Quick Definition
Extreme performance tends to be followed by more moderate performance, simply due to statistical probability.
Definition
Regression to the mean is a statistical phenomenon where, after an extreme event or measurement, the next measurement is likely to be closer to the average. This occurs because extreme outcomes are often partly due to luck or random variation, which is unlikely to repeat.
When an outcome is unusually good or bad, it typically reflects a combination of skill or underlying ability plus random factors. The random factors are unlikely to align the same way in subsequent measurements, so performance tends to "regress" toward the average. This principle is often misunderstood or misattributed to other causes.
Origin & History
The concept was first systematically described by Sir Francis Galton in 1885, after studying the heights of parents and their children. Galton observed that tall parents tended to have children who were tall but not as tall as themselves—a regression toward the mean height.
Galton coined the term "regression" in statistics from this observation. Daniel Kahneman and Amos Tversky highlighted regression to the mean as a major factor in human misprediction of future performance, showing that people systematically fail to account for the role of luck in extreme outcomes.
Key Principles
- Establish the baseline - Determine what the average or expected performance is
- Identify extreme performance - Note when an outcome is significantly above or below baseline
- Consider the role of luck - Ask what proportion might be due to factors beyond consistent skill
- Expect moderation - Anticipate the next measurement will likely be closer to average
- Resist causal misattribution - Before attributing to causes, consider random variation
When to Use
- Evaluating performance of individuals or teams
- Making attributions about causes of outcomes
- Expecting future outcomes
- Interpreting test results or measurements
- Evaluating investment returns
- Assessing training or intervention effectiveness
How to Apply
- Establish the baseline - Determine what the average or expected performance is
- Identify extreme performance - Note when an outcome is significantly above or below baseline
- Consider the role of luck - Ask what proportion might be due to factors beyond skill
- Expect moderation - Anticipate the next measurement will be closer to average
- Resist causal misattribution - Before attributing to causes, consider random variation
- Avoid overreaction - Don't change strategies based solely on extreme outcomes
- Gather more data - The more measurements, the more you can distinguish signal from noise
Real-World Example
Sports Performance: A basketball player who shoots exceptionally well in one game is likely to shoot somewhat worse in the next game—not because they are lazy or unlucky, but because their extraordinary first-game performance likely included factors that won't repeat. Similarly, a terrible shooting night doesn't indicate a permanent decline.
Common Pitfalls
- Misattribution to intervention - Crediting improvement after poor performance to corrective actions when regression would have occurred anyway
- Misattribution to punishment - Blaming decline after good performance on criticism when regression alone explains it
- Small sample overconfidence - Drawing conclusions from one or two measurements
- Forgetting base rates - Not considering how extreme an outcome is relative to normal distribution
- Regression fallacy - Assuming regression is the explanation when there might be a real causal factor