Definition and Galton's Discovery
Regression to the mean is a statistical phenomenon where an extreme measurement tends to be followed by one closer to the average. Francis Galton's analysis of parent-child height data showed that tall parents' children were, on average, not as tall, and short parents' children were not as short. This observation gave the statistical term "regression" its name.
How It Creates Illusions of Causation
Regression to the mean easily generates false causal beliefs. If a student performs extremely poorly and is scolded, then improves next time, the improvement may be attributed to the scolding. The score could have moved toward the average without any intervention.
Similarly, praising someone after an exceptional performance followed by a decline leads to the misconception that praise causes complacency. One way to separate regression to the mean from the effect of an intervention is a comparison with a control group.
Sports Examples
The "Sports Illustrated cover jinx," where athletes featured on the cover decline the following season, can be explained by regression to the mean. If selection for the cover reflects an unusually strong period, a return toward the average afterward could happen without any curse.
Ranking Fluctuations and Regression
If you achieve a temporarily high ranking on MyRank, a subsequent drop does not necessarily mean your ability has declined. It may simply reflect measurement variability and regression to the mean. The average of multiple measurements is less affected by this variability than a single result.