged neighborhood of Hyde Park stands out as an outlier on all three maps.
Geographers take heed: conformity to Tobler's Law
that “near things are more related than distant things,”
is dependent on socioeconomic status.
A simpler way of visualizing this relationship is through a neighborhood-level scatter plot
between education or crime, and mobility.
Again, the strong (non-linear) relationship
suggests that these mobility measures are not simply collecting noise.
It is striking and important to recognize that people from different backgrounds
engage their immediate surroundings in measurably different ways.
However, the mobility concepts are neither conceptually nor statistically identical with these more-familiar measures of social status:
this is not simply a “round-about way of measuring income.”
I show in the paper that clustering and nearest-neighborhood interactions
in fact offer independent predictive power for property and violent crime rates.
As sociological theory would predict, better integration (higher social closure)
is associated with lower crime rates.
The methodology presented is distinguished by its reproducibility:
cell phone data are increasingly available and
allow for inexpensive comparisons of social life across time and space.
Copyright © 2018 James Saxon