Posted in Behind the News

A note about virus numbers

, by John Daniszewski

Numbers have been key to reporting on the current coronavirus health crisis. Data on the geographic spread and the acceleration or deacceleration of the contagion have been the foundation for understanding the battle. National and global metrics on cases, deaths and recovery are the battlefield statistics we need to tell this story.

Keeping up with the data around the COVID-19 outbreak can be a challenge. With a multitude of dashboards and data sources out there, the AP has decided to use the data collected by the Johns Hopkins University Center for Systems Science and Engineering as our prime source for outbreak caseloads and death counts for the United States and globally.

Temperatures are checked before people enter a local city hall during a community quarantine aimed to prevent the spread of the new coronavirus in Manila, Philippines, March 24, 2020. (AP Photo/Aaron Favila)

Among the points that played into the decision: The Hopkins data is now available at the county level in the U.S., which will allow us to see local trends. Hopkins has been collecting this data for the longest time, starting months versus weeks ago — gathering data from a variety of sources and vetting and double-checking it before publishing. Its data project is led by actual epidemiologists and doctors familiar with disease spread.

As with everything, there are exceptions: If a mayor or governor announces a number that is different from what is in the dashboard, then it’s fine to use that number and attribute to the source. But we should still use Hopkins for global figures, and not be afraid to note when their numbers might not match up against the latest announcement.

Unless the story is about a milestone, it’s better to speak about the numbers in generalities, or to round off the figures in stories, scripts and captions. Chances are any exact figure will be outdated as soon as you push the button, so “at least” and “more than” are the best tools to avoid looking dated.