Visualization Guide

Explore graphics made with the COVID Tracking Project dataset along with tips to help you present the data in the clearest and most accurate way possible.

Complete, up-to-date testing and outcomes data is essential to a successful public health response to the US COVID-19 outbreak, but that data is not being published by the Centers for Disease Control and Prevention (CDC) or any other governmental entity.

The COVID Tracking Project exists to help fill that gap. We collect the best available testing and patient-outcomes data from every US state and territory and the District of Columbia, with full daily updates to the entire dataset.

Please note: Our data will always be an undercount. We can only track tests that states report, and not all states report all tests. More significantly, per-capita testing levels in the US remain low, which means that an unknown but probably very large number of people are sick, but aren’t being tested. But this is the data we can collect, and it provides the most detailed information available about the shape and relative severity of outbreaks in US states and territories.

According to our data, US metropolitan areas have been hit the hardest. New York leads the rest of the country in the number of positive cases by more than 120,000, followed by New Jersey, Massachusetts, and Louisiana.

Because COVID-19 testing and reporting are inconsistent among states, it’s easy to misinterpret the data. That makes it especially important to create clear and accurate visualizations. Otherwise even simple and minimalistic graphics can be misleading. If you plan to display data from the COVID Tracking Project yourself, please closely follow these design and visualization guidelines.

Consider normalizing the data.

If you’re creating a choropleth map (where each state is shaded in proportion to a statistical variable), make sure you encode a population-controlled rate, such as “positive tests per one million people.” If you want to show absolute numbers, such as the number of new positive cases per day, use a symbol map.

Choose colors carefully.

Readers are likely experiencing some latent anxiety, so do your best to neither make light of the situation nor be alarmist about it. One application of this is in your color choice: You don’t want your map’s color scheme or design to minimize the situation by being overly playful or lighthearted. You also don’t want to select colors that suggest the worst possible outcome.

Include the denominator.

Testing is one of the most important tools in controlling an outbreak. When universal testing is implemented, people who are infected with the virus can be isolated from folks who test negative. This functions as a targeted social distancing technique and can help slow the outbreak.

Charting the number of positive tests alone is often problematic. Simple case counts show where people are being tested, not necessarily where people are sick. To illustrate the point, a state that reports three cases of COVID-19 after testing 2,000 people is probably in a different stage of its outbreak than a state that reports three cases but has only tested 20 people. But if all you have is a case count, those states look exactly the same. That is why we need to include the total number of tests as a denominator.

Be mindful when comparing states.

By comparing positive tests to total tests in each state and territory, we can get a sense of how widespread a state’s testing regime might be (while keeping in mind that population densities vary widely across the country).

Don’t ignore data uncertainty.

Though this is a national crisis, each US state or territory reports its data differently. We track numbers provided by each state, but the quality and frequency of reports vary widely. Transparency is crucial. Be honest about what is in the data that you are charting and what isn’t. Footnotes and annotations can help you make these disclaimers. We publish details about inconsistencies in data reporting for every state and territory.

Use absolute numbers for death counts.

An organized, collective, and timely response from the government and other authorities is a key factor in saving lives. One metric you can use to measure the response’s effectiveness is the number of deaths attributed to the virus per day and/or per state or territory.

We recommend using total numbers for plotting deaths to compare one US state or territory against another. In this case, adjusting per capita adds a layer of abstraction to the graphic. This reduces the data’s power and the reader’s comprehension. It’s easier to picture 200 fatalities than 0.0001 fatalities per capita, as John Burn-Murdoch, a data journalist at the Financial Times, pointed out.

Remember that even death counts are uncertain.

Data that tracks COVID-19 death counts is still a gray area. Though some experts prefer to measure the pandemic’s severity using the number of deaths instead of total cases, several factors could bias COVID-19 mortality data.

There are some concerns that official death statistics may overcount COVID-19 fatalities by assuming any patient who tests positive for the virus was killed by it. But Marc Lipsitch, a Harvard University epidemiologist, told FactCheck.org that “the number of such cases will be small.” Undercounting is a bigger problem, he says. “A greater issue is errors in the other direction.”

If people die from COVID-19 before they are tested, their death might not be included in the official tally. For example, a WNYC/Gothamist investigation found that as of Tuesday, April 7, around 200 New Yorkers had died at home every day without access to testing and medical treatment. That’s 10 times higher than NYC’s typical at-home death rate. These deaths are likely caused by COVID-19. If this effect is widespread in the United States, that means official statistics undercount the disease’s fatality rate.

To get better grounding when interpreting death rates, consider comparing COVID-19 mortality rates for a given location since the beginning of the outbreak to fatalities during the same period in previous years.

Be clear and honest.

While news is moving faster than ever to keep up with the pace of the pandemic spread, designers and visualization experts’ goal is to present COVID-19 data in a clear and honest way. Provide context, and consider the tips above to avoid common pitfalls in data reporting as we seek to inform people during this time of crisis.