06 July 2022

The Making of the Do No Harm Guide

In today’s increasingly data-driven world, it is important that the way data are presented and communicated reflects the spirit of diversity, equity, and inclusion (DEI). The Do No Harm Guide1 is a resource for data analysts and communicators that explores how data can be visualized in a thoughtful and inclusive manner. It discusses how various aspects of data visualization design, from language to data ordering to color selection, should be utilized to not perpetuate biases, stereotypes, or other types of harm.

Though the guide was released in June 2021, the idea and motivation for creating such a piece started much earlier. In 2019, the data visualization team at the Urban Institute, a nonprofit public policy research organization based in Washington, DC, sought to update its data visualization style guide, a document that defined styling requirements for charts and maps made at the institute. At the time, I was working as a developer on Urban’s data visualization team, where I applied my knowledge of data visualization best practices to build interactive, data-driven narrative features and data tools while also reviewing static charts and maps produced for research publications and blog posts. To lead the effort to update and expand the style guide, I was paired with Jon Schwabish, a senior fellow at Urban and renowned data visualization expert whose specialty is in improving clarity in data communication.

From the outset, Jon and I knew we wanted to not just update the advice and guidelines on how charts should look but also add entire sections and make the style guide a more complete and holistic document. One of these new sections was going to tackle how to incorporate DEI thinking into the way data are visualized. We did not want researchers at Urban to just make “technically” good charts – we wanted them to make ones that did not perpetuate biases or reinforce stereotypes. However, as we looked to write this section, we realized existing resources and thinking on this area of data viz were very thin. We were unable to turn up much in the way of research or guidance on this issue compared to other topics in data viz, so progress languished for a while, particularly so after the start of the COVID-19 pandemic.

However, when George Floyd was murdered in May 2020 and racial justice protests erupted across the US that summer, Jon and I realized we could not continue to delay and thus decided to take a first stab at articulating some thoughts on creating equitable and inclusive data visualizations. We consolidated ideas from the handful of resources that we were able to find and added a few of our own, drawing on our experiences of working in the field of data viz to produce a short four-page paper. We felt that doing this was our way of contributing to the moment of racial reckoning sweeping the nation in our area of expertise. Simultaneously, Urban as an organization was also reflecting on how to improve its own internal DEI efforts and provided a backdrop that resonated with this work we were doing.

The success of our first paper led to an opportunity to dive deeper and expand upon our initial work thanks to funding from Tableau, a visual analytics company. To expand on our initial paper, Jon and I quickly realized we would need to speak directly with people who were thinking about issues related to equity or bias in data, so we decided to conduct a series of interviews. Our list of people to reach out to started with the authors of the pieces we cited in our original paper but soon expanded to include others whose work we felt were relevant, ranging from authors of books on social justice in data science to individuals leading DEI efforts in data-driven organizations. We reached out to a variety of folks representing a wide range of industries including academia, journalism, and the forand nonprofit sectors, asking if they would be willing to sit down for an hour-long Zoom interview with us.

To prepare for these interviews, Jon and I drafted a brief agenda with a list of questions we wanted to cover. Our questions reflected our desire to explore new topics that went beyond the look and feel of charts, topics such as the process of creating an equitable data visualization, how to build relationships with communities reflected in these visualizations, and the role organizational culture plays in producing inclusive communications products. Even at this initial stage, we had a sense that we wanted this project to take a wider lens to the issue of DEI in data visualization than our original paper did.

As we reflected on our interviews and started an outline based on the themes and points we had synthesized from the interviews into a full written draft, Jon and I quickly came to agree that we wanted “empathy” to be the key overarching theme that would undergird this guide. We felt it was critical that data communicators and visualizers remember that the data they are using represent the lives and experiences of actual human beings and thus to be conscious of how the charts and maps they create would be received by those whose lives were reflected in them. As one of our interviewees, the journalist Kim Bui, asked, “If I were one of the data points on this data visualization, would I feel offended?”

Jon and I wanted this guide to be a practical and actionable resource with takeaways that people could easily incorporate in their own work. We did not want this project to be an academic thought piece laden with dense, inscrutable theory that would be ignored by practitioners. Thus, we sought to include as many concrete examples as possible to illustrate our points so users could better absorb what we were writing about.

We sent a draft of our document to the people we had interviewed as well as other reviewers at both Urban and Tableau to solicit feedback and then started the polishing stage of revision, design work, and copyediting. Finally, after all of this work was done and almost a year later, the Do No Harm Guide was launched on June 9, 2021!

Working on the Do No Harm Guide was a very meaningful experience filled with surprises that I learned so much from. First was the scope of the project – this project ballooned into something much bigger and more ambitious than either Jon or I had initially anticipated. We did not think we would end up speaking to nearly 20 people and writing a paper that was over 40 pages long; our initial estimate was to interview four to five individuals and write approximately 10 pages. The enthusiasm, however, that we received as we embarked on our work really spoke to the need for such a resource and how underexplored this topic is. The additional people and work our interviewees referred us to is also a reminder of how much more there is to be uncovered. In addition, doing such qualitative interviews was a new style of working for me, as someone with a mostly quantitative background. Being able to ask open-ended questions and probe more deeply into these issues with our interviewees further reinforced the value of qualitative research approaches and how they can complement quantitative analysis of numbers.

Another takeaway I had from working on this project was the importance of casting a wide net. We spoke to a variety of folks in various fields and sectors and looked for people working at the intersection of data and equity, even if they were not data visualization practitioners specifically. Doing so opened our minds to new issues we had not considered ourselves. They also referred us to more people, organizations, and resources to incorporate into our work.

Another important point was the importance of having institutional or formal support. Jon and I could not have done this work if we had not received dedicated funding to do so. There is no way we could have spent 20 hours interviewing and an additional 40+ hours each writing and editing this guide as a side project. It is important that DEI commitments are backed with concrete resources. On a similar note, compensating people for their time was a theme that emerged from our conversations and was admittedly something we were only partially good at. We unfortunately were unable to compensate our interviewees for the time they spent talking to us, but we were able to compensate our internal reviewers for providing feedback on our draft. This is definitely an area of improvement for the future.

Finally, I had a great experience working with Jon on this guide. We brought different perspectives and styles and I felt that we really balanced each other out. As someone in the earlier stages of my data visualization career, I am particularly grateful to have had the opportunity to work on a project such as this with him. Jon is so incredibly knowledgeable and experienced in the field of data viz and has been an amazing mentor that I have learned a lot from. This project has expanded my horizons when it comes to thinking about how issues of equity and inclusivity intersect with data, and I am honored to have been able to make my own small contribution to this body of knowledge.

Looking forward, Jon and I hope that the Do No Harm Guide will be a living, breathing document that continues to be updated and expanded on as the field of data visualization matures and as society and technology further evolve. From the outset, we were careful to frame this guide as a collection of issues to consider rather than a set of definitive, ironclad rules. We hope that our efforts on the Do No Harm Guide move us one step closer to a world in which data are visualized and communicated in ways that live up to the title of this project.

Alice Feng is a data visualization developer based in the Washington, DC, area. She is passionate about using design to make data and information more accessible to broader audiences and recently has been exploring ways to bring more diversity, equity, and inclusion into the way we visualize data. Alice is one of the coauthors of the Do No Harm Guide, a resource for data visualizers and communicators on how to work with and present data with an equity lens.

1 Schwabish, J. & Feng, A. Do No Harm Guide: Applying Equity Awareness in Data Visualization. https://www.urban.org/research/publication/do-no-harm-guide-applying-equity-awareness-data-visualization (2021).