A struggling athlete has a terrible game. The coach screams at him. Next game, he plays better. The coach concludes that yelling works. But what if the improvement would have happened anyway?

This is the regression fallacy, one of the most common reasoning errors in everyday life. When we observe an extreme outcome followed by a more typical one, we instinctively credit whatever happened in between. Yet statistics has a quieter explanation: extreme values tend to be followed by less extreme values, regardless of intervention. Learning to recognise this pattern protects us from inventing causes where none exist.

Imagine measuring the height of every adult in a city. Most cluster near the average, while extreme heights are rare. This shape, the familiar bell curve, governs countless natural phenomena, from test scores to blood pressure to athletic performance.

Now consider any single measurement that lands far from the average. Performance on a given day reflects two ingredients: an underlying ability and a dose of random variation. When someone performs exceptionally well or poorly, both ingredients usually contributed. Their next performance will likely reflect their true ability plus a fresh, smaller dose of luck. The result? They move closer to average.

This phenomenon, called regression to the mean, was first identified by Francis Galton in the 1880s. He noticed that tall parents tended to have children shorter than themselves, and short parents had children taller than themselves. Nothing was pulling them toward average except mathematics itself.

Takeaway

Extreme outcomes contain extreme luck, and luck doesn't repeat. Expect the spectacular and the catastrophic to soften on their own.

Consider a familiar pattern. A patient seeks treatment when symptoms are at their worst. The doctor prescribes something. Days later, the patient feels better and credits the treatment. But people typically visit doctors during symptom peaks, and peaks, by definition, tend to subside.

The same logic applies broadly. A school launches a reading program after a year of dismal test scores; scores rise the next year. A business hires a consultant after a quarterly slump; performance recovers. A speed camera is installed at an intersection with unusually high accidents; accidents drop. In each case, the intervention might have helped. It might also have done nothing while regression did the work.

This is why anecdotal evidence misleads so persistently. We notice interventions following extreme moments precisely because extreme moments demand action. The improvement that follows feels like proof, but it's often just the statistical tide pulling outcomes back toward their natural level.

Takeaway

When you intervene at an extreme moment, improvement is almost guaranteed regardless of what you did. The harder question is whether you intervened better than doing nothing.

How do we distinguish genuine cause from regression's illusion? The classical answer is the controlled comparison. We take two groups starting from the same extreme position, give one the intervention, and leave the other alone. If both improve equally, the intervention did nothing. If the treated group improves more, we have evidence of a real effect.

This is why medicine relies on randomised controlled trials and why thoughtful researchers always ask, compared to what? Without a baseline showing what would have happened naturally, any claim of causation rests on faith. The control group reveals the invisible force of regression by letting us see it operate untouched.

You can apply this logic informally too. Before crediting a coach, a diet, or a policy with success, ask whether similar cases without that intervention also improved. If extreme outcomes tend to soften everywhere, your favoured cause may be claiming credit for work it didn't do.

Takeaway

A claim of cause means little without knowing what would have happened otherwise. Always ask, compared to what?

Regression to the mean is invisible because nothing seems to cause it. There is no agent, no force, no intervention. Just the quiet mathematics of variation returning to its centre.

Once you see this pattern, you see it everywhere: in business turnarounds, medical recoveries, athletic slumps, and the rise and fall of reputations. The discipline of good reasoning is not to deny that interventions work, but to demand the evidence that distinguishes real effects from the natural rhythm of extremes returning home.