The Importance of Data Visualization in Combating a Pandemic

An unprecedented volume and variety of data have been produced to analyze and monitor the changing dynamics of the COVID-19 pandemic. Within the backdrop of this data deluge, the use of data visualization, and especially dashboards, served as a core component of the public health response to disseminate and distill key indicators of community spread and to contextualize mitigating actions. Individuals also began to create dashboards of their own on personal websites or public platforms, such as PowerBI or Tableau Public, adding a personal interpretation to the broader public health narrative. At no other point in history has there been such an extensive collection of data, visualizations, and dashboards of a singular virus and the disease it causes. As the pandemic continues into its third year, it is important to reflect on the effectiveness of these dashboard creation strategies and what we can learn from them.

WHY DASHBOARDS?

The importance of data visualizations for informed decision-making has long been examined and demonstrated in use cases as divergent as therapeutic decision aids, infectious disease management, and policymaking.^1,2 These studies of data visualization focus primarily on simplistic ways of communicating individual risk metrics.³ However, the pandemic has underscored the utility of more complex visual displays to engage and inform the public, including embedding visualizations within text narratives (i.e., news articles), infographics, and dashboards. Among these different approaches, dashboards stand apart in their specific goal of displaying data in a way that is glanceable but also creates channels to engage in further exploration or analysis; in contrast, narratives and infographics are intended for deeper exposition, although they can limit opportunities for exploration and alternative analysis.⁴ The glanceable views of dashboards are constructed from multiple visualizations that are combined, ideally, to provide an effective overview of the data. I argue that it is these characteristics of dashboards that made them a key tool that many public health organizations used to communicate and contextualize the changing dynamics of the pandemic.

EXAMINING DASHBOARD CREATION STRATEGIES

The ways that dashboards are constructed is an important factor in their efficacy. Different dashboard configurations can vary from the types of data and visualizations they show to even nuanced choices of individual design elements such as color, text, and degrees of interaction.^5,6 The efficacy of individual visualizations is as important as their combination into a dashboard.⁷ Yet, despite research exploring different visualization and dashboard creation strategies,^5–8 a clear consensus on an effective approach has failed to emerge.

From the perspective of business analytics, where dashboards first gained widespread use, the purpose of dashboards is to relay high-level metrics.⁴ Many public health dashboards did conform to this constraint and emphasized key indicators of the pandemic’s changing dynamics, including case counts, hospitalizations, and deaths. However, data underlying the pandemic and public health responses are also more complex and include changes over time, geography, viral genomics, and networks of transmission.⁹ Forecasting models showing the possible trajectories of the pandemic, including the response to interventions (i.e., vaccines, therapeutics, and nonpharmaceutical interventions), were also influential in shaping public perceptions. These statistical and machine learning models also add uncertainty that influences how information is visualized, interpreted, and acted on.⁸

These complex types of data push the boundaries of dashboard creation, but they are essential for telling a complete and informed story about the pandemic. An example is Nextstrain,¹⁰ which provides a real-time visual snapshot of viral evolution and spread. Nextstrain integrates multiple visualizations, including a phylogenetic tree, a map, and a Muller plot, to convey the extent of viral variation over space and time. Nextstrain had been developed over the better part of a decade, initially to monitor seasonal influenza, but it also was tested and refined in the Ebola and Zika outbreaks. Over time, the creators arrived at a set of visualizations that provided an effective overview. Early into the pandemic, researchers with the Seattle Flu Study shared a Nextstrain visualization with a concerning conclusion: the then unknown novel coronavirus had already been circulating in Washington State for approximately 6 weeks. Within 1 week of this finding, many employers in the Seattle area shifted to remote work, with others across the United States and beyond soon adopting a similar strategy. This scenario underscores the important synergistic relationship between data analysis and visualizations; it demonstrates the capabilities of visualization to contextualize the public health implications of analytic results.

THE POTENTIAL FOR DIS- AND MISINFORMATION

Despite the demonstrated benefits of visualizations and dashboards, there is also evidence that they can misinform, either intentionally or not. Intents to mislead with data and visualization are active efforts of disinformation; examples of disinformation using data visualization have been actively observed throughout the pandemic and shared widely on social media.¹¹ More poorly understood is misinformation that does not result from malice but is unintentional, such as by failing to update dashboard data, accidentally displaying an incorrect metric, or failing to include appropriate contextual information.¹² Poorly crafted visualizations that exaggerate data trends, such as via truncated axes or filtering out or obscuring relevant contextual information, also affect how people read and interpret dashboard information. As the pandemic continues, the presence of “stale dashboards”—those whose data has not been updated in some time—may also serve as concerning sources of misinformation; at first glance, the reader may not understand that the information is out of date and may come away with an incorrect interpretation. The reader’s own data and graphical literacy baselines will also factor into their interpretation of dashboards. Without knowing how these baselines are distributed among the general population, we are not able to assess the efficacy of dashboards to inform or the susceptibility of the public to visual dis- and misinformation. We also lack insight into how dashboards can build or (intentionally or not) erode trust in data and public health organizations.

Unfortunately, the pervasiveness of dashboards that misinform, and their impact on public understanding of the pandemic, has not received much attention. In retrospective analyses of the pandemic, the role that data visualizations and dashboards play in the public’s perception, including their contributions to dis- and misinformation, needs additional scrutiny.

THE CASE FOR A DATA VISUALIZATION RESEARCH AGENDA

Data visualization has long been viewed as an afterthought in data analysis. Yet, the current pandemic makes a compelling case that uses of data visualization and the creation of dashboards are key instruments in the public health toolbox, alongside methodologies from statistics, epidemiology, and qualitative research. Dashboards are also tools that can be misused, both intentionally and not, to mislead rather than to inform. Although research exists that demonstrates the importance and impact of data visualization, there is also a lack of studies that examine visualizations in more complex presentations, such as a dashboard, and that are robustly tested with a broad audience. This knowledge gap means that data visualizations and dashboards are relatively untested and poorly understood even as they are being widely deployed to inform and persuade the public during a pandemic. We should do better. A public health research agenda needs to include data visualization not as an afterthought but as a core component. This agenda should consider the role of visualizations and dashboards not only as a communication tool but also as an intervention that both informs and can modify behavior, from individual choices to larger public health policy. It is critical to understand both the scope and the limitations of such an intervention and discern its effect on how people think critically with data.

Beyond the present pandemic, public health interventions will be deployed to address other pressing challenges, from the opioid epidemic to the climate crisis. Then, as now, data visualization will serve as an important tool to inform and persuade. A research-informed and evidence-based approach needs to guide our data visualization strategies and evaluate their impact on the public.