Unintended Consequences

As noted by the authors of Freakonomics, unintended consequences are prevalent—and data visualizations are no exception. Even the most carefully constructed and accurate data presentations can lead to misinterpretations and poor decision-making.

For instance, early on, certain graphics highlighted the considerably higher mortality risk of COVID-19 among the elderly compared to younger populations.

Conversely, another graph indicated that more individuals aged 20-44 were being hospitalized than those aged 75-85, based on early U.S. data.

These two graphs convey different conclusions. Both are accurate, yet they emphasize distinct aspects of the disease's impact. While many readers might reconcile the two interpretations, casual viewers could easily draw drastically different conclusions regarding the risks faced by different age groups.

This oversimplified understanding early in the pandemic led many to believe that younger individuals were not at significant risk from the virus.

Media coverage often showcased younger people disregarding lockdowns to attend parties and visit beaches. Was there a connection? It’s difficult to say for sure, but the influence of such graphs is plausible.

This example underscores how a simple graph can communicate varying messages using the same data. Let’s consider another instance that illustrates the power of language in discussing fatalities, comparing current figures with projections.

Total Fatalities

Many of us can recall discussions from early 2020, where the total number of COVID-19 deaths was often framed as still being in the tens, hundreds, or even thousands—merely a fraction compared to seasonal flu or daily road fatalities.

This perspective fails to grasp the alarm's essence. The concern surrounding SARS-CoV-2 was rooted in its potential for widespread damage, not just the immediate toll.

However, headlines like the following may not have helped clarify the situation:

This article lacks context—why the virus poses such a significant threat, its high contagion rate, and the dire projections if it spreads unchecked. While avid news followers might be aware, not everyone has the time or desire to read multiple articles.

In contrast, another article published just days earlier conveyed a vastly different narrative:

This article warned that COVID-19 could result in tens of millions of deaths! Even with the same underlying facts, presenting a projection alters the tone dramatically.

Now, let’s examine how the same dataset can be visualized in various ways, leading to different interpretations.

What to Plot, and How?

Regardless of the risk of misinterpretation, accurately presenting data related to a rapidly evolving, infectious disease is challenging.

Consider these two graphs from Our World in Data. The first depicts total deaths in the countries most affected.

In this visualization, the steep curve of the U.S. stands out alarmingly, while China’s initial outbreak figures are notably high.

On the other hand, the subsequent graph normalizes the same data per million people in each country.

This perspective dramatically shifts the narrative. Per capita, China's figures barely register, while the U.S. numbers appear less alarming. Spain and Italy's statistics, however, emerge as particularly concerning.

For someone puzzled by the situation, what message does the latter graph convey? What implications does each graph hold regarding which country has managed the crisis better or worse?

Given that COVID-19 is an infectious disease, should overall population figures serve as the denominator? Is the total number more relevant? Should we evaluate based on the number of epicenters, or perhaps consider population density due to the virus's nature?

Crucially, how will decision-makers at various organizational, governmental, or local levels interpret these visuals and act accordingly?

Without domain expertise, I struggle to assess these data points effectively. Moreover, I wouldn’t know where to begin interpreting these figures or which insights might produce the most favorable outcomes, as Thaler and Sunstein would suggest.

Lastly, consider the challenges of mapping data:

Mapping Data

Visualizing data is equally complex. Take a look at this graphic from The New York Times' COVID-19 tracking page for the U.S.

This map illustrates total cases as of April 13, 2020, at the county level, presenting a daunting view where few areas appear untouched by the outbreak.

Conversely, consider this next image that provides a contrasting (though still troubling) perspective.

This graphic shows that the rate of new cases has drastically slowed on the west coast, while the east coast and southern states continue to see rapid increases.

Given the significant infection numbers in states like California, this finding may surprise many viewers. Once again, the same dataset yields vastly different interpretations.

For example, someone analyzing this map might seek to understand the reasons behind these disparities and how to replicate that success on the east coast, insights that the first graph might not reveal.

In selecting, filtering, manipulating, and presenting data, we make countless choices that significantly influence the output and readers’ perceptions.

The core message here is clear. Those of us involved in data presentation and visualization cannot claim that our interpretations are neutral or free from bias.

While the data remains unchanged, the choices made in selecting, filtering, manipulating, and presenting that data greatly impact the final output and the impressions of the audience.

Data visualization serves a specific purpose. It is not an end in itself but a tool that must align with the underlying intent—effectively conveying the intended message. Authors must consider what that message is and what it aims to achieve.

In the context of a pandemic that has already disrupted millions of lives, displaced individuals, and could cost countless lives, the stakes have never been higher.

As Amanda Makulec aptly stated, let’s #VizResponsibly.

Stay safe, everyone.

If you found this insightful, feel free to share your thoughts or follow me on Twitter for updates. Additionally, check out my piece on various countries' healthcare capacities: