x
Breaking News
More () »

VERIFY: How a viral chart of COVID-19 cases by state partisanship came together

A viral video on Twitter visualizes the change in which states have had the most COVID-19 cases over time since the start of the pandemic.
Credit: AP
This electron microscope image made available and color-enhanced by the National Institute of Allergy and Infectious Diseases Integrated Research Facility in Fort Detrick, Md., shows Novel Coronavirus SARS-CoV-2 virus particles, orange, isolated from a patient.

A video showcasing a graph tracking COVID-19 cases by state and partisanship since June has gone viral multiple times this past week. 

Multiple versions of it have gained thousands of likes and shares as some people remark on the visualization and others question the accuracy or methodology. One Twitter post featuring the chart has racked up more than 8.5 million views. 

What's shown in that viral tweet is that when the chart begins in June, it shows "Blue States" toward the top for those with the most coronavirus cases, but as time goes on the illustration shows more "Red States" filling the top slots. 

So where did this graph come from? Is the data accurate? Why is it presented this way?

WHAT WE FOUND

The graph comes from Dan Goodspeed, a web developer who runs his own website and blog. A few questions can be answered just by reading the site where he's posted the chart. 

First, this is one of several COVID-19 charts he has on his website that presented in the "bar chart races" style. That's when a visualization shows how bar charts compare as time goes on. He's also made charts showing weekly new COVID-19 cases and deaths by state and for total cases and deaths by state. Those four charts all start on March 1 and measure from there. 

The chart that's been going viral on social media starts in June and is the only one that shows the partisanship of each individual state.

The COVID-19 data he uses comes from data compiled by the New York Times that's provided in real-time. The New York Times explained “the data is the product of dozens of journalists working across several time zones to monitor news conferences, analyze data releases and seek clarification from public officials on how they categorize cases.” The outlet added that the data is based on reports from local and state health agencies. While the New York Times data isn’t the exact same as the Centers for Disease Control and Prevention data, it's fairly similar. 

Goodspeed explained on his website that he uses the Cook Partisan Voting Index for determining the political lean of each state. The Cook PVI estimates how much more Republican or Democrat a state is than the national average,  based on recent voting patterns.

This measure does not consider the party-affiliation of the state’s lawmakers. A state that leans Republican on the PVI can have a Democratic governor and/or a Democratic majority in the state legislature, and a state that leans Democratic can have a Republican governor and/or a Republican majority in the state legislature. 

Goodspeed explained beneath the chart that's gone viral that “June 1 was chosen as it was around that time that countries worldwide that had been sucker-punched months before had the opportunity to apply proven preventative measures.” Essentially, he is arguing that by that point scientists worldwide had determined the best ways to slow COVID-19 and so differences in total cases by that point are notable.

He said he plans to make a similar chart showing partisanship and deaths next. He said he made a chart for cases first because it’s more steady.

RELATED: VERIFY: How a Trump campaign ad misrepresents a quote by Anthony Fauci

RELATED: VERIFY: Why Abraham Lincoln chose not to appoint a Supreme Court justice during 1864 election

That steadiness is something he corrected for even within the charts tallying case totals. Goodspeed’s website explains the bar charts are normalized, meaning that if 1000 new cases were reported in a day after there were about 50 cases reported the two days prior, those 1000 cases would be spread about the three day period in a method that treats the sudden case surge as a correction for missing data the days before.

Have something you'd like VERIFIED? Click here to submit your story. 

Before You Leave, Check This Out