Let’s say you’re walking along a river. It’s a nice autumn day in someplace like New England, and you come upon a road leading to a bridge going over the river. It’s one of those old covered bridges, which looks like a barn sitting astride the waterway, with a stretch of bridge leading out to the covered part. You can see a couple of people standing at the edge of the bridge, next to the covered part of the bridge, looking over the edge. As you get closer, they jump.
You run to the bridge and scramble down the bank. Did they fall? No, it looked like were standing shoulder-to-shoulder and they jumped at the same time, together. When you get down to the riverbed, you see not only the corpses of those two people you saw jump, but two more—all roughly in a line. All freshly dead.
So what happened here? Any reasonable, rational, logical person would come to the conclusion that all four people jumped from the same bridge together, at the same time. Right? It’s the simplest, and most logical explanation for what happened. Any other explanation would require a tremendous amount of coincidence and serendipity to explain.
Remember this story in the coming months. When it comes to figuring out the real toll from the COVID-19 epidemic, a lot of people are going to try to convince you that the only people that you can say jumped from that bridge and died are the two that you saw with your own eyes, and that you can’t use reason and logic to add the other victims to the bridge-jumping tolls.
The Difference Between Direct Observation and Statistical Analysis
The situation I described above is basically the difference between direct observation and statistical analysis. The way that we’ve been measuring COVID-19 infections and deaths thus far have been from direct observation. Testing tells us who has the virus. Those who test positive and die (or are tested post mortem) are added to the death ledger that we’re all-too familiar with these days. (A real-time, daily body count the likes of which we haven’t seen since the Vietnam War.) These figures are reliable in their certainty (minus the relatively small false positives and negatives to which all tests are prone.)
Generally, for an epidemic, this type of data collection is only useful for small-scale diseases where we need a very detailed data set to study, those diseases that are particularly lethal and we need detailed positive testing to keep it contained (like ebola), and those diseases that we know little about and thus are wary of extrapolating too much about. At first, the new coronavirus fell into this last category, but as we understand it better and realize that the spread of it is much wider than we have from direct testing, statistical models are going to be much more useful.
Statistical models are used for much more widespread diseases whose impact that can’t be adequately cataloged by merely using direct measurements. Illness and resulting deaths are compared against past years and correlated against reports of symptoms, and this results in a far more accurate number than direct measurements could ever supply. This is how the CDC measures flu outbreaks each year. (If the CDC required all flu victims to test positive for the flu virus before being directly measured, the official count of people dying from the flu would practically be zero … well, or very, very low because almost no one gets tested for the flu. You generally don’t need to. The CDC’s figures are derived from reports of symptoms and post mortem statistical analyses.)
Logically, statistical analysis makes sense when you want to make sure that you capture all cases, especially so that you can better analyze the situation in the future—accounting for margins of error, of course—without the irregularity and variability of testing capabilities and thoroughness. Direct measurement is only useful when you need to keep the sample size small and controlled (better suited to a scientific study instead of a real-world situation).
You’re standing on the bank of the river, wondering why four people would all jump off a bridge to their deaths at the same time, when another person walks up. You look up. “These four people all jumped to their deaths together,” you say.
“Did you see them all jump together?” he asks.
“No,” you say. “I saw two of them jump and then saw the other two when I got down here.”
“Then you can’t count the other two people,” he says. “You can only count the ones you saw jump with your own eyes.“
Apples-to-Oranges Comparisons
Be wary of anyone trying to force the use of direct measurement on the larger real-world COVID-19 situation. They either don’t understand the difference between it and statistical analysis, and why the latter is preferable, or they have an ulterior motive to keep numbers down. Also, these individuals tend to compare the direct measurement figures of COVID-19 deaths against the statistically derived figures of flu deaths and equivocate the two.
They’re not equal. The number of COVID-19 deaths is the floor—the minimum number of people who have died from this virus. We don’t yet know what the ceiling (or upper level) is, and we won’t until we have more extensive death figures from all nations and states. Everywhere that has seen an outbreak has reported large numbers of deaths above normal levels, many of whom died at home, often suddenly. The vast majority of these are going to have been from some COVID-19-related condition. We’ve seen that the virus can lead to blood clotting and resulting secondary problems like strokes, heart attack, and other major organ failures. We don’t know the full extent of this yet, but we know it’s more than just that the respiratory disease can lead to pneumonia.
The actual number of people dead as a result of COVID-19 is probably twice what has been reported. More specifically, it’s probably a little under twice reported in developed nations, two to three times what has been reported in hotspots in developed nations (like New York City), and much, much higher in developing nations (like Ecuador).
But those who are trying hard to dispel how dangerous the virus is are going to work just as hard to convince you that you should compare COVID-19’s apples versus the flu’s oranges. That comparison drastically inflates the number of flu deaths while purposefully minimizing the coronavirus deaths.
As you stand on the riverbed, you look downstream to where there is another bridge like this one. Underneath that bridge, there are two more dead bodies. You look over to the other person and ask, “What about them?”
“Well, they clearly jumped from that bridge,” he says.
“But how do you know that?” you ask. “Did you see them?”
“No,” he says. “But people jump off of that bridge every year.”
“Then isn’t it correct to conclude that all four of these people, and those two downstream, all jumped from the bridges to their deaths?” you ask.
“But you only saw two of these four jump, so you can’t count the the other two,” he says. “So there’s two people who jumped here and two who jumped there.”
And that’s the illogical argument that you’re going to be having with far too many people over the coming months. They’re going to keep trying to convince you that COVID-19 is no worse than the flu, but by using an apples-to-oranges comparison that drastically undercounts those we’ve lost to this pandemic. Don’t let them get away with it. We owe it to all of those we have lost this year to get the ultimate count right.
Image at top uses photo from Nick Carson at English Wikipedia.