Skip to content
Omnikalc

Reliable change index calculator: is a change in score real or measurement error?

Enter a person's score before and after, the scale's standard deviation and its reliability to get the RCI, the p-value and the smallest reliable change.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Value out of range

Results

Reliable change index (RCI)

-2.556

Reliable change (beyond measurement error)
Standard error of measurement (SEM)
3.873
Standard error of the difference (Sdiff)
5.477
Smallest change that is reliable at 95 %
10.74
p-value (two-sided)
0.0106

The reliable change index, or RCI, tells you whether a person’s change in score is larger than measurement error. The reliable change index calculator takes the scores before and after, the standard deviation of the scale and its reliability, and gives the standard error of measurement, the RCI, the p-value and the smallest reliable change.

How the calculation works

  • Compute the standard error of measurement: SEM = s × √(1 − r), where s is the standard deviation of the scale in a reference group and r its test-retest reliability.
  • The standard error of the difference between two scores of the same person is Sdiff = √2 × SEM, because both scores carry measurement error.
  • The reliable change index of Jacobson and Truax is RCI = (score after − score before) ÷ Sdiff.
  • An absolute RCI above 1.96 means the change is unlikely, with 95 % confidence, to come from measurement error alone. The smallest reliable change is 1.96 × Sdiff.

Example

A person scores 62 before treatment and 48 after, on a scale with a standard deviation of 10 and a reliability of 0.85. SEM = 10 × √0.15 = 3.873, Sdiff = 5.477, and RCI = −14 ÷ 5.477 = −2.556. The p-value is 0.0106 and the smallest reliable change is 10.74 points, so the 14-point drop is a reliable change.

Why it matters

Scores move from one occasion to the next even when nothing has changed. Without a yardstick it is easy to over-read a small difference as progress or decline. The RCI gives a clear threshold for when a change is more than noise.

Good to know

The index treats the reliability and the standard deviation as known values from a good reference sample. It does not correct for practice effects or regression to the mean, which can make a second score drift even for a stable person. Read it as a screening threshold and not as proof of change.

The standard error calculator covers the error of a sample mean, and the paired t-test calculator tests change at the group level.

Frequently asked questions

What is the reliable change index?

It is a way to judge whether the change in one person's score between two occasions is larger than what measurement error alone would produce. Jacobson and Truax proposed it in 1991 for therapy research, and it is now used in clinical psychology and neuropsychology.

Why is it different from a t-test?

A t-test compares group means, while the RCI judges a single individual. It uses the reliability of the scale to estimate the noise in one person's scores, so the result applies to that person and not to an average.

Does a reliable change mean a meaningful change?

No. A change can be larger than measurement error and still be too small to matter in daily life. The RCI answers whether the change is real, and a clinical cut-off or minimal important difference answers whether it is large enough to care about.

Which reliability should I use?

Use test-retest reliability if the scores come from two sessions apart in time, as it includes day-to-day variation. Internal consistency such as Cronbach's alpha is often used in practice, but it gives a smaller error and so a smaller threshold than test-retest would.

Was this calculator helpful?

Thanks for your feedback!

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