That sounded good. But it ignored a few key qualifiers. One was that about half of those counties didn’t have any confirmed cases in the first place, somewhat reducing the importance of their not adding any additional cases of the contagious virus. The other was that Trump’s description of those counties as “30 percent” of the country misrepresented how broad the effects were. While they constituted about 30 percent of the number of counties, they only represented 6 percent of the population. Cases were added in counties where 94 percent of the population lived.
A week later, Trump again rolled out a number meant to indicate that the country was defeating the virus.
“As we continue our battle against the virus, the data and facts on the ground suggest that we’re making great progress,” Trump said. “In 23 states, new cases have declined in the [week]. Forty percent of American counties have also seen a rapid decline in new cases. Forty-six states reported drop in patients showing coronavirus-like symptoms. That’s a big number.”
The last measure is hard to evaluate, since the Centers for Disease Control and Prevention releases weekly data on the prevalence of coronavirus-like symptoms by region. Let’s focus instead on the first number, those declines by state over the past week.
Nationally, the peak that Trump identified a week ago appears not to exist. There’s some volatility in new cases by day, but a three-day average of new confirmed cases shows that the country has been pretty consistently adding just under 30,000 new cases. That this correlates with a consistently flat number of daily tests is probably not a coincidence, as the Atlantic has reported.
Technically, the three-day average on April 16 was slightly higher than the average on April 23, the most recent day for which Johns Hopkins University has full data. Over that week, the national three-day average dropped by about 3.5 percent, from about 29,000 to about 28,000.
Trump’s 23-state figure, though, is hard to ascertain. Considering the same metric — change in the three-day average, which accommodates somewhat for daily fluctuations — only 13 states saw a drop from April 16 to April 23. Happily, that includes a number of the worst-hit states such as New York and Michigan.
(States are listed from smallest to largest drop.)
All of the other states, though, saw an increase. That includes 21 states where at least 200 cases are being added a day at this point, helping to drive the national increase in confirmed cases.
Most concerning are the states where the three-day average of new cases has at least doubled over the last week. It includes several states with low population densities, the sorts of places that Trump has waved off as unlikely to be significantly affected by the virus given their spaciousness.
Granted, it’s easier to double the number of new cases when you start at 10 cases. Combined, though, these six states are adding more than 1,300 cases each day on average. Only Wyoming is adding fewer than 100.
This question of when and if states are hitting peaks is best evaluated by looking at the long-term trend. We can visualize the three-day average in each state as a function of the maximum value each state has seen. This interactive shows how new case totals have evolved since March 20. (It includes data on deaths, tests completed and the percentage of tests returning as positive. The test data are from the COVID Tracking Project.)
Show data for
While few states are at their peaks — good news — many have seen drops in new cases turn around and head back up. We’ve expected to see neat, bell-shaped curves on these metrics, but the picture is instead much less clear.
It is certainly the case that a consistently daily increase in new coronavirus cases is better than an exponentially growing number of new cases each day. But, again, the White House’s numbers on how successful we’ve been is questionable, even once one accounts for the cherry-picking.