Regression to the mean examples always fascinate me because they show up in so many areas of life. The basic idea is that extreme measurements tend to be followed by more average ones.
I've seen this in sports (the "Sports Illustrated jinx"), in business performance, and even in test scores. But I'm looking for more concrete regression to the mean examples that people might encounter in their daily lives.
How would you explain this concept to someone who's never studied statistics? And what are some common mistakes people make by not understanding regression to the mean?
Sports performance is full of regression to the mean examples. A rookie has an amazing first season, then performs worse in the second season. People call it a sophomore slump," but it's often just regression to the mean.
The rookie's first season performance was probably above their true ability level. The second season is closer to their actual average performance.
Same with the "Sports Illustrated jinx" - athletes who appear on the cover often perform worse afterward. But they were probably performing at an unusually high level to get on the cover in the first place. Subsequent performance regresses toward their mean.
In business, regression to the mean explains why companies that are extreme performers one year often become more average the next. The best companies to work for" lists - companies that score extremely high one year often score lower the next, not because they got worse, but because extreme scores tend to regress toward the mean.
This causes problems when managers reward or punish based on single measurements. A salesperson has a fantastic month, gets a big bonus, then has an average month. The manager thinks the bonus caused complacency, but it might just be regression.
Understanding regression to the mean helps avoid overreacting to random fluctuations.
Test scores are another classic example. Students who score extremely high on one test often score lower on the next, and vice versa. This isn't because they got smarter or dumber - it's because test scores have a random component.
Measurement error means that a student's observed score = true ability + random error. Extreme scores are more likely to have positive error. On the next test, the error is likely to be smaller or negative, so the score regresses toward the true mean.
This is why we shouldn't make big decisions based on single measurements. Multiple measurements give a better estimate of true ability.