I admit this probably sounds particularly nerdy, but one of
my favorite classes in graduate school was a course in statistics. I’ve always
liked math okay, but not enough to make a career out of. However, I’m
fascinated by polling, sampling, and the methods by which we can accurately
assess, predict, and understand the realities amongst large bodies of people by
careful analysis of a much smaller group.
In those days, since it was part of my workload for a
Political Science degree, the topics we dealt with involved polling, public
opinion, and voting trends. But in these days so heavily dominated by pandemic,
death charts, and politicians’ pleas to “flatten the curve,” I find my mind
starting to apply some of those academic principles to our current situation.
And I’m concerned – not just about the disease, the deaths, and the devastation
to our economy. I’m concerned about our ability to understand what’s happening
and thus respond to it effectively.
A few days ago I saw this story out of North Carolina:
North Carolina health officials are not conducting random testing for the coronavirus, despite a push from one of the state’s top Republican lawmakers to provide policymakers with better tracking of the virus and more complete data on the outbreak.
That lawmaker, state Senator Phil Berger, is on to
something. Dr. Fauci himself has said that the data we get out as conclusions
is only as good as the data we put in. And what has Berger (and me) concerned
is the quality of the data we are putting in.
To put it bluntly, we don’t know how widespread coronavirus
is. All our leading medical experts acknowledge that in many instances, people
can have the virus without exhibiting almost any symptoms whatsoever. So you
conceivably have hundreds of thousands of infected victims whose bodies possess
the immunity to prevent any dire consequence from the virus. Add to that the even
larger portion of the population that may experience a few symptoms, but
nothing serious enough to take them to the hospital for testing. How many are
Well, there are smart people, like this
one from Harvard, who think it could be a massive amount:
In a column for the Washington Post, Harvard epidemiology professor Marc Lipsitch writes that as even as the number of US cases of coronavirus continues to mount, the real number may be exponentially greater – and that the implications of that fact are both complex and counterintuitive.
What he means by counterintuitive is this: it would seem to
be a bad thing if there were, as he supposes, ten times as many cases of
coronavirus in the United States than have been tested and confirmed. But in
actuality it’s a great thing.
“Paradoxically, given the level of distress we are seeing in the health care system, it is good not bad that we have many more cases. That means that the horrible outcomes we are seeing (because they are getting tested) are the tip of a bigger iceberg of milder cases.
“We can only hope the proportion of unobserved cases is large, because then we are closer to achieving herd immunity, and each bad outcome brings with it a larger number of mild outcomes that contribute to herd immunity. The math of epidemics is weird.”
Weird, yes; but incredibly important. Consider that if we
know that for every 100 people with a confirmed case of coronavirus, 1 of them
dies. That puts the mortality rate at 1%, tragically high. But if there are ten
times as many with the virus that are not counted, that would mean 1 person is
dying for every 1,000 people who have it – a mortality rate of .1%. That’s a
big difference. So big it would undoubtedly alter how we confront and respond
And this is why statistics matter. We know that random
sampling can accurately pinpoint, with remarkable precision, larger trends
based on much smaller samples. Every state governor worth their salt needs to
begin immediately commissioning these kinds of studies. Don’t wait on the
federal government to move. North Carolina needs to listen to Phil Berger, and
every other state needs to do the same.
Input affects output. And output is affecting our state and
federal policy. Too many lives and too many livelihoods are at stake to be
relying on bad input.