Applying Racial Equity Awareness in Data Visualization

Abstract

A data visualization style guide does for graphs what the Chicago Manual of Style does for English grammar: it defines the components of a graph and their proper, consistent use. At the Urban Institute, a nonprofit research institution based in Washington, DC, our data visualization style guide defines these styles for our research and communications staff. In the process of revising and expanding the style guide, we are taking a more diverse, equitable, and inclusive (DEI) perspective to our research, data, and visualizations. To date, our approach has been to create a set of recommendations and issues to consider, rather than a set of rules that researchers must follow. In this paper, we discuss eight techniques that data visualization producers should consider when creating visuals with this DEI approach. As with our existing style guide, we consider this a first step in our thought process of creating graphs and producing content through a DEI lens.

Data science training for official statistics: A new scientific paradigm of information and knowledge development in national statistical systems

“…findability, accessibility, interoperability and reusability [26], concerning fairness, privacy, security, inclusiveness, transparency and accountability. Some frameworks address specifically ethnicity problems such as the Diverse, Equitable, and Inclusive framework (DEI) [27] or the Privacy, Security, Accountability, Reliability, Safety, Fairness, Inclusiveness, Transparency (PARFIT) of Microsoft's responsible AI principles and guidelines [28]. These are important in producing official statistics to serve the public needs not only by data and information but also by empowering social responsibility, social awareness, and sustainability.…”

Designing equitable and inclusive visualizations: An underexplored facet of best practices for research and publishing

“…A second dimension of inclusivity concerns representation equity within visualizations themselves. Visualizations can perpetuate systemic social inequities (D'Ignazio & Bhargava, 2020; Dörk et al, 2013; Kennedy et al, 2016; Schwabish & Feng, 2020). Two major mechanisms by which visualizations can perpetuate systemic inequities are (1) reinforcing the position, status, and power of dominant social groups and (2) obscuring realities or lived experiences that should have been captured by the visualization.…”

“…An example of the first mechanism is a visualization that depicts multiple groups and places White students or White men as the first group, implicitly presenting them as the default category to which all others must be compared. A simple redress is alphabetically listing groups (Schwabish & Feng, 2020) to avoid equating the graphical position of dominant groups with their structural position of power and privilege in engineering. An example of the second mechanism is a bar plot depicting only posttest learning scores by demographic group rather than depicting pre‐to‐post change scores.…”

“…By always showing pre‐to‐post changes (change scores), the graphic preserves the meaningful learning all groups exhibited. Schwabish and Feng (2020) also identify a variety of other ways in which visualizations can challenge or reinforce systemic inequities, including using language and colors with racial equity awareness, considering what groups are missing, questioning default visualization approaches, and exhibiting empathy for visualized populations.…”

Visualization for Villainy

“…We believe this is a critical juncture to carry out this sinister work. Recent efforts to use visualization for social good [1, 3, 4] and the emerging thread of research focused on the ethical issues facing data visualization practitioners [21,29,67] suggest a growing interest in using visualization for the benefit of marginalized people [26,61], disabled people [45,50,53,80], and people in general. We therefore hold that there is an urgent need to intervene now and provide countermeasures to these do-gooder efforts.…”