How do you measure a city’s diversity? If a city has a lot of different racial and ethnic groups in their own segregated sections, is that diverse?

A blog called priceonomics recently ranked major American cities on diversity by looking at the percentage of major racial and ethnic groups within the city’s limits. The District of Columbia came in 21st, slightly less diverse than Oklahoma City.

Photo by Eric Hews, erichews.com posted with permission.

However, while this analysis is useful, it it doesn’t reveal whether the neighborhoods in each city are themselves diverse, or whether the city boundary just encompasses some all-black areas, other all-white areas, and so on.

If we modify this methodology to measure the average diversity of a city’s neighborhoods, rather than of the city as a whole, we are able to quantify how integrated these place are. On this new measure, the District performs even worse.

A neighborhood-level calculation changes the results

Consider Chicago. With roughly equal-sized black, white, and Latino populations, the Windy City ranks as the fifth most diverse city in the country on the priceonomics scale. However, if we instead use priceonomics’ same methodology (it took the percentage of black, white, Asian, Latino, and other people in the city, then used a Herfindahl-Hirschman Index to combine those numbers into a single score) for each of Chicago’s individual census tracts, then take the weighted average, Chicago suddenly drops to 38th out of 45.

Chicago, as a whole, is diverse, but its neighborhoods are not. The average Chicago census tract is less diverse than a typical tract in Portland or Colorado Springs, both relatively homogeneous cities that scored near the bottom in the original citywide index. Both are close to 70% white, but the non-white population isn’t all clumped in a small non-white area.

Chicago’s diverse population is largely segregated. Sacramento is diverse, and so are its neighborhoods.

Here are the scores for all of the cities in the analysis. You can click on a column in this table to sort it. Click on the name of any city in this table to see a map of that city’s Census tracts and their diversity levels.

Name New rank Diversity index

(neighborhood)

Original rank Diversity index (citywide) Rank change
Sacramento 1 0.324778 2 0.249345 1
Oakland 2 0.367682 1 0.234220 -1
Long Beach 3 0.404824 4 0.285367 1
Fresno 4 0.419840 15 0.332616 11
San Jose 5 0.419912 6 0.295963 1
San Francisco 6 0.420050 8 0.311810 2
Las Vegas 7 0.439887 17 0.338106 10
San Diego 8 0.468351 9 0.312709 1
Fort Worth 9 0.478435 13 0.324671 4
Albuquerque 10 0.482462 27 0.399064 17
Charlotte 11 0.495846 16 0.336905 5
Boston 12 0.496351 7 0.310615 -5
Austin 13 0.500739 19 0.371211 6
Oklahoma City 14 0.502493 20 0.374375 6
Virginia Beach 15 0.510308 33 0.455436 18
Raleigh 16 0.512390 22 0.381092 6
Houston 17 0.518201 10 0.313724 -7
Tucson 18 0.518248 26 0.399055 8
New York 19 0.520827 3 0.260531 -16
Jacksonville 20 0.524788 25 0.398575 5
Los Angeles 21 0.526386 18 0.339228 -3
Dallas 22 0.531520 12 0.321215 -10
Denver 23 0.536938 24 0.387416 1
Nashville 24 0.538537 29 0.408857 5
Seattle 25 0.545698 37 0.476876 12
Mesa 26 0.553278 38 0.492694 12
Phoenix 27 0.556085 23 0.384986 -4
Indianapolis 28 0.565431 31 0.422630 3
Columbus 29 0.565687 32 0.430533 3
Colorado Springs 30 0.565902 41 0.528233 11
Portland 31 0.569062 43 0.539417 12
San Antonio 32 0.574919 35 0.474636 3
Kansas City 33 0.579408 28 0.399368 -5
Milwaukee 34 0.589951 11 0.320661 -23
Philadelphia 35 0.599411 14 0.331147 -21
Washington 36 0.611801 21 0.378045 -15
Omaha 37 0.613333 39 0.501041 2
Chicago 38 0.632993 5 0.290745 -33
Louisville 39 0.656964 40 0.518145 1
Memphis 40 0.670075 34 0.474338 -6
Atlanta 41 0.670933 30 0.416675 -11
Baltimore 42 0.681552 36 0.475331 -6
El Paso 43 0.706141 44 0.663345 1
Miami 44 0.732796 42 0.536259 -2
Detroit 45 0.795764 45 0.674185 0
Note: Priceonomics used the 2013 1-year American Community Survey estimates for their analysis. This analysis uses the 5-year estimates, because it is available at both the Place and Census Tract levels. As a result, the citywide index scores may vary slightly from the data presented by Priceonomics.

California cities dominate the adjusted rankings, accounting for the top six spots: Sacramento, Oakland, Long Beach, Fresno, San Jose, and San Francisco. Virginia Beach moved up 18 slots, representing the largest jump of any one city.

DC, on the other hand, drops into the bottom quartile, neck and neck with Omaha. Like Chicago (well, not quite as bad as Chicago), the District’s citywide diversity doesn’t extend to diversity within most of its neighborhoods.

The diversity of each census tract in DC.

How citywide diversity relates to neighborhood diversity

There is a correlation between diversity in a city and diversity within its neighborhoods, although places like Chicago and DC remind us that it is not necessarily as strong relationship. Here’s a scatter plot comparing the citywide and neighborhood average diversity indices for each of the 45 cities:

Diversity within neighborhoods compared to overall city diversity, with the most integrated and most segregated cities labeled.

Cities above the trend line have less diverse census tracts than the city’s overall diversity would suggest. These are therefore relatively segregated. Chicago and DC fall into this category.

Miami is among the least diverse cities on the entire list (remember that according to this methodology, “diversity” only considers 5 distinct groups, lumping together, for example, everyone who identifies as Hispanic/Latino), but on a neighborhood level it’s even more segregated still.

Cities below the trend line have neighborhoods that are more diverse than comparable cities at their level of citywide diversity. This group includes Sacramento, which is both diverse and integrated, as well as Portland, which is not diverse, but relatively well-integrated.

Diversity and integration are both important, and the District has a long way to go on both measures. What do you notice?

Cross-posted at R.U. Seriousing Me?