Regression to the mean

When something is measured as extreme once, the next measurement is usually closer to average. This is regression to the mean. It happens whenever measurements include some luck or noise.

Where it was first seen

In the 1880s, Francis Galton noticed that very tall parents tended to have children who were tall, but less extremely tall than themselves. Very short parents had children who were short, but closer to average. Nothing was "pulling" heights; extreme results simply include extra luck that doesn't repeat.

Why it fools people

• Praise and punishment: a pilot who makes an unusually bad landing is criticised, and next time does better. The instructor concludes criticism works. The next landing would probably have been better anyway.
• "Sports Illustrated cover jinx": athletes appear on the cover after an exceptional season, and then their performance drops back.
• Treatment that "works": patients often seek help when symptoms are at their worst. They tend to improve anyway.

In research

If you select people, models or runs because they were extreme, expect them to look less extreme next time. That's why studies need a control group: the control group regresses to the mean too, so you can see what the treatment adds.

RememberExtreme results are partly luck, and luck doesn't repeat. Use a control group before claiming anything caused an improvement.