MONDAY, JANUARY 25, 2021
And to what extent do we care?: Have we Americans possibly gotten lucky? Have we managed to avoid a predicted surge of Covid infections and deaths?
Yesterday, atop the front page in print editions, a news report in the Washington Post examined that question. Fenit Nirappil started his front-page report as shown:
NIRAPPIL (1/24/21): The United States appears to have avoided the worst-case coronavirus scenarios that officials feared would overwhelm hospitals in the aftermath of Christmas and New Year’s gatherings. But experts caution that the threat from the virus has not diminished and could intensify with the emergence of new variants.
Even as hospitalizations begin to stabilize, they do so from record heights. The country’s hospitals averaged more than 130,000 covid-19 patients a day over seven days this month, far exceeding summer and spring surges. The death toll from cases contracted before and after the holidays will stretch into February. Authorities reported nearly 4,500 deaths Wednesday, a new single-day record.
According to Nirappil, it seems we may have dodged a bullet, though things could still get worse.
A person could say that Nirappil was possibly hedging his bets. For ourselves, we were most struck by the last sentence we've highlighted, in which Nirappil told readers this:
"Authorities reported nearly 4,500 deaths Wednesday, a new single-day record."
Nirappil's statement can be defended as technically accurate. It could also be seen an example of cherry-picking of a familiar kind.
It's true! According to the Washington Post's data set, 4,440 deaths from Covid-19 were reported on Wednesday, January 20. That's almost "nearly 4,500 deaths," though it's closer to 4,400.
Something else in Nirappil's statement is true. Within the Post's data set, that was the highest number of Covid deaths reported for any single day since the pandemic began in March 2020. Inevitably, that may seem like a significant fact.
Then again, other things are true. Wednesday's reported number of deaths had been very high. But as we noted last Friday, these were the numbers of Covid deaths reported on the three days before that:
Reported deaths from Covid-19:
Sunday, January 17: 2,068
Monday, January 18: 1,418
Tuesday, January 19: 2,166
Wednesday's number was very high. By way of comparison, Monday's number had been very low. And just for the record, these numbers—these numbers of "reported" deaths—followed a three-day holiday weekend, when the reporting anomalies which occur every week tend to be larger than usual.
As he started, Nirappil reported only one number from the previous week. Was that an example of solid reporting? Or was it "cherry-picking?"
There's no ultimate way to answer that question, but Nirappil omitted more information than he chose to include. Meanwhile, here's the way he and/or his editor chose to end his lengthy report:
NIRAPPIL: About 3,900 deaths were reported Friday in the United States, suggesting Wednesday’s peak may not be an anomaly. The seven-day average of deaths in the weeks after Thanksgiving hovered around 2,500.
“We are getting, in some way, numb to the numbers,” said John Brownstein, chief innovation officer at Boston Children’s Hospital. “The numbers of hospitalizations, cases and death are still incredibly high. Even if we see a pullback, those numbers are still incredibly scary.”
As he closed his lengthy report, Nirappil made a comparison. Plainly, it was a comparison of the apples-to-oranges kind:
Nirappil compared the number of deaths reported on Friday (about 3,900) to the smaller average number in the weeks after Thanksgiving (about 2,500 deaths per day). In doing so, he showed that he knows about the important statistic known as the 7-day rolling average.
What he didn't say was this:
After Friday's reporting, the 7-day average for the preceding week stood at 3,101 deaths per day. After Saturday's reporting was done, the 7-day average had dropped to 3,084. In other words:
Where the 7-day average once stood at "about 2,500 deaths" per day, it now stands at just over 3,000. That's an apples-to-apples presentation, although it still has its flaws.
There is no perfect way to report the number of Covid deaths. That said, there are plenty of ways to misreport the number of such deaths.
One such way involves the cherry-picking of individual days. To wit:
Even after November's election, Trump spokesperson Kayleigh McEnany would report the (lower) number of deaths reported on some particular weekend day, thereby giving the impression that things just weren't so bad.
By way of contrast, people like Maddow and Tapper would reliably report the (higher) number of deaths reported on some particular midweek day, thereby pushing viewers' perceptions in the other direction.
No one reported the 7- or 14-day averages, the only respectable way to report such deaths. No one bothered explaining the weekly reporting glitch which drives numbers higher on midweek days, much lower over the weekend.
Nirappil picked the highest number, simply ignored the others. Eventually, he reported the 7-day averages post-Thanksgiving but ignored the 7-day average which obtained as his report went to press.
At present, that average is down by roughly ten percent from where it stood just ten days ago. That may be a holiday-based statistical glitch, or it may suggest that some modest improvement is taking place.
At any rate, the current average—3,070 Covid deaths per day—is higher than it was at any point during the year 2020. That's a basic statement of fact, with high and low numbers thrown in.
Atop page one, in paragraph 2, Nirappil cited one number—the highest one-day total ever. He didn't mention the lower numbers which had come in the days before.
He didn't work with 7-day averages until the very end of his report. At that time, he offered an apples-to-oranges comparison instead of saying something like this:
"The seven-day average of deaths in the weeks after Thanksgiving hovered around 2,500. Today, the seven-day average stands at just over 3,000 deaths per day, though it may be leveling off."
Even there, the recent holiday weekend may be introducing noise into the statistics. But that point could be mentioned too, and at least you'd now be comparing apples to apples, with a word of caution thrown in.
We were struck by the way Nirappil (and his editor) shot that largest number right to the top of the pile. Cable stars have done that for months, even as McEnany was cherry-picking the smallest single-day numbers.
Nirappil's choice brought a question to mind. It's an anthropological question, and we plan to regard it that way.
Our questions goes like this:
To what extent are we humans wired to care about such entities as accuracy and truth? Enjoyably, we plan to discuss that question this week with respect to some well-known movies.
"What is truth?" Pontius Pilate once asked. Our own question will be somewhat different.
To what extent are we humans wired to care about such entities as accuracy and truth? This week, we'll examine the question with respect to the way major critics have reviewed certain films. Next week, we expect to move on to prevailing questions of gender and race.
Are we wired to care about truth? Or is it Storyline all the way down?
In our view, the answer to this is in no way clear, not even here in Our Town!
Tomorrow: Roger Ebert, Mel Gibson, three stars