What to consider when choosing colors for race, ethnicity, and world regions

Data on race and ethnicity is often important. (Especially U.S.) newsrooms, NGOs, and others should and will use data to show that racism exists — until the day it doesn’t. But just as with data visualizations on gender, visualizations of race can come with subconscious biases and even reinforce racist stereotypes. This article explains what to keep in mind when choosing colors for a visualization with racial categories – so that all readers feel respected.

Avoid stereotypical skin colors
Avoid brown and olive
Avoid gray to diminish
Avoid using blue for Europe or white people
Consider less saturated colors
Keep shuffling your colors

Two disclaimers: First, take the ideas in this article as food for thought on how people might respond to your colors. Keep them in mind, but don’t be intimidated by them. Depending on the visualization type you’re choosing, and how you phrase the text in and around your visualization, almost any color can work.

Second: As a white German, I’m often uncomfortable talking about race. In German, the word is strongly associated with colonialism and Nazis — and as someone who grew up in a very homogeneous environment, I still feel naive about race from time to time. I’ll try my best.

Avoid stereotypical skin colors

In almost all cases, it’s insensitive to represent racial categories using stereotypical skin colors. Avoid using black for data about Black people, white for white people, yellow for Asian people, etc.

At the beginning of the last century, that was still widespread:

It was also considered okay to color Asia as yellow ⬤, Africa as black ⬤, and Europe as white ⬤:

But skin color doesn’t define race; race is a social construct. We may talk about “white” and “black” people, but our actual skin colors are hardly a differentiator, as the beautiful project Humanæ shows which you can see at the top of this article.

Nowadays, most visualizations about world regions and race use colors that have little to do with stereotypical skin colors, and rightly so:

An exception to this rule is when white, gray, or black is a backdrop to what’s actually important in your visualization. In the chart below, the “white” category is the largest but also the least important, and hence colored with a warm light gray ⬤ . The chart is about how people of color are catching up, so they’re shown in more saturated colors ⬤⬤⬤⬤.

Using white, gray, or black should stay the exception, though – you can find the reasons for that further down. If in doubt, ask coworkers or friends for their opinion.

Avoid brown and olive

Certain shades of brown and olive can also be remindful of skin tones. What’s more, people tend to like them less than other colors in general. Consider avoiding brown and olive — people of any race might be unhappy to see data about themselves in these colors.

Avoid gray to diminish

As Jon Schwabish and Alice Feng point out in a “Do no harm” guide, a useful question to ask when designing data visualizations is: “If I were one of the data points on this visualization, would I feel offended?” Making sure that data subjects feel respected is always important, and especially so when visualizing sensitive topics like gender, religion, or race.

That’s why using gray for “other” or “multiracial” categories can be inappropriate. You’re visualizing data about people, after all, and gray can communicate “the people in this category are not as important as the others.”

The following charts all give their “Other” or “Multiracial” categories an equal visual importance:

Avoid using blue for Europe or white people

It might be tempting to represent Europe in blue. Europe’s flag is blue ⬤. It’s represented as blue in the popular board game Risk ⬤. The ring in the logo for the Olympics representing Europe (according to the IOC in the 60s) was a dark blue ⬤. Most people in this little poll of 62 people said they’d assign a blue to Europe ⬤.

But blue is also often used to represent professionalism, competence, even royalty. Using it for Europe — or white people — can reinforce the absurd idea of a superior race.

If you can, consider giving blue to another racial category:

Or simply don’t use a (strong) blue in your palette at all:

If blue is a must, consider shifting the hue in the direction of purple or turquoise, or showing more than one race in different shades of blue:

Consider less saturated colors

Most saturated (crayon) colors have strong associations: We’ve seen that a strong dark blue can mean competent and royal. A green can be interpreted as positive, or right, while a strong red can mean danger or important. You shouldn’t want any of these adjectives to be associated with any of your chart’s racial categories.

A way to get around this and still use the full palette is just to tone the colors down and shift their hues slightly. A pale red — almost rose — ⬤ doesn’t have as strong an effect as a vibrant red ⬤.

Keep in mind that these less-saturated colors can be harder to distinguish for colorblind people, and that bright, pastel colors might fail color contrast accessibility tests. To get around that, you can work with darker or more saturated outlines:

Keep shuffling your colors in different projects

Almost all the guidelines in this article can be summed up with the following: Don’t lean on stereotypes or go with the first associations that come to mind, especially if they’re unconscious (“It just feels right to use yellow for Asia”). Instead, ask yourself why you have them, and go against them.

But all colors come with some associations. Which is why no race should be permanently linked to one specific color. So if you find yourself picking colors for racial data again in another project, consider a different set than last time. Your Black category came in pink last time? Maybe this time, go with turquoise. And next time, with a dark blue.

---

Thanks for reading! Have you found good examples of visualizing race that this article is missing? I’d love to see and include them. Send me a link to lisa@datawrapper.de, or mention them in a comment below.