MONDAY, FEBRUARY 8, 2021
Statistics are very hard: What follows doesn't matter at all, except as anthropology.
As anthropology, it's painfully instructive. At issue is the following question:
Nationwide, how many people have been dying of Covid-19 on a daily basis?
As you know, such numbers tend to be presented in the form of a "seven-day rolling average." According to the Washington Post's data, these were the averages which obtained at the end of the past five days:
Deaths from Covid-19 per day, nationwide
Seven-day rolling average
Wednesday, February 3: 3,070
Thursday, February 4: 2,906
Friday, February 5: 2,910
Saturday, February 6: 4,112
Sunday, February 7: 2,816
According to the Washington Post, an average of 2,910 people per day were dying from Covid as of last Friday.
By Saturday, the average had jumped to 4,112 per day! One day later, the average had gone back down to 2,816.
Remember—those numbers don't represent the number of deaths occurring on, or recorded on, each particular day. In each case, those are supposed to be the average number of deaths per day over the prior seven days.
Having said that, you're right! Those numbers don't seem to make sense.
More specifically, that sudden one-day jump, followed by a return to that much lower number, doesn't seem to make much sense. Also, for what it's worth, the numbers at the New York Times indicate no such one-day jump in the average number of deaths.
For what it's worth, we can't tell you why that Post is showing that anomalous one-day jump. Intuitively, the sudden increase might seem to be related to an anomalous, very high number of deaths reported by both newspapers for Thursday, February 4—but even there, we can't see why the Post's rolling average didn't jump until Saturday, then suddenly came back down.
What explains the anomalous, very high number of deaths reported for last Thursday? The Post doesn't explain, but the Times does—that very high, one-day number reflects the fact that, on that day, "Indiana announced about 1,500 deaths from previous months after reconciling records."
Stating the obvious, a whole bunch of deaths "from previous months" probably shouldn't be considered in computing current seven-day averages. Based on appearances, the New York Times didn't include those 1,500 deaths in its computation of current seven-day averages.
Having said that, alas! Even those deaths from previous months can't explain why the Post's seven-day average took an unexplained one-day jump two days later, then came back down again.
The Post's graphic doesn't seem to make sense; neither do the Post's data. Two days later, no one at the Post seems to have noticed how weird its graphic looks, or how odd its data seem.
This is where the anthropology starts:
Unless you work in the building or moon-shot trades, statistics turn out to be very hard for members of our species! Our major newspapers have made this point abundantly clear in a wide array of ways over the course of the past several decades while covering, or failing to cover, a wide array of topics.
National test scores were very hard, even when they were going way up. For that reason, they weren't reported, discussed or explained. Major pundits just kept saying that nothing was working.
Data about lead exposure were very hard, even when we were pretending to discuss the water problem in Flint. It was more thrilling to drop the P-bomb, scaring Flint's parents and kids.
It's hard to adjust economic data for inflation. It's hard to adjust for population in reporting different nations' death rates from Covid-19.
Back in the mid-1990s, it wasn't hard, it was impossible, to explain whether Newt Gingrich's Medicare proposal would require "cuts" to the Medicare program, or would simply "slow the rate at which the program would grow." On TV, our major pundits were puzzled for a full year.
In a somewhat similar vein, it would be both boring and hard to discuss the lunacy of our nation's per-person health care spending. For that reason, and perhaps for others, the major newspapers in Our Town never report such data at all.
("Too many numbers," as Chris Cuomo often says, out loud.)
Anthropologically speaking, nothing about such concerns isn't extremely hard. Anthropologically speaking, only recitation of Script and Storyline can be said to be easy.
"Al Gore said he invented the Internet?" Every journalist recited that script. Memorization is very easy, but also reassuring and fun!
Everybody says the same thing! No one has to be nervous! Our reporting is based on advice from major credentialed top experts.