There has been much discussion in recent years extolling the virtues of ethnic diversity in cities, especially as it relates to the richness of urban life. Metropolitan areas like New York, Chicago, and Los Angeles offer a richness of diverse cultural and culinary experiences that few cities can rival.
Often, however, this richness at a metropolitan scale can give the wrong impression that these cities are racially integrated. We used the diversity measure built into our Snapshot API, which ranges from a value of 0.0 (no diversity) to 1.0 (complete diversity) with the national score at 0.52. Two counties which are on the diverse side of the scale – Sacramento CA and Cook IL – provide an interesting comparison when viewed at the local level. The two maps at the same scale and shading scheme provide an interesting and stark contrast between these two cities. The Sacramento map shows very limited areas of low diversity, primarily towards the affluent northeastern suburbs, but in general shows much higher neighborhood diversity than is found in Chicago. In Chicago, there are large areas where diversity scores are very low, despite the known diversity of the city as a whole. Sacramento is relatively integrated at the neighborhood level, whilst Chicago remains a diverse set of segregated neighborhoods.
This is a demonstration of one of the classic problems in spatial analysis, which is that the scale effect (the size of the units) and the zone effect (the particular geographic aggregation utilized) can have a tremendous effect on the results. In general, it is part of the problem known as MAUP, or modifiable areal unit problem, which was most famously written on by my friend and colleague Stan Openshaw, who wrote that “the areal units (zonal objects) used in many geographic studies are arbitrary, modifiable, and subject to the whims and fancies of whoever is doing, or did, the aggregating” (Openshaw, Stan (1983) The Modifiable Areal Unit Problem).
Analysts should always be cognizant of MAUP, which truly pervades almost all spatial analytics. Data is most often available for standard administrative zones (states, counties, ZIP codes), and yet using such zones for analysis can often result in maps which can mislead users because they are both inconsistent in terms of scale and because they are arbitrarily defined with respect to the data being analyzed. For example, a population density map at the county level will mask the dramatic differences between the western fringe and Mojave Desert areas of Riverside and San Bernardino counties in California. With the explosion of geospatial applications and the widespread dissemination of maps in business, it is imperative that we be aware of, and attempt to minimize, these effects in our work.
Gary Menger, President