Reliable Change Index
Also known as: RCI, Reliable Change Index (Jacobson-Truax), Jacobson-Truax Reliable Change, Reliable Change Criterion
The Reliable Change Index (RCI) is a statistic that tells whether the change in an individual client's score on a measure, from before to after an intervention, is large enough that it is unlikely to be an artifact of the instrument's measurement error. Introduced by Neil Jacobson and Paula Truax in 1991 as one half of their two-part definition of clinically significant change, it converts a pre-post difference into a standardized value and compares it against a critical cutoff, typically 1.96, so that practitioners and researchers can classify each client as reliably improved, unchanged, or reliably deteriorated.
Key highlights
- Gives a per-client, not just per-group, verdict on whether change is real, directly serving accountable individual practice.
- Requires only pre and post scores plus published reliability and standard deviation, so it is computable in everyday settings.
- Anchors the interpretation of change to the measure's own error, preventing over-reading of trivial score movements.
- Forms half of the widely adopted Jacobson-Truax framework, enabling standardized reporting of recovery rates across studies.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use the RCI when you have pre and post scores from a reliable, normed measure and you want to judge, for each individual client, whether their change is more than measurement noise — a core need in single-system evaluation, routine outcome monitoring, and practice accountability. It pairs naturally with a clinical-significance cutoff to form the full Jacobson-Truax classification. It is inappropriate when you only have a single time point, when the measure's reliability is unknown or poor, or when you want a group-average effect, for which standard effect sizes are more suitable. It also assumes the instrument's error structure in your client resembles that in the normative sample.
Strengths & limitations
- Gives a per-client, not just per-group, verdict on whether change is real, directly serving accountable individual practice.
- Requires only pre and post scores plus published reliability and standard deviation, so it is computable in everyday settings.
- Anchors the interpretation of change to the measure's own error, preventing over-reading of trivial score movements.
- Forms half of the widely adopted Jacobson-Truax framework, enabling standardized reporting of recovery rates across studies.
- Depends entirely on the accuracy of the reliability estimate borrowed from a normative sample, which may not match the client population.
- Assumes measurement error is equal and normally distributed at both occasions; skewed or floor/ceiling-bounded scales violate this.
- Treats the cutoff (1.96) as a sharp threshold, so clients just inside and just outside the band are classified differently despite near-identical change.
- Ignores regression to the mean unless a corrected variant (e.g., Hsu or Speer) is used, which can bias classification for extreme baseline scores.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
Why is the critical value 1.96 and can I use a different one?
The value 1.96 is the two-tailed z critical value for a 95% confidence level under a normal distribution, so an RCI beyond