Esprit critique
← All biases
bias · Probabilities

Regression fallacy

Neglect of regression to the mean, Négligence de la régression vers la moyenne

Attributing a specific cause to a return towards normal that is simply explained by statistical variability.

Mechanism

An extreme result is often partly due to chance. The next measurement therefore tends to be closer to the average. Since we spontaneously look for causes, we attribute this return to whatever happened between the two measurements.

In practice

Example

An athlete makes the cover of a magazine after an exceptional season, then their results drop. People speak of the “cover curse”, when an exceptional season is rarely followed by another.

Famous case

Daniel Kahneman tells how an Israeli Air Force instructor claimed that praise made trainee pilots perform worse and that reprimands made them perform better. It was regression to the mean: after an exceptional flight, the next one is on average worse, whatever the instructor’s reaction.

Defending yourself

How to protect yourself

Before attributing an improvement or a decline to an intervention, ask whether the starting point was extreme. Compare with a control group that did not receive the intervention.

Don’t confuse

Post hoc ergo propter hoc (fallacy)

Post hoc attributes an effect to whatever precedes it; regression to the mean explains why that attribution so often misleads after an extreme result.

Going further

Source

Francis Galton (1886), on the height of the children of very tall or very short parents.

Practise

In the app, each entry offers ways to practise recognising the concept. No sign-up: your progress stays on your device.

Open the interactive entry