Your questions, answered
“Can anyone explain why The Washington Post's chart of ‘new deaths reported per day’ seems to show cycles, peaks and valleys in a frequency of approximately 7 days?” — Michael in Washington
It's because of the way state health departments report their daily covid-19 numbers, which has been problematic since the beginning of the crisis. Hospitals, labs and government agencies generally have fewer staff working over the weekends, which leads to artificially low statistics early in the week, and higher figures later.
“Even within that, there is a difference between the rhythm of the rises and falls between cases and deaths,” said Jacqueline Dupree, a Washington Post editor who oversees our covid-19 tracking project. “Death reports are lowest on Sundays and Mondays and generally highest on Tuesdays and Wednesdays. New case reports are lowest on Sundays, Mondays and Tuesdays (because not as many tests are run on weekends), rising to their highest generally on Thursdays and Fridays. And if there's a three-day weekend, then everything shifts another day.”
Here's a picture of what she's talking about:
Some politicians appear to be taking advantage of the daily ups and downs. “Today Texas had the fewest covid-19 fatalities since March 30th,” Texas Gov. Greg Abbott (R) tweeted Monday. “We've also had the fewest Texans testing positive for covid-19 in the past six weeks.”
Strictly speaking, that was almost true. Texas reported six deaths and 593 new cases Monday. But Monday counts are always low, and both figures roughly tripled the very next day as they followed the weekly cycle. In fact, Texas reported one of its highest daily totals on record Tuesday: 1,688 new cases.
That's why our tracker now shows a seven-day moving average for each state and the U.S. as a whole, which smoothes out the daily fluctuations. The average makes it clear that Texas's infection rate has risen sharply in the past several days, even though Monday's post-weekend count was particularly low.
Dupree mentioned that some other news organizations track deaths and positive tests on the day they occurred, rather than the day they were reported.
“Each [method] has its pros and cons,” she said. “For instance, there are complaints that the states that choose to graph their new cases by the test dates rather than the report dates are obscuring the trends, because the most recent days on those graphs will always be lower given that it can take a few days for tests to be completed and then reported to the state. We feel that using the date the state reports cases and deaths gives a clearer picture sooner of whether overall new cases and new deaths are rising or falling.”
The trade-off is the jagged daily totals Michael asked about.