Actually, correlation kinda does imply causation
Causality inference is a potent defining feature of human intelligence that serves us well in many situations. Our ability to suss out how things work is largely based on this “one weird trick” that our brains can do. Flick the switch, light turns on … probably causally related!
- touch fire 👉🏻 get burned
- throw rock 👉🏻 break window
- eat too much 👉🏻 get sick
There are countless simple correlations like this that we master effortlessly before we can even tie our shoes. Infants can actually see simple causality in billiard-ball animations! (Probably.) We see B follow A and we just kinda get it that A caused B.
Unfortunately, we also get it wrong a lot. Like when we believe that a treatment worked because our pain seemed to ease right after it. Maybe it did, and maybe it didn’t.
We especially fumble causality detection in health and medicine. Like when we are trying to judge the effect of a treatment for pain. Or interpret the meaning of gua sha bruises. Or grok the correlation between stretching and reduced mortality. Or decide if a tendon tore because of a drug we’re taking. Or whether noisy joints are doomed to fail from arthritis. To name a few thorny examples from just the last couple months on this blog.
This problem underlies practically everything on PainScience.com. Posts like this dig down to the bedrock of how we figure out what actually helps with pain and injury. Take us to Nerd Factor 9, Mr. Sulu!
Causality inference is a double-edged sword
We are bad at understanding complex causality for the same reason we are good at detecting simple causality — we’re “sensitive” to it. So we get causality right constantly when the variables are simple and readily observable. But we rarely get it right in health care, or any other complex endeavour, where there are many variables and many are subjective or otherwise murky. What’s really going on in a causal relationship almost always turns out to be different and waaaaay more complicated than we thought.
Our failures in this department are legion and disastrous. By far the most important thing anyone needs to understand about the relationship between correlation and causation is that A did not necessarily cause B just because B followed A, and making this mistake is one of the Greatest Hits of human thinking glitches.
That should be emphasized every time correlation is discussed, because, as Barker Bausell put it (Snake Oil Science), we have a problem with “confusion between correlation and cause on an industrial scale.” Many amusing examples of spurious correlations have been mined from data. This problem has been trumpeted ad nauseam by so many smart people for so long that it seems like an unassailable edifice.
And yet …
Actually, correlation kinda does imply causation
The famous rule — “correlation does not imply causation” — is in fact a misleading oversimplification. At the very least it’s missing a word, and it should be “correlation does not necessarily imply causation.” (Although “equal” sometimes replaces “imply,” the more common usage is definitely imply.)
Or maybe we should just rephrase it entirely. Edward Tufte, an American statistician who made the same point quite a while ago, suggested that a good informal re-wording would be:
“Correlation is not causation but it sure is a hint.”
Edward Tufte
Because correlation actually does “imply” causation, and many (if not most) events that occur in sequence that appear to be causally related are in fact causally related. Their correlation is not a coincidence. Clapping makes noise, braking stops cars, coffee makes you poop.
Our brains are tuned to detect those relationships, and that mental superpower served us well as we grew up as a species. The problem is that we’re so good at it that we overdo it. Our brains are reckless with it. We see causality everywhere, even when it isn’t there — “illusory correlation” (Chapman, 1967), yet another form of apophenia (or “patternicity,” Shermer, 2008), “the tendency to perceive meaningful connections between unrelated things.” We aggressively over-detect and over-interpret all kinds of things: faces, creepy crawlies, kinship cues, symmetry, emotionality in voices, heights and edges! And many, many more. Brains are stuffed with these biases.
So we think we understand causality in many situations where it is actually much too difficult to know. And very few situations are as bewildering and counter-intuitive as physiology and medicine.
The difference between general and specific causality inference
One of the things that trips up our causality inference superpower the most is the tricky difference between inference of causality and the attribution of mechanism. General versus specific causes, basically. We can and routinely do correctly detect causes when correlation gives us a strong enough hint, but we routinely screw up exactly what caused what.
In other words,
For instance, most people will assume that when a very stubborn old pain goes away during a one-hour acupuncture session that the experience must have caused the relief, because the relief followed the experience. And that assumption is probably correct. The appearance of relief probably isn’t a coincidence, probably not just regression to the mean (too quick). Something about the experience was probably helpful in some sense.
But what and how exactly? Most people will then (carelessly or self-servingly) move on to the next “logical” assumption: that the treatment caused the relief because acupuncture works as advertised. Alas, it almost certainly does not, despite seemingly credible evidence of efficacy.
We can be right about the causality in a wide view, but still be hopelessly wrong about what specifically caused what. Most people will simply ignore the possibility that the true mechanism of relief was not the efficacy of acupuncture, but the power of a caring professional promising aid and performing fascinating rituals that reek of implied potency. And the power of “surely no one would do this if it didn’t work!” These factors are wildly underestimated by most acupuncture patients. And acupuncturists.
The hint is real enough; what it’s hinting at is often not what you think.
This blog post is a slightly adjusted excerpt from one of my weirder full articles, The Power of Barking: Correlation, causation, and how we decide what treatments work. Among other things, it goes on to answer the question, If a treatment works, who cares if it’s not for the reason we think? Which maybe the most common objection to skepticism about any kind of medicine people want to believe in — and one of the hardest to respond to.