Regression to the mean: extreme measurements on one trial tend to be less extreme on the next, because the first trial included luck as well as skill. Galton documented it; Kahneman shows humans lack intuitive grasp of it.
Misread as causation
Flight instructors concluded punishment works and praise fails: after great landings, performance worsens; after poor landings, improves. The true story is regression — they praised good landings and scolded bad ones, so regression looked like treatment effect.
Same pattern: tall parents have slightly shorter children; high first-year students regress in GPA; golfers' day-2 scores regress from day-1 extremes.
Prediction fix
Intuitive forecasts are nonregressive — as extreme as the evidence. Correct method: start at baseline (mean), move toward intuition by a fraction equal to the evidence-outcome correlation. See inside-view-outside-view and taming intuitive predictions in thinking-fast-and-slow.