True confessions time: I’m a podcast addict. I listen when I’m cooking dinner, when I’m walking the dog, when I’m knitting. Most of the podcasts I listen to are about politics, which I chalk up to having spent 30 years of my life in Washington, D.C., where politics is the local business, sport, and familiar background noise.
The hosts of my favorite politics podcasts seem like personal friends after all these years, and so it gives me no joy to report that, as a rule, they sort of stink at science. Worse, they freely admit that they don’t get science. Indeed, sometimes they even gloat about it, as if it’s okay to say, “One huge area of knowledge, critical for understanding the complex world around me, is just a big swirling mess as far as I’m concerned! Ha-ha! Funny!”
Honestly, I love you, David Plotz, but you are a serial offender in this regard.
Nevertheless, I have @davidplotz to thank for bringing to my attention the subject of this article: a study that suggests that wearing eyeglasses may protect you from contracting the coronavirus. This study is a great example for teachers to use to show students how to evaluate scientific evidence, and especially how to ask good questions about correlation and causation.
First of all, let’s observe that humans have a handy skill: the ability to recognize patterns. Night follows day; rotten food makes you sick; lists of three examples are more amusing when they end with an incongruous item.
Humans are not the only animals that can recognize patterns, of course. I suspect your dog knows that when you start putting on your shoes it’s time for a walk (yay!), unless you’ve put on your party shoes and grabbed your car keys (which means it’s time for a nap). But clearly humans take the skill to a higher level.
Many of science’s great discoveries were the result of recognizing a pattern: Edward Jenner noticed that milkmaids never contracted smallpox, for example, recognizing a pattern that then led him to develop the very first vaccine.
Pattern recognition can go wrong, of course. For starters, we can detect patterns that aren’t actually there, like the constellations that the ancient Greeks discerned in the night sky, and we can fail to recognize patterns that are there, like the correlation between puerperal fever mortality and lack of handwashing that Ignaz Semmelweis’s colleagues failed to recognize in mid-19-century Vienna.
A more complicated way in which pattern recognition can go wrong involves getting so attached to a perceived pattern that exceptions or contradictions are overlooked or ignored. (The formal term for this is confirmation bias.) For example, pathologists were so convinced, for decades, that the stomach was too acidic for bacteria that they failed to see bacteria that were clearly visible in biopsies of stomach ulcers. They simply filtered out information that was inconsistent with their mental pattern. Besides, physicians had decided that another pattern explained ulcers: people with high stress lives got ulcers; therefore stress caused them. It took Barry J. Marshall and J. Robin Warren to demonstrate that, in fact, ulcers are caused by a bacterium called Helicobacter pylori, a discovery for which they won the Nobel Prize for Physiology or Medicine in 2005.
Pattern recognition is only the beginning, of course. Once we see a pattern, we think that there has to be a reason for it. The pattern itself is the “what”; the next step is to seek the “why.” Figuring out the “why” is a big part of science, but it’s tricky. Suppose that we have recognized a simple pattern, say, the presence of H. pylori is correlated with the occurrence of ulcers. Then how do we decide among the possible hypotheses, including that the bacteria cause the ulcers, that the ulcers encourage the bacteria, that a third factor causes both the bacteria and the ulcers, that the pattern is just a coincidence? And that’s just a simple pattern! Volumes could be—and have been—written about the best ways to develop and test scientific explanations of observed patterns.
OMG, Ann! Are you ever going to get to the eyeglasses study?
Yes. Here we go. Doctors in a hospital in China noticed that very few of their COVID-19 patients seemed to wear glasses (aha: a pattern). They decided to investigate. They interviewed all 276 patients hospitalized with COVID-19 in their hospital between January 27 and March 13, 2020, asking them whether they wore glasses more than eight hours per day. They found that only 16 patients reported wearing glasses all day—just 5.8%.