Reflecting on the first eleven months of the global pandemic
The COVID-19 global pandemic has been unprecedented in its scale, disparate physical reactions, global-reach, and economic repercussions, as well as its generation of mass quantities of real-time, readily accessible data. Due to faster Wi-Fi speeds, cloud storage, and software advances, COVID-19 analysis and visualizations (by both experts and novices) have proliferated as rapidly as the virus itself. The ubiquitous graphics, although creatively inspiring, were at times exhausting.
Opting to work with COVID-19 data for my final project, I was conscientious of this visualization fatigue. I reflected on the frustrations I experienced when viewing other graphics, and sought ways to create visualizations that would answer my outstanding questions.
My first decision was to look at individual state data in conjunction with global country data. Although somewhat unorthodox (and possibly, unjustly prioritizing the United States), one set of numbers cannot accurately capture the pandemic’s spread across all 50 states. The varied state data reflects that the country has lacked a unified, national response to the virus – information I sought to capture in my visualizations.
The objective of my research was to reflect on the past year’s data, specifically the accumulations and fluctuations of cases and deaths. I sought out patterns to the virus’s escalation, and logic to the path of its global spread.
Visually, I knew my data story would require maps and graphs, but I also aimed to create something new. I wanted to generate at least one graphic that stood out from the standard forms of visualization that I regularly encountered.
DATA & PROCESS
To begin my project, I merged two open-access datasets. I accessed daily, global data from Our World in Data (OWID) and United States daily data from The New York Times API. My initial analysis was completed in Excel where I normalized and unified the files. I built out the NYT’s data to mirror the robustness of the global data. By incorporating state populations, I was able to view both cases and deaths as per of 100,000 people. To allow for mapping variations, I added a column to associate each state with its region (as designated by the US Census Bureau). Wanting to capture monthly averages, I focused on data from January 1st through November 30th.
After this initial analysis, I brought my dataset into Tableau, where I generated dozens of worksheets to explore and visualize. My immediate idea was to create maps (for both cases and deaths), displaying monthly averages. Although informative, I found these graphics to be ordinary and underwhelming. This story could be better told with a motion graphic that rapidly played each day of the pandemic while allowing the user to pause and access detailed information. (Consideration for a future project when I acquire more coding knowledge.)
Contemplating COVID-19 in terms of days and months, I envisioned a calendar representation of the data. Although not a standard form of visualization, a calendar (like a map) is readily recognizable information spanning over time. I was additionally inspired by the periodic table, and its successful use of information hierarchy in a confined space. Initially attempting my calendar design in Tableau, I quickly moved to Adobe InDesign, where I could easily establish a complex grid to keep information aligned and ordered. This visualization required the most iterations regarding what to include, what to make most dominant, and where and how to utilize color encoding. While working in InDesign, I also created a bar chart depicting COVID-19’s ranking in the leading causes of deaths.
I returned to Tableau for the remainder of my visualizations. Using area plots, I looked at monthly percentage changes, in the world, the US and New York. Regretfully, my world data referenced total global population, rather than the combined populations of the countries that have provided COVID-19 data, resulting in significantly lower averages.
I created a map to display COVID-19’s cumulative impact on the four US regions: Northeast, Midwest, South, and West. Pie charts presented the monthly variations in the regions. (A happy accidental discovery, expanding the size of the pie chart in Tableau results in these more visually appealing, box graphics – which I have nicknamed brownie charts.)
Wanting to capture the rapid acceleration of increased cases, I used a simple bar chart to visualize all days of the pandemic, annotating the milestones: 1 million cases, 10 million cases, and so on. Additional annotations capture the increased pace of spread.
Lastly, I combined all visuals into a Tableau story. Incorporating JPEGs of my InDesign graphics was not ideal. Some image quality is lost, and more importantly there is no interactivity with the information. In an attempt to not frustrate the viewer, my subheads provide directions as to what visualizations offer further exploration.
I conducted UX research with two subjects throughout my design process. I purposely chose one subject with a design background and one with an analytical background. The designer provided positive feedback. She found the case milestone bar chart to be a new and especially informative graphic. She was also excited by the information-dense calendar. The analyst, although appreciative of the graphics, had data-related questions. As I reflected on the data in a variety of ways (cases, deaths, as per 100,000 people, monthly totals, grand totals, etc), all labels needed to be accurate and clear in order to make the visuals understandable. He also questioned the validity of certain numbers, which was extremely helpful, as my ability to answer (or not) revealed the extent to which I understood and could support the data. In addition to finding the milestone bar graph insightful, he was most interested in the macro to micro area plots of cases and deaths. I had initially annotated only two percentage changes on each graph, but ultimately expanded to show all months in the hopes of telling a more complete story.
Viewing the calendar graphic, I have a greater appreciation of the global health community. I recall first learning of COVID-19 towards the end of January – yet, to my eyes, the January data is far from alarming. (Noted that these numbers are most likely unrealistically low due to inaccurate data provided by China.)
The New York area plot depicted that the state was impressively successful at quelling the virus after April’s apex. Seeing a second, upward peak begin in September is discouraging. This timing coincides with people returning to the city (after summer escapes) for both work and school.
I had not fully appreciated how rapidly the virus spread was increasing, until viewing the milestone bar graph. Ten million new cases in a span of 17 days is nearly impossible to comprehend. And we are now on track to reach the next milestone (70 million cases) in only 16 days. At least the forthcoming vaccines offer some hope.
More than one cohesive novel of data, my analysis and visualizations read as if they were a series of short stories on the pandemic. The world maps displaying monthly cases and deaths could sit well with the milestone bar graph. The analysis on US Regions could easily be expanded and elaborated on for a future project. The macro to micro area plots feel like the start of an interesting idea that has yet to be fully fleshed out.
The information dense calendar would work best as a stand alone piece. Although not stylistically similar, this visual was inspired by the work of Goirgia Lupi. Her graphics present multiple layers of data – some readily apparent, some requiring more effort to reveal. Presenting the calendar as an independent visual would allow for more creative freedom, ideally resulting in a more exciting graphic.
I included the deaths in the perspective bar chart, because I found this data to be especially enlightening. However, I view this visualization as a start to an entirely different COVID-19 story. This future story will require more perspective, time and research. It will tie in data from past pandemics, global health statistics from the years effected by the virus, and ideally some sort of end date due to vaccination success.
Throughout my research, I found myself reflecting on one quote regarding the 1918 pandemic. Although I cannot recall the speaker / author, the general sentiment was “For all I have read about the influenza pandemic, its true magnitude is captured in this one fact: in 1918, the average lifespan in the US decreased by 12 years.” COVID-19, a very different virus, will not significantly alter life expectancy. But provided enough time and perspective, maybe there will be one data point powerful enough to capture the immensity of this pandemic. I look forward to continuing my COVID-19 research and seeking it out.