Regression to the mean is the tendency for unusually extreme measurements to be followed by values closer to the typical average when the same process is measured again. The core aim of Science mastery is not to teach students that extreme values are somehow pulled toward the mean by a physical force. It is to help them understand how random variation makes extreme observations less likely to repeat exactly.
For students and parents searching for regression to the mean, regression toward the mean, regression to mean examples, extreme values, repeated measurement or why scores return to average, the most useful principle is this: an extreme observation often contains both a real component and an unusually large random component. When measured again, that random component is unlikely to be equally extreme.
This matters because ordinary statistical behaviour can look like a treatment effect.
The 60-Second Idea
Suppose someone is measured on an unusually bad day.
The next measurement is likely to be closer to their typical level.
Likewise, an unusually excellent performance may be followed by a more ordinary one.
No intervention is required.
The change can arise simply because the first measurement was extreme.
Wait, What? Improvement After Treatment Does Not Always Mean the Treatment Worked?
Correct.
People often seek treatment when symptoms are unusually severe.
Even without effective treatment, symptoms may later return closer to their normal level.
If there is no control group, this natural statistical tendency can be mistaken for treatment success.
This is one reason controlled studies matter.
A Worked Example: Test Scores
A student usually scores around 75.
One day, because of poor sleep and difficult questions, the student scores 48.
On the next test, the student scores 72.
The increase does not necessarily mean a new study method caused a 24-point improvement.
The first score may have been unusually low relative to the student’s typical level.
A Worked Example: Sports Performance
An athlete records a personal-best performance far above their usual level.
The next performance is closer to the athlete’s average.
This may look like decline.
But part of the original personal best may have reflected unusually favourable random factors.
Regression to the mean explains why exceptional performances often do not repeat immediately.
Why Extremes Regress
Suppose a measurement contains:
true underlying level + random variation.
An extreme score often requires:
- a high underlying level;
- plus unusually favourable random variation;
or:
- a low underlying level;
- plus unusually unfavourable random variation.
On the next measurement, the random component is unlikely to be equally extreme.
Regression to the Mean Is Not Measurement Error Alone
Measurement error can contribute, but regression to the mean can occur whenever repeated observations contain random variability.
Examples include:
- blood pressure;
- sports scores;
- school performance;
- weather measurements;
- biological markers.
Why Control Groups Matter
Suppose patients are enrolled only when symptoms are severe.
After treatment, symptoms improve.
Without a control group, possible explanations include:
- real treatment effect;
- placebo effect;
- natural recovery;
- regression to the mean.
A control group helps distinguish these possibilities.
See Control Group.
Regression to the Mean and Placebo Studies
Placebo-controlled studies are especially important when participants enter because symptoms are unusually severe.
Both treatment and placebo groups may improve partly because of regression to the mean.
The difference between groups helps isolate the active treatment effect.
See Placebo Effect.
Selection on Extreme Values
Regression to the mean is strongest when people or samples are selected specifically because their first measurement is extreme.
Examples:
- lowest-performing students;
- highest blood pressure readings;
- worst-performing factories;
- best-performing athletes.
When remeasured, average performance often moves inward even if nothing changes.
A Worked Example: School Intervention
A school selects the 20 lowest-scoring students for a new programme.
On the next test, average scores rise.
Possible explanation:
the programme helped.
But another explanation is regression to the mean because the group was selected for extreme low scores.
A stronger evaluation would compare them with a suitable control group.
Regression to the Mean vs Trend
Regression to the mean is not the same as a long-term trend.
Trend:
values systematically increase or decrease over time.
Regression to the mean:
extreme values are followed by values closer to typical levels because of random variation.
Students should not confuse the two.
Regression to the Mean vs Causation
A before-and-after change does not automatically prove the intervention caused the change.
Whenever participants are selected because they are extreme at baseline, regression to the mean should be considered.
Primary Science Foundations
Primary learners can understand the simpler idea:
One unusually high or low result may not repeat because measurements naturally vary.
This encourages repeated measurement instead of overinterpreting one extreme result.
Secondary Science Regression to the Mean
Secondary students should increasingly connect the idea to:
- extreme-value selection;
- before-and-after studies;
- control groups;
- random variation;
- treatment evaluation.
How to Practise
For any before-and-after study, ask:
- Were participants selected because their first value was extreme?
- How variable is the measurement?
- Is there a control group?
- Could ordinary variation explain part of the change?
Common Mistakes
- thinking regression to the mean is a physical force;
- assuming every improvement after treatment is causal;
- ignoring extreme-value selection;
- confusing regression to the mean with trend;
- using one extreme baseline without repeated measurements.
Frequently Asked Questions
What is regression to the mean?
It is the tendency for extreme measurements to be followed by measurements closer to the typical mean when random variation is present.
Why does it happen?
Extreme measurements often contain unusually large random components that are unlikely to repeat exactly.
Why does it matter in experiments?
It can make an intervention appear effective when part of the change would have happened anyway.
How do control groups help?
They show how much change occurs without the active intervention, helping separate treatment effects from regression to the mean.
Useful eduKateSG Routes
The Core Aim
Regression to the mean warns us not to overinterpret movement away from an extreme starting point.
Repeat measurements. Use control groups. Separate ordinary variation from real intervention effects.
That is the core aim: avoid turning statistical return-to-normal into a false story of cause and effect.
Properly taught kids shine a bright light into the future.