All the world's a donut: Introducing a new way to visualize our ancestries and origins

Last year, my colleague David Jarman introduced a dataset combining several different questions from the Census Bureau’s American Community Survey into one massive spreadsheet giving detailed information on the ancestries and origins of the population by congressional district. What, though, can one do to make sense of more than 200 categories of information like this? Break out the colors, of course!

Example with every category given equal weight

The goal was two-fold: to be able to see broad patterns, akin to census race and ethnicity categories, as well as to highlight the diversity within these categories. The solution was simply to use color. Lots and lots of bright, shiny color arranged on a wheel—or as we like to think of it, a donut. The high saturation makes it easier to distinguish shades, and the shape is compact. Since the goal is qualitative interpretation, we’re not sacrificing anything this way.

Every category is assigned a unique color determined mainly by geography. Color families are roughly, but not precisely, aligned with census race and ethnicity categories. Starting at 12 o’clock and moving clockwise:

• Black or African-American is yellow.
• Specific sub-Saharan African nationalities are yellow-orange.
• Islands of the Caribbean are orange.
• Central and South America are shades of red.
• North American nations that predate European colonization vary from pink to fuchsia.
• Asia and the Pacific Islands are shades of blue and purple.
• Europe is green.
• White colonial identities, such as American, French Canadian, and Cajun, are lime green, just to the left of the 12 o’clock position.

Map of color assignments

Again, the goal here is not to be able to identify a category by its color (although with practice that is certainly possible for several dozen of the most common), but rather to identify color families and variation within color families. So first, what's the big picture? You can eyeball the relative importance of each regional heritage just from the color families, or, if you’re colorblind, from the location/pattern of shades (admittedly a much more difficult task; future use of these charts, though, will be accompanied by alternate means of interpretation as well).

In a few days, you will be able to find donuts like these for every congressional district in our Atlas of the 117th Congress—just head to the “Snapshot” column. Below we’ll discuss a few examples representing a variety of districts, with the district number in the middle of each graph, starting with a trio of starkly different locales:

From left to right, we have populations that are mainly white, split between white and Black, and mainly Black and Latino.

There's more to learn, of course. What specific ancestries are present in substantial numbers? It’s important because there can be huge variety in broad categories, and, from a practical standpoint when it comes to elections, they can often be linked to political preferences. Here are three more examples:

All three districts are about three-quarters Latino, but as the color patterns show, you don't want to think of these three Latino populations as the same.

The same is true even when immigration of the groups in question mainly took place decades or even centuries ago. The next three districts have very different political histories despite all of them being overwhelmingly white:

Communities with more recent immigration histories often have more localized populations, with substantial numbers in just a few districts. In these cases it’s often helpful to identify specific communities, as opposed to the umbrella racial categories typically used, as you can see just below:

This sort of disaggregation can be important, as community differences within a racial category can be just as significant as across racial categories. For example, the districts shown above all have a median income of about $70-80,000. But the median income of Hmong in MN-04 is around $55,000, while the median income of Filipinos in NV-03 is about $83,000, and the median income of Indians in NJ-06 is about $115,000.

Recent immigrant groups are not the only ones typically concentrated in just a few locations. Native American nations, which of course predate European colonization and subsequent immigration, are also usually localized. Below are three districts with a large Native American presence:

Politicians ignore the interests of these nations at their peril—partisanship is variable and not necessarily set in stone.

Sometimes, what’s important is not the identity of groups but simply the amount of diversity within groups. The next three districts are not only very diverse but include a broad range of different origins within each larger category, as you can see from the highly variable shading:

Although we cannot easily identify many of the groups in the above charts, it’s clear that there are indeed a large number of groups present in some sectors, a conclusion that has value in its own right.

Finally, all categorization schemes will have problems. Here’s some examples highlighting this:

Cape Verdean communities typically are co-located with significant populations of Portuguese-Americans. Based on geography, however, Cape Verde, a former colony of Portugal off the coast of Africa, is in the orange sector, while Portugal is green.

Likewise, Guyanese immigrant communities, typically ethnically Indian, are usually found where there’s large Indian-American communities, but India is blue and Guyana is pink (although arguably it could also be Caribbean orange).

On the other hand, we have the countries of the Middle East and Central Asia, where the concern is not so much colonization as much as inadequate census categories. Here, the use of geography is clarifying.

Again, to find the donut for your district or any other, please check out the “Snapshot” column of our Atlas of the 117th Congress, starting next week.

Source: Daily Kos