Image by Editor
There’s something almost embarrassing about how confidently humans assign cause and effect to things that are just… happening near each other. Ice cream sales go up, and so do drowning deaths. A rooster crows, and the sun rises. You wore your lucky socks, and your team won. The brain doesn’t care that correlation and causation are different concepts — it’s too busy building stories to stop and check the math.
The Brain Was Built for Pattern Recognition, Not Statistics
Humans survived for millennia by connecting dots fast. Hear a rustle in the bushes, assume a predator, run. No time for a regression analysis. That instinct is still running in the background today, even when you’re looking at a scatter plot instead of a savanna.
The brain is wired to see sequences and immediately promote them to relationships. A happens, then B happens — so A must have caused B. It’s called post hoc reasoning, and it’s basically the brain’s default factory setting.
The problem is that this shortcut, which kept our ancestors alive, is a statistical nightmare. It doesn’t account for confounders, it ignores base rates, and it completely skips the question of whether there’s any plausible mechanism connecting the two things at all.
Spurious Correlations Are Everywhere (And Surprisingly Convincing)
Tyler Vigen’s famous Spurious Correlations project showed that U.S. spending on science, space, and technology correlates almost perfectly with people taking their own lives, with a whopping correlation of about 99.8%. The r-value is genuinely high. The chart looks convincing. And yet nobody in their right mind would argue for a causal link there.
But here’s the thing — when the data involves something we already believe, we stop being skeptical. Studies showing that coffee drinkers have lower rates of certain diseases get treated as near-proof that coffee is protective.
The correlation is real, but the leap to causation requires ruling out a dozen alternative explanations: lifestyle factors, socioeconomic status, and dietary habits. Coffee drinkers might just be a particular kind of person, and that person’s overall habits are doing the work.
Confounders Are the Real Villains Here
A confounder is a third variable that’s quietly influencing both of the things you’re measuring, making them look related when they’re not really talking to each other at all. It’s the hidden hand behind a lot of misleading statistics.
Shoe size and reading ability are correlated in children — not because bigger feet make you smarter, but because older kids have both. Age is the confounder.
Once you control for it, the relationship disappears. The tricky part is that confounders aren’t always obvious. Sometimes they’re buried in demographic data, and sometimes they’re genuinely hard to measure.
Randomized controlled trials became the gold standard in research precisely because randomization distributes confounders evenly across groups, neutralizing them. Observational data, which is most of what exists in the real world, doesn’t have that luxury.
Why We’re So Bad at Updating When We Should
Even when people are shown that a correlation has a confounding explanation, they often hold on to the original causal story. There’s a well-documented phenomenon in psychology called belief perseverance.
How it works is this: once a causal narrative takes root in your head, contradicting evidence has to work twice as hard to dislodge it. The story feels right. It fits the mental model. Replacing it with “well, there’s actually a third variable” is cognitively unsatisfying in a way that’s genuinely uncomfortable.
This is compounded by media coverage. Headlines are written for attention, not precision. “X is linked to Y” becomes “X causes Y” somewhere between the press release and the tweet.
Readers absorb the causal claim, not the correlation caveat buried in paragraph four. By the time a correction or a follow-up study comes out, the original story has already calcified into received wisdom.
The Role of Plausibility (And How It Can Work Against You)
One of the subtle ways causal thinking sneaks past our defenses is through plausibility. If a correlation has a story that makes intuitive sense, we’re far more likely to accept it as causal without demanding evidence of the mechanism.
Red wine and heart health, for example — there’s a reasonable-sounding biological story about antioxidants and resveratrol, which made it easier to believe the correlation was telling us something real. It took years of more rigorous work to complicate that picture significantly.
Plausibility is genuinely useful as a starting point. Scientists use it to generate hypotheses worth testing. But it’s not the same as evidence, and treating a good story as a substitute for a controlled study is where a lot of people, including researchers who should know better, get into trouble.
How to Actually Think About Causality
When you see a statistical relationship being reported as causal, there are a few honest questions worth asking. What’s the study design? Observational data can reveal associations, but it can’t establish causation on its own without very careful controls.
What are the plausible confounders, and did the researchers account for them? Has the finding been replicated, or is it a single study with a catchy result?
It’s also worth sitting with the discomfort of not knowing. Sometimes the data shows a relationship, and we genuinely don’t know yet whether it’s causal, confounded, or coincidental.
That’s not a failure of statistics — it’s just where the knowledge currently stands. Tolerating that ambiguity is one of the harder parts of thinking statistically, but it’s also one of the most important.
Final Thoughts
The pull toward causal thinking isn’t a flaw you can reason your way out of entirely. It’s structural. But knowing it’s there changes how you read a chart, how you respond to a headline, and how much confidence you put in a pattern before you’ve asked the harder questions. Statistics doesn’t give you certainty — it gives you better tools for being wrong less often. That’s not a consolation prize. That’s actually the whole game.
