Skip to content
Omnikalc

Regression to the mean calculator

Enter the population mean and standard deviation, the correlation between the two measurements and an extreme first score to get the expected second score, how far it moves back towards the mean and a prediction interval.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Value out of range

Results

Expected second score

121.00

Expected move back towards the mean
9.00
Share of the gap that disappears
30.0 %
Moderate regression
Standard deviation of the second score
10.712
Lower 95 % limit of the second score
100.00
Upper 95 % limit of the second score
142.00
First score in standard units
2.000
Expected second score in standard units
1.400

The regression to the mean calculator gives the expected second score after an extreme first score, the share of the gap that disappears and a 95 % prediction interval.

How the calculation works

  • Enter the mean and standard deviation of the population.
  • Enter the correlation between the two measurements and the first score.
  • The calculator gives the expected second score, the move back, the share of the gap and the interval.

Example

An intelligence score has a mean of 100, a standard deviation of 15 and a test-retest correlation of 0.7. A first score of 130 is 2 standard deviations above the mean. The expected retest is 100 + 0.7 × 30 = 121.00, so the score moves back by 9.00, which is 30.0 % of the gap. The 95 % interval of the retest runs from 100.00 to 142.00.

Why it matters

The calculator helps students in psychology, medicine and education, researchers who design before-and-after studies, and managers who judge the results of a program after an exceptional period.

Good to know

The formula assumes that the two measurements have the same mean and variance and follow a bivariate normal distribution. The effect is only about chance in the measurement and in the situation, not about any actual change. To test an intervention fairly, use a control group selected in the same way, or use the first measurement to select and a separate one to judge. The same logic explains why the best performers of one year often do less well the next, and why a bad week often looks better the following week. When the correlation is negative, the formula still applies, and the second score is expected to fall on the other side of the mean.

The z-score calculator standardizes values, the percentile calculator ranks them, the normal distribution calculator gives probabilities and the reliable change index calculator tells whether a change exceeds measurement error.

Frequently asked questions

What is regression to the mean?

When a measurement is partly due to chance, an extreme result tends to be followed by a less extreme one, even if nothing changes. The expected second score is the mean plus the correlation times the first deviation: μ + ρ × (x − μ).

Why does it matter?

People selected for an extreme score, such as the worst patients or the best students, will on average move toward the middle on retest. Without a control group, this natural drift can be mistaken for the effect of a treatment or a program.

What correlation should I use?

Use the test-retest correlation of the measure. The lower the reliability, the stronger the regression. A correlation of 1 means no regression, and 0 means the second score is expected to equal the mean.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.