Introduction

We are pleased to announce the launch of our first interactive visualization, Diversity and Integration in America.

This visualization lets you see what areas are diverse and integrated and what areas aren’t. It begins at the nationwide level, as shown above. But you can also zoom in to very local areas and get statistics on any one of over eighty thousand individual U.S. Census tracts as shown in figure 1.

To explore the Diversity and Integration in America visualization start by looking at the country as a whole. This is the view you will have when you first load the page. See if the areas you expected to be diverse are diverse and if the areas you expected to not be diverse are not diverse.

Now try clicking on the small button at the top right. This switches the map colors from diversity to integration. You will notice that some areas are just as integrated as the are diverse, while in other areas the colors get a lot lighter, indicating that although those areas are diverse, they are not integrated.

Next, zoom into the area where you live. If you zoom in close enough, you can click on individual census tracts and you’ll see a popup like in figure 1 that tells you how diverse and integrated the tract is, and exactly how many residents of each of the racial and ethnic groups counted in the underlying data live in that tract. On some parts of the map, large numbers of adjacent tracts are about the same level of diversity. On other parts, you will find quite diverse areas very close to areas than are not diverse at all.

We hope you find this visualization interesting and informative.

How are Diversity and Integration Computed?

Diversity and integration are both numbers between zero and 100%. Diversity measures the degree to which people from different racial and ethnic groups live in a given area, like a city, town, zip code or census tract. Integration measures whether people of different racial and ethnic groups who live in a given area are uniformly distributed or segregated. For example, each group lives in their own neighborhood without any members of any other groups, then the area is not integrated.

Diversity

Diversity is defined as the likelihood that when two random people in a geographic area encounter one another that they will be from different racial or ethic groups. If there is one dominant demographic group in an area, then this is not likely to happen, so diversity is low. If there are lots of residents of many different groups in the area, then people are likely to encounter others unlike themselves, so diversity is high.

Figure 2 shows an example of an area with a diverse population:

Figure 3 shows an example of an area with a not very diverse population:

Integration

Integration measures the average diversity of a smaller area, such as a neighborhood, within a larger area such as a town. A town as a whole could be diverse, but if none of the neighborhoods in the town are diverse, then the town is not integrated. The diverse population we showed in figure 2 was also integrated, because there were people of each different group sprinkled all over. But diversity does not have to imply integration. Figure 4 shows an example that is just as diverse overall as figure 2, but not integrated at all:

The way we define diversity and integration, an area that is not diverse cannot be integrated. The diversity score sets an upper bound on how integrated the area can be. So the non-diverse population in figure 3 is not integrated either, even though the small number of individuals who are not members of the dominant group are spread out.

The Math Behind Diversity and Integration

If you are interested in more mathematical details, you can read all about them, and see the code we used to do the computations, in the divintseg open source package. The key thing to remember is that both diversity is measured from 0% (completely homogeneous) to a theoretical maximum of 100% (everyone is different). Integration is measured on a similar scale, but the maximum integration score an area can have is its diversity score. If an area is 60% diverse, it could be 60% integrated, but it cannot be 61% integrated. It could, of course, be much lower, possibly down to 0% integrated if it looks like figure 4.

Data

The data we used to compute diversity and integration come from the U.S. Census Bureau. The dataset we chose to use is the Decennial Census P.L. 94-171 Redistricting Data from 2020. In particular, we used two groups of variables called P1 and P2 that consist of counts of people who self identified as members of different ethnic and racial groups in the 2020 U.S. Census. In P1, only race is considered, not ethnicity. People are divided into groups based on their self-reported identification as members of one or more racial groups, such as Black, white, asian, American Indian and Alaska Native, and so on. In P2, people who identify as Hispanic or Latino are counted in a single group and those who do not are further divided into racial groups as everyone was in P1.

We also wanted to know the racial breakdown of people who identify as Hispanic or Latino. Combining the data from P1 and P2, we were able to compute this. The number of people who identify as being members of a given racial group who are Hispanic or Latino is simply the total number of people who identify as being in that racial group (from P1) minus the number of people who do not identify as Hispanic or Latino who identify as being in the racial group (from P2).

This Census data is available in aggregated form at a variety of geographic levels, from the nation as a whole down to individual blocks. We used data at the census tract and block level. A census tract is an area that typically has one to three thousand residents. A block will typically have on the order of a hundred residents. Overall, we looked at over eighty thousand census tracts and over eight million blocks. So on average each census tract consists of about a hundred block. We computed diversity at both levels, and used the diversity at the block group level to compute integration at the census tract level. This was done using the math described in the documentation of the divintseg open source package.. The interactive map we published allows you to zoom from the whole country down to the census tract level.

Give it a Try!

We hope you’ll give the interactive Diversity and Integration in America visualization a try. If you enjoy the map or if you learn something from it, please share it with your friends.

Update [Feb. 18, 2023]

The data behind the map is now available for download. See Diversity and Integration Data for details.