Regression Bias
The tendency to misinterpret natural regression to the mean as being caused by specific actions or interventions, leading people to take credit for changes that are statistically expected.
THE IDEA
Understanding Regression Bias
Regression bias occurs when a change following an unusually high or low result is attributed to an intervention without considering regression to the mean. If an extreme result partly reflects chance, a later observation may be closer to the average even when nothing has changed.
This is a statistical tendency, not a requirement that every extreme result be followed by an average one. To assess an intervention, compare outcomes with what would be expected without it rather than relying on a single before-and-after change.
When to use it
Evaluating employee performance after a record-setting month
Interpreting medical treatment efficacy when symptoms naturally wax and wane
Making investment decisions based on past extreme returns
Adjusting coaching strategies after an athlete's outlier performance
FROM THEORY TO EVERYDAY LIFE
In practice
Stock Market 'Expert'
After unusually high returns, a fund’s performance drops toward its average. The manager blames market conditions. Regression to the mean is another possibility; these two observations alone do not identify the cause.
Student Performance
A student scores 95% on a midterm, well above her usual average of 80%, then scores 82% on the final. A tutor treats the decline as evidence that tutoring failed. The unusually high midterm score and possible regression to the mean also need consideration.
A MOMENT TO REFLECT
Put it to work.
When have I attributed a change in outcomes to my actions when they might have simply regressed to the mean?
How can I distinguish between genuine improvement and regression to the mean when evaluating performance?
In what situations am I most prone to take credit for natural fluctuations?
Often confused with
References
- Thinking, Fast and Slow — Daniel Kahneman (2011)
- Regression towards mediocrity in hereditary stature — Francis Galton (1886)
CONTINUE THE CONVERSATION
Source texts
- 60 Cognitive Biases That Ruin Everything We Do by David Scott