Visualizing the Sustainable Development Goals
A look back at how far we’ve come on the world’s greatest challenges, with dataviz designer Maarten Lambrechts
January is the time of year when many turn their attention backward to help move forward. This year, looking backward to 2020 means reflecting upon a year that laid bare many gaps in economic and social development, exacerbated by the pandemic, while also revealing humanity’s remarkable capacity for cooperation and innovation. The new year also seemed like a good time, then, to check in with data visualization designer Maarten Lambrechts about his work on the 2020 Atlas of Sustainable Development Goals.
The Sustainable Development Goals (SDGs) are a set of 17 goals, adopted by the United Nations in 2015, that guide global action to tackle the world’s greatest challenges — challenges like healthcare access, poverty, hunger, gender inequality, and the climate crisis.
Progress toward the SDGs is measured by the World Bank’s World Development Indicators. The 2020 Atlas summarizes this data in 17 chapters, one for each goal, each packed with colorful charts and interactive visualizations.
The visualizations were developed by the four-person team of Maarten Lambrechts, Yaryna Serkez, Jan Willem Tulp, and Elbert Wang. I was able to sit down with Maarten in late November to chat about his experience working on such a large project. As Maarten explained, each chapter was assigned to one member of the dataviz team: he worked on Chapters 2–5, 7–9, 11 and 14. Outlines for each of the chapters — and later the chapter text itself — were developed by staff at the World Bank.
Here, Maarten shares his thoughts on some key themes, including designing as part of a team, considering data literacy, using data storytelling to inspire action, and his affinity for weird charts or so-called “xenographics” (Maarten launched the website xeno.graphics in 2017). Maarten’s responses have been edited for length and clarity.
On designing as part of a team:
Maarten Lambrechts: For the Atlas, Yaryna built a template that we — the dataviz team — could use to develop the chapters. (The template was Create React App, and the visualizations were rendered with React, though many use D3 helper functions for things like scales. We pulled in the chapter text from Google Docs and the data from Google Sheets.)
The dataviz team had a weekly meeting to discuss things about that template. We sometimes shared ideas, or someone would present on the current state of a chapter. We weren’t really collaborating much, just getting feedback and making sure that the design didn’t stray too far between chapters. We had some suggested color palettes, but we could use them quite freely. We tried to maintain consistency, but if you look closely, you’ll notice that some of the chapters use slightly different colors.
On automating workflow and “sketching with code”:
ML: To stay organized on a large project, you have to think carefully about the project setup. Data management can become a real mess, so you have to think about it upfront and avoid too many manual data transformations, because those will cost you time.
Early on, I switched from using the Excel and CSV files that people were sending me to using R to get data directly from the World Development Indicators API. All my data transformations were captured in code. I use the ggplot package in R to quickly explore different ways of looking at datasets. You can switch from bars to lines to other things quite easily; you can aggregate data; you can make maps. That’s the main technology I use when I want to explore ideas and make “visual sketches” with code. It’s important to sketch with real data, not dummy data, so you can see what is going to actually work for your dataset.
Because this was a project that spanned multiple months, some datasets were updated while we were working on the chapters, but the only thing that I had to do was run my script again. From the script I could also upload the data to Google Sheets, where it was grabbed by the chapter templates.
On considering data literacy while experimenting with non-standard chart types:
ML: The team at the World Bank was supportive of us experimenting with visualizations. This was the first time that the World Bank published one of their big reports online-only, so we wanted to make something that stands out. We were given a lot of freedom to design non-standard chart types, so in many chapters you’ll see things that are not line charts or bar charts.
We tried to make sure that whenever we used complex visualizations, we built up the complexity in steps. For example, we use a lot of scrollytelling modules, which often serve to explain the complexity of a chart. We first explain what the axes mean, then what the colors mean, then we add another layer of complexity with the data.
I have had experiences in the past where I would suggest a weird chart type that I thought might work quite well for data and the message, but the client’s reaction would be, “No, this is way too complicated for a reader.” We had none of that from the World Bank. The World Bank, of course, wanted to ensure that the reader was always on board, but they were open to the charts that we presented to them. The charts always had to fit the message — it was ultimately the shape of the data and the message we wanted to tell that determined the charts we used.
On using data storytelling to inspire action:
ML: The first thing that we tried to do with the Atlas was to lower the barrier for people to get into the data and the research by explaining things in a visual way. Then, on the editorial side, we tried to avoid making it all doom and gloom. You’ll see that we have a lot of bad news in the chapters, but we also have some good news. We try to end each chapter on a positive note, though in the year 2020 that was not always easy.
Of course, the Atlas is about global issues, and people can feel quite helpless when they read about problems that are so big. The Atlas is not meant to be a direct call to action; it’s more about bringing the numbers to as many people as possible, to hopefully increase awareness and inspire broader change. As an example, I have heard from people in education who are planning to use the stories and visuals in their teaching.
On learning about the state of global development from the Atlas:
ML: There are many interesting insights in the data. One of the things that surprised me is in Chapter 2, which is about hunger and undernourishment. The first charts in that chapter are simple area charts where I highlight regions where people are becoming more food insecure. Over the last couple of years, it appears that in many regions the trend is going up, meaning that more people are suffering from undernourishment or food insecurity.
