American Scientist

In episode two of D&I ComSci—American Scientist’s science-for-all podcast—we are exploring all things data visualization: why using visuals to represent measurements and numbers is so important; how scientists can visualize inclusively, speaking to audiences of different cultures, visual abilities, and scientific experience levels. We’re bringing in a diverse group of data visualization experts—Allen Hillery and Alice Feng—to speak on behalf of their experiences in inclusive data visualization and what they believe can improve inclusive data visualization moving forward.

Transcript

Jordan Anderson 00:00

This is a revised version of this podcast.

Science communication!
Inclusive science communication!

Here's some data. Humans buy 1 million plastic water bottles per minute. Let's take a second and think about that. We know 1 million is a huge number. But how do we interpret how significant 1 million bottles per minute is? Simon Scarr and Marco Hernandez, two data visualizers for Reuters news agency, helped us to visualize the scale of 1 million bottles per minute. In 2019. They designed a graphic called drowning in plastic that drops icons of 1 million bottles onto a three-dimensional space and compares the pile to a scale image of a person and a dump truck. Data visualizations—such as Scarrs' and Hernandez's—take numerical information and relate it visually through pictures, maps, and graphs. Viewers can better grasp the information presented because the human mind is much better at comparing concrete images and identifying patterns and outliers than it is at scaling numbers and statistics. Inclusive data visualization goes a step further. It makes sure that each visual representation takes into account how individuals might respond to pictures, maps, and graphs by considering those with disabilities and those whose cultures might affect how they interpret data.

Jordan Anderson 01:54

There are many resources that teach us how to account for visual differences. Font size, touch, and audio options and design style are options visualizers are considering to aid different communities’ comprehension of a visualization. But even accountability efforts can be implicitly biased. One example of this we discussed is colorblindness. Today, there are hundreds of visualizations out there that account for colorblindness. But when it comes to colorblindness, studies show most people who are colorblind belong to one group, that group being white men.

Blind and low vision are more prevalent in populations, yet visualizations are only just beginning to account for these communities. In such communities, even calling data visualization a visualization becomes non-inclusive. According to Josh Miele, a blind accessibility researcher at Amazon, who spoke at the visualization conference “Adapting Comics for Blind and Low-vision readers,” all visualizations must consider that blind is a diverse community. Those with nystagmus—uncontrolled rapid eye movement—for example, often identify as blind despite being able to interpret structure and color.

Nearsightedness, farsightedness, and ocular albinism can affect how we consider visualizing different graphics. For some, adding braille is beneficial; for others, dynamic coloring and audio options might be better.

When it comes to culture, reading left and right is just one factor. Color usage alone can evoke certain responses to a visualization depending on an individual's culture and environment. Red might evoke more of a passion or warning response, whereas green might evoke more calmer safety. Different cultures might even view the same color differently. In China, for example, white is worn at funerals and could elicit the same sort of response as the color black in America. With all that data visualization must account for, inclusivity might seem an incredibly daunting task even with current technology. But some scholars even decades ago made a mark with examples of inclusive data visualization that still remain valid today. One of these scholars might surprise you.

W.E.B. DuBois 04:02

Ladies and gentlemen, socialism and the American Negro.

Jordan Anderson 04:05

W.E.B. Dubois—most people who took American History have heard his name. He was a sociologist and prominent civil rights leader during the segregation era. But did you know Dubois was a revolutionary inclusive science communicator?

W.E.B. DuBois 04:20

And so, I changed from studying the Negro problem to propaganda.

Jordan Anderson 04:28

In 1899, while W.E.B Dubois was a professor at the historically Black Atlanta University, his friend, Black journalist and lawyer, Thomas Junius Callaway, approached DuBois with an offer to help him display African American progress since slavery at the Paris Exposition of 1900.

W.E.B. DuBois 04:48

Knowledge wasn't enough. But even if people were ignorant of essential matters, which they had to know, they wouldn't correct their actions.

Jordan Anderson 05:01

Slavery had just ended in 1865 with the ratification of the 13th Amendment. Barely a few decades later, Dubois was one of the only Black men to attend the Ivy League Harvard University. With Callaway’s offer, he knew he had the opportunity to make a massive impact, and decided to create sixty handmade visualizations to document the ways in which society still held back Black America. I interviewed Allen Hillery, a Black freelance writer, data literacy advocate, and expert on DuBois and his visualizations to learn more about Dubois’s role as an inclusive science communicator and why his visualizations are still regarded as such a strong example of inclusive data visualization even with all the accommodations of today.

Allen Hillery 05:47

The theory was that pretty much African Americans would have died out due to their inferiority. So, he presented this exhibit… to show growth in the community and he used different metrics like home ownership, educational attainment to make his case and that there had been growth in spite of the different, well, Jim Crow, pretty much, that was there, that would have hindered a lot of the basic things that other people were able to accomplish in America. So, what he did do was that he allowed the data to speak for itself. And he used the metrics that were common across, you know, all Americans. And he also was able to let his message be told with the visualizations—the color that he used, the font sizes. And also he had some photographs, and he also had a lot of the law—the law that was, you know, Jim Crow law to show that okay, and here is this achievement in spite of these laws. Instead of him maybe having a more emotional, you know, speech or presentation, he let—he balanced that out with data.

Jordan Anderson 07:00

DuBois’s visualizations used vibrant colors, word choice, and style that not only captured his audience, but also revolutionized data and statistics in a way comparable to the Scottish engineer, William Playfair, who invented the line chart, bar chart, pie chart, and circle graph between 1785 and 1805.

But style and other visual accommodations that we discussed earlier, such as braille or audio options, aren’t always enough for effective inclusive data visualization. DuBois’s visualizations were supported by the facts, the data itself. And this allowed his work to be much more effective.

Allen Hillery 07:44

So when you want to have an effective argument, you want to make sure that you are using credible sources, which he did. He used census data. He used some other demographic data. He actually used the law. He had words there. He had some emotion—he evoked emotion with the colors and also with the achievements—and then he also had the data.

Jordan Anderson 08:02

DuBois was a powerful communicator, but he did have other factors outside of visualization that made his work more significant. Despite being a Black man in the early 1900s, DuBois had strong credibility as a Harvard Ivy League graduate. He also tailored his visualizations to attract the dominant audience at the time. He used French typography and European language mnemonics despite the statistics he was using being inherently about the Black community. This approach isn't always the most effective. And next we're going to discuss why visualizing from the dominant perspective can sometimes present challenges to the data itself, which in all visualizations is the most important factor.

Last month, I met with Alice Feng, a data visualization developer at the Urban Institute for Social and Economic Policy Research, to discuss her work in inclusive data visualization. I told her I had just met with Allen Hillery to discuss W.E.B. DuBois, and how DuBois tailored his work to fit both the minority and dominant perspective. When I asked her whether or not the dominant perspective can sometimes skew data or the underrepresented experience, she presented a very current example of research conducted at the University of California - Los Angeles regarding the COVID-19 pandemic.

Alice Feng 09:15

Yes, there was a group at UCLA, I think they are the Center for Health Policy Research, I think. They were doing a lot of work focused on Pacific Islander communities. And I think especially with regards to COVID-19 pandemic, one of the areas that they've really focused on has been collecting data about the impact of the pandemic on this particular group, because, you know, Pacific Islanders, being a smaller racial ethnic group in the U.S., I think, unfortunately tend to get overlooked and a lot of times don't get captured very much in the data.

Jordan Anderson 09:54

Yes. So when we first met, you told me about how data had misrepresented these communities and overlooked members outside of what society designated underserved areas.

Alice Feng 10:03

So, I think it was a really, really interesting work. But yes, like you said, right, it's, it's very much focused on one particular community. And that's not to say that there aren't other groups within the U.S. whose experiences have also been overlooked, marginalized, not captured at all in the data. And so, I think there's definitely a balance there. I think they've not only like collected data and done some cleaning and transformation to make it more accessible to others who might be interested in this data. I think they've also built some visual products about it as well. And they definitely have, it sounds like, forged really strong connections with that community. I remember them telling us about how they have strong partnerships with local Pacific Islander groups. How they went about building those relationships with those groups. Things like: the importance of, you know, reaching out to these groups through some sort of mutual connection, having somebody who can introduce you to another, not cold calling them; the importance of meeting these people in their locations—going to them rather than having them come to you.

