WEDNESDAY, DECEMBER 2, 2020
Let the Post's editors show you: As we've often noted, reporting statistics is very hard, especially for major top journalists.
That's even true with respect to major topic like pandemic infection and death. This very morning, the editorial board of the Washington Post helps us see how hard it can be to fight our way through such statistics.
In a newly published editorial, the board says that Sweden's experiment with "herd immunity" has majorly failed. As far as we know, that assessment seems to be accurate.
As far as we know, Sweden's experiment has been a major failure. But along the way, as they attempt to prove their case, the highly distinguished upper-end editors haplessly tell us this:
WASHINGTON POST EDITORIAL (12/2/20): Sweden is now caught in a wave of pandemic pain—and reversing course. Sweden has 48.9 new confirmed cases per 100,000 population, compared with 21.7 in Denmark, 8.2 in Norway and 7.7 in Finland. Sweden is averaging about 42.6 deaths per day, compared with 6.9 in Denmark, 3 in Norway and 2.1 in Finland. Sweden’s total 6,798 deaths, predominantly among the elderly, dwarf the toll in the other Nordic nations combined.
The editors went to the finest schools. But how much good did it do?
In this passage, the editors seem to be claiming that Sweden's current rates of Covid infection and death are much higher than the infection/death rates of three Nordic neighbors.
As far as we know, that's true. But for starters, just consider the very first claim they made:
"Sweden has 48.9 new confirmed cases per 100,000 population, compared with 21.7 in Denmark, 8.2 in Norway and 7.7 in Finland."
In that statement, the editors have adjusted for population. For each of the countries, they're reporting the number of new cases per 100,000 population—but they fail to state the time span in question.
Sweden is said to be recording 48.9 new cases per 100,000 population. But is Sweden recording that number of new cases on a daily basis? Is that perhaps a weekly average?
The reader has no way to know. In a typical manifestation, the editors fail to say.
Now, consider the editors' second statement:
"Sweden is averaging about 42.6 deaths per day, compared with 6.9 in Denmark, 3 in Norway and 2.1 in Finland."
In this case, the editors have defined the relevant time span; Sweden is averaging roughly 42 deaths per day. But is that Sweden's average number of deaths per day unadjusted for population? Or is that Sweden's average number of daily death per 100,000 population?
The editors failed to say—and by now, we were puzzled. Here's why:
A forgiving person might want to assume that both sets of statistics represent these nations' average daily occurrences per 100,000 population.
A forgiving person might want to make that assumption. But is it possible that Sweden is averaging 49 new cases per day, but as many as 42 deaths?
On its face, that didn't seem to make sense. For that reason, we decided to click the one link the editors provide. We were taken to this borderline bewildering web site maintained by the Financial Times.
Even at that major Covid web site, the FT does a strikingly poor job explaining what its various charts and graphs are recording. Such confusions have been remarkably common as elite journalistic cadres have attempted to keep us abreast of the most significant Covid-19 statistics.
Making a long story short, we journeyed to some other sites which are easier to interpret. And sure enough:
According to this useful site, Sweden is currently averaging roughly 42 Covid deaths per day; Denmark is averaging 7. But those are average numbers of daily deaths nationwide, unadjusted for population.
That said, Sweden's population is almost twice the size of each of the other three Nordic nations. And as we've noted a million times, it doesn't make sense to draw comparisons of this type between countries without adjusting for population.
As a general matter, it doesn't make sense to do that. But our major journalists, from Rachel on down, routinely do this. (Needless to say, it's generally done when it makes a preferred storyline better.)
Again, we're not suggesting that the board's overall assessment about Sweden is wrong. As best we can tell, Sweden's attempt to ride the herd immunity hobby horse is an experiment which has failed.
We're speaking here about something different. Consider our point this way:
Fred Hiatt, the Post's editorial board editor, is a Harvard graduate (class of 1974). And not only that! His father, Howard Hiatt, was a medical researcher and dean of the Harvard School of Public Health!
Even with a background like that, Hiatt couldn't induce the board to offer a coherent presentation about the way these Nordic nations currently fare in this major public health crisis. Hiatt is a thoroughly sensible person, but in an intellectually capable world, that passage in this morning's Post would be viewed as a comically puzzling mess.
It wouldn't be hard to rewrite that passage so that it made perfect sense. You'd have to make some basic adjustments:
It made no sense to present any figures in that passage without adjusting for population. Most strikingly, it made no sense to adjust the figures for Covid cases but not for Covid deaths.
For various reasons, it would have made better sense to present the daily numbers of cases and deaths per million population. But all in all, that passage from the board is an incompetent mess.
That said, can we talk? This sort of thing is amazingly common when Covid data are reported by the upper-end press. This brings us back to the conceptual mess Governor Kristi Noem loosed on the world back on November 18.
Noem is governor of South Dakota. Her father was a farmer/rancher. He wasn't dean of anybody's school of public health.
With that in mind, to what extent might Noem have thought that her presentation made sense? We can't hope to answer that question, but just to refresh you, her hopelessly jumbled, grossly misleading presentation went exactly like this:
NOEM (11/18/20): Across the country and around the globe, cases [of Covid-19] are increasing. Over the past week, cases are on the rise in 48 states.
Some have said that my refusal to mandate masks is a reason why our cases are rising here in the state of South Dakota, and that is not true.
Others have said that my refusal to advance harsh restrictions like lockdowns is another one of the reasons why our cases are rising, and that is also not true.
There are 41 states that have some kind of a mask mandate. Cases are on the rise in 39 of those 41 states.
Now, some in the media have said that South Dakota is the worst in the world right now, and that is absolutely false. I'd encourage you to look at the Johns Hopkins Coronavirus Resource Center. There you'll see that there are other states with far higher new confirmed cases per one hundred—one thousand people, compared to South Dakota.
That presentation was a mess. How did the press corps react?
Tomorrow: What in the world had she said?
For figure filberts only: For the record, the FT site records these current daily death rates for the four Nordic nations. These are average numbers of Covid deaths per day per 100,000 population:
Norway: 0.06Finland: 0.06
Even after adjusting for population, Sweden is doing substantially worse. But dag, those numbers look small!