I have to admit, much of the news in the Atlas is not good news, especially on biodiversity and climate change. We have a chapter on deforestation, and it’s kind of depressing. But there is also good news. You have many more people connected to electricity grids, for example, which is shown in Chapter 7. It’s not all bad news, luckily.
On adapting the project because of the pandemic:
ML: Chapter 3 is about health: how can you publish an article about health on a global scale and not talk about the virus? Initially, the Atlas was scheduled to be published earlier, but the pandemic hit the plan hard.
Many chapters had to be rewritten in some way to include charts or text about the coronavirus. Of course, many of the statistics that we use are from 2019, so the pandemic isn’t reflected yet in the data. Instead, we were mainly trying to find angles to say something about the virus in the stories. For example, in Chapter 3, we have numbers on the impact on tuberculosis. We know COVID-19 is taking resources away from the healthcare capacity to address other diseases, so the pandemic has an impact on those other diseases as well.
On visualizations that worked well:
ML: I’m proudest of two visualizations. Chapter 14 is about oceans, and it contains many maps. Standard world maps center on the continents, but for the maps in Chapter 14, I used a projection that keeps the oceans whole and splits up the land instead. That was technically challenging. Also, making it work on mobile was hard. There, I rotate the map so north is to the left. That’s a bit weird, of course, but for mobile sometimes you have to do things like that.
My other favorite is… it doesn’t really have a name. (That was something that the dataviz team needed to do as well — we invented names for custom charts.) In Chapter 4, we use a “shifted Marimekko.” It’s a Marimekko chart, but you also shift the bars vertically. This chart shows “learning poverty” in different regions of the world: in some regions, many children are out of school, but you also have areas where children are in school but not performing well. The chart shows the distribution of students across these classifications. It’s one of the charts where we explain the complexity through scrollytelling. It was a challenge to develop because the layout is kind of tricky and I had to calculate things myself, but I always like developing things that I haven’t done before.
On visualizations that didn’t work well:
ML: One visualization that didn’t work out was also for Chapter 4. The World Bank team wanted to include something about the impact of school closings from COVID-19. They found a study about school closings in Pakistan where the authors looked at school performance in children based on how far they lived from the fault line of a 2005 earthquake. You could see the influence of school closings on the performance of these pupils — if they lived far away from the earthquake, it didn’t really matter, but closer to the fault line (where schools were closed for many months), more pupils were dropping out or performing badly. We initially wanted to make a map to show how far people were living from the earthquake, but it proved too complicated. In the end, we used a fairly straightforward chart, nothing fancy. That was one chart I worked hard on, and it ended up being just a copy of the chart in the original paper.
On one of my (Claire’s) favorite visualizations, a globe depicting nighttime lights in Chapter 7:
ML: I also like that one very much, although I initially thought the idea (from Andrew Whitby on the World Bank team) was way too crazy. But in the end, it was not so hard to develop because this was one of the last chapters we worked on and we could adapt elements from the others. For example, we already had a drop-down module that I could use so that people could easily look up cities. There was a good React globe component that I could use almost directly. Andrew wrote the code to compare areas with similar populations, but with different levels of nighttime lights. The trickiest part was actually calculating the closest city when people click on the globe. In the end, it’s probably one of the most fun and interactive things that we have in the stories.
On xenographics:
ML: Before I launched the website xeno.graphics, I was reading about visualization and seeing charts on Twitter, and every once in a while, I would see something that I hadn’t seen before. I was always intrigued. I thought it would be nice to have a collection of these weird charts that could be useful for certain datasets or certain situations.
Then there was a call for speakers for the 2018 OpenVis conference, and I submitted a proposal to talk about xenographics. To build the collection, I scanned my Twitter feed; I scanned the archive of papers from the IEEE VIS Conference; I scanned Wikipedia for the term “diagram”; I scanned blogs to find specific visualizations; and I put them all on the website.
One issue in the area of visualization is that we have researchers doing interesting stuff, but it’s not easy to get that body of knowledge from the research to the practitioner. So, at OpenVis, I talked about how to get these xenographics more into the mainstream: the role scientists have to play and the role practitioners have to play.
Since then, I have presented some of the xenographics I found to clients — I even used some xenographic techniques in the Atlas. One of these is in Chapter 5. The chart — called an upset plot — shows whether countries have adopted gender equality reforms in different areas, like parenthood or the workplace, and how that overlaps with performance on another measure. Traditionally Venn diagrams show overlap, but the upset plot is actually better suited to communicate both overlap and quantities.
Many thanks to Maarten for sharing his insights and reflections. For more from Maarten (including his favorite xenographic!), be sure to check out this week’s issue of The ‘Gale.
Claire Santoro is an information designer with a passion for energy and sustainability. For 10 years, Claire has worked with governmental agencies, non-profit organizations, and higher education to accelerate climate action by communicating complex information in an engaging, approachable way. Claire holds an M.S. in environmental science from the University of Michigan.