Jordan Anderson 11:24

But why go through all of the trouble to build these connections, when similarly to DuBois’s time, it's the dominant community that tends to have influence and can impact society to advance these communities?

Alice Feng 11:36

I think it often just really comes down to what is your message or the story you're trying to tell? And who are you trying to tell it to, right? So, if you have if you know from the outset that you have a very explicit goal of I'm looking at this one particular group, I wanted to understand them and I want my research to be primarily used by this group, to be found useful by this group, then, in that sense, I would say that, yes, it makes sense then that you're, like this data that you're collecting, this dashboard that you're building will just focus on the segment that the Pacific Islander community and maybe not capture other racial ethnic groups in the U.S.. But yeah, if you are, if your focus of your project is very much as on, you know, understanding the impact of COVID on a national scale, that it behooves you, yes, to be as inclusive as possible.

Jordan Anderson 12:31

Today, there are more efforts than ever to bring new voices and perspectives into science communication. We've seen a ton of efforts to raise awareness of diversity and inclusion in post-George Floyd alone. And so, the question now becomes “How well are we maintaining diversity and how comfortable are underrepresented individuals in these spaces?”

There are a number of sources out there that suggest minority communities tend to feel uncomfortable working with science or in scientific spaces, including in the tech world of data visualization. Last episode, we touched on that a bit by talking about science's exclusive history and how in the past science has been both privatized and used to inhibit various communities. But how much does this hold true in the scientific field now?

Alice Feng 13:22

You know, the inclusivity means you kind of baked into the entire research process, or the data analysis process, right? Creative or eye-catching visualizations out there, those are the kinds of things that I think the community tends to focus most of its attention and discussion and kind of excitement on. And so things like inclusivity, or things like accessibility as well, I think, which is a very closely related concept, have unfortunately kind of been not forgotten about, but just something that most people… it has been a lower priority. If your data is biased, if it is missing certain groups, if it was, you know, produced under kind of a really flawed, bias methodology, if the analysis you did on top of it was itself also biased or flawed, you know, making your data viz inclusive at the end isn't going to fix those sorts of problems, right? It's not going to magically solve, like, racist analysis or racist data just because you use the right labels in your graph, or use the right colors in your graph, right? So like, if you truly want I think to embrace inclusiveness, it has to start at the very beginning, and it has to be a value that's upheld throughout your entire project, not just at the end when you're trying to visualize and communicate your results. So, that I think also goes back to you know, making sure that the people who are doing the data collection who are doing the analysis are also diverse, hopefully. At a minimum, at least aware of, you know, what are their identities and what privileges and biases might that cause. Being also an organization that itself also values diversity, equity inclusion is really important. I mean, I have a hard time imagining an organization that doesn't uphold those values can produce inclusive, effectively, truly an effectively inclusive communication products at the end. So I think that's the other thing I would just keep in mind is like, don't just be inclusivity in data viz in isolation from the rest of the process that has to, you know, that has to happen for you to even make that data visualization.

Jordan Anderson 15:40

As we approach the end of this episode, I want to reflect on our conversations with Allen Hillery, and Alice Feng. It's important to remember that inclusive data communication involves understanding your audience and recognizing the diversity within individual visual experiences.

Inclusive science communication involves thinking about these spaces and fostering the sorts of connections with individuals representing these communities. Data visualization is just one example of how scientists can reach audiences by empathizing and with communities and individuals outside of their own, and by creating compelling and dynamic artwork that is both captivating and persuasive. In future episodes, we'll continue to explore other ways scientists can make their work more inclusive, and the nuances that may cause these techniques to be more or less relevant in these spaces.

Today's episode of D&I ComSci has been brought to you by American Scientist and Sigma Xi, The Scientific Research Honor Society. Special thanks to Allen Hillery and Alice Feng for speaking with me today. Today's music choices come from Mr. Simmons "Group Selection" and Free Music Archive. If you want to hear more from Mr. Simmons, please be sure to check out rapguide.com also mentioned on our previous episode. Also, be sure to check out our last episode, "Science and Hip Hop: Using Music to Communicate Science" on American Scientist dot org. If you like what you heard today, follow American Scientist or follow me on Twitter @Jordan_artsci. I'm your host Jordan. Thanks for listening!