Pages

March 14, 2020

Coronavirus, science literacy, math, and behavioral psychology

A number of people have shared around that New York Times article of a few days ago that claims the COVID19 version of the coronavirus could kill as few as 200,000 Americans. Over the hours today, I've done some mental gestating, and realize that people need to think this through better. (And no, I'm not linking to the story, for several reasons, some of which will become apparent below.)

First, reactions to the story illustrate two things.

One is that Homo sapiens as a species is NOT "rational" by nature. Our australopithecine ancestors weren't and we didn't involve any hugely greater degree of rationality. Nor, contra the hopes and claims of some philosophies like Stoicism, has cultural evolution promoted a much greater degree of average rationality growth. That's simple fact.

Two is that the basic principles of behavioral psychology, per Dan Ariely's excellent book of that name, that show humans are "Predicably Irrational," are quite, quite true.

Specifically, per it and its related behavioral economics, humans do (on average, this is a social science) show predicable — and predictably irrational — behaviors in many scenarios. One of these is loss aversion. That means that we tend to overreact to many of our fears of losses. A related behavior is "loss attention." And, obviously, death is a loss. A huge loss, whether it's our own medical diagnosis or a statistical estimate.

With that, let's dig in.

(Update: Let's have some common sense while you're out shopping and other things, too, folks.)

First, you may say, "I heard about a New York Times story that coronavirus could kill as many as 10 million Americans."

That's the same story.

But, because of your loss aversion and related psychology, you noticed the 10 million. And probably read past the 200,000.

I also noticed the 200,000.

I had already, previously, based on estimates of both the virulence and lethality of COVID19, estimated 250,000 deaths.

That's not nothing, of course.

But PUBLIC health is about social statistics, not individual anecdotes.

Since seasonal flu kills an estimated 50,000 Americans a year, that puts this into some context.

Turns out I am not the only one spreading the "lessen the alarm" stance. Georgetown U. psychology prof Dr. Jelena Kecmanovic has some good stuff. She includes: Accept uncertainty, don't underestimate resilience, and, per what I've said, don't overestimate threats.

Further context?

Alcohol, from DWI fatalities to workplace fatalities to suicides to liver cancer to other cancers and ultimately to cirrhosis, kills almost 100,000 Americans a year. (That's more than all illicit drugs combined, and that number may be low; I suspect it is.)

Cigarettes kill almost 500,000.

So?

If you're overreacting to COVID19 while either smoking a Marlboro or else drinking one too many Buds or shots of Jack Daniel's, I don't want to hear from you. (And, in case you're wondering? Exact numbers are even harder to come by, but stroke, heart attack and other similar deaths, the numbers of them preventable from better diet and exercise, slot between alcohol and tobacco. So, if all the exercise you do is walking out the door to your car, I don't want to hear from you, either.)

Let's continue.

There's one other thing many Americans don't know about the experts estimating death rates of between 200,000 and 10 million. They probably think that means an average of 5.1 million. And they'd be wrong.

But, if they think of a simple brainteaser that stumps many of them, they'd maybe think they're wrong without knowing what the actual median might be, which I'll tell you in a minute.

That brainteaser is a simple one.

If bacteria in a petri dish double every minute, and the dish is filled in one hour, when is it half filled?

Most people divide one hour in half and say "30 minutes."

The correct answer, of course, is "59 minutes." It's half full in 59 minutes and in one more minute, with another doubling, it's full.

Many growth, or decline, and change issues in the natural sciences work by this "geometric progression" that involves multiplying growth, rather than an "arithmetic progression" that involves additive growth.

This is how half-lives of radioactive elements work.

It's also an issue in social sciences, as it's how compound interest works.

The geometric mean between 200,000 and 10 million is 1.4 million.

NOT 5 million.

That is still a lot of people. But it's a lot less than 5 million.

Personally, I think odds are less than 10 percent the US even hits 1 million, for a variety of reasons. That's still a lot of people. But it's a lot fewer than many people will expect.

But, I'm promoting your loss aversion.

So, let me give you MY geometric mean estimate. I think there's a 50 percent chance COVID19 deaths in the US stay below 400,000.

==

There are other factors at work.

Globally, elected politicians, once they see a critical mass on coronavirus worries in elective democracies, are going to tilt toward overreaction.

In the US, there's also an ignorance of population densities and population demographics.

If one takes the "Acela Corridor" and extends it 100 miles inland, that part of the US is as densely populated as the Benelux countries.

Elsewhere? Even at 327 million people? Not so much. Nationally, the US population density is about the same as Venezuela. See here. Cut out the Acela Corridor and, though more sprawled, the Southland of SoCal, but to be fair, whack Alaska as well. The rest of the "lower 48" is probably about the same as Peru on average.

Per state rankings, Texas is about the same as Columbia. East Texas is about the same as Iran, which has had some worries, yes. Or maybe halfway between Iran and Spain. But it's far below Italy, far far below Northern Italy where the coronavirus worries hit there, and WAY below South Korea.

OK, the biggies? Since both have densely populated coastal areas, but lost of mountains and deserts? US at 34 people per square mile vs China at 145 people per square mile? No contest. Going more granular? Wuhan itself is 8 million people or so and the metro area is 19 million and it's more densely populated than US metro areas, with the possible partial exception of a NYC or Bay Area. It's in Hubei province, which at 78,000 square miles is about 12-13 percent bigger than Oklahoma, but with 58 MILLION people. Understand?

Population density is no guarantee in offering a measure of protection, and of course, I'm not claiming that it offers anywhere near full protection. But, in much of less-populated America, it DOES offer a measure of protection, in my guesstimate.

===

Update: Per a Facebook discussion which I just exited, this is part of why it's not totally fair to compare "the West" to "Asia." The person is, for now, still a FB friend but he's been moved to "acquaintances."

As far as Asian countries controlling it well? China's been lying since the start and I wouldn't trust its numbers. Singapore and Hong Kong are city states, one still semi-authoritarian and the other under the thumb of Beijing. South Korea is, it seems, doing a good job, but it's the size of Indiana with 50 million people. Taiwan is the size of Maryland. So, that alone isn't fair.

Nor is looking at only the US and Italy and saying that's "the West." As I told said person before exiting the conversation, that ignores countries like Germany that are handling it well.

No comments:

Post a Comment

Your comments are appreciated, as is at least a modicum of politeness.
Comments are moderated, so yours may not appear immediately.
Due to various forms of spamming, comments with professional websites, not your personal website or blog, may be rejected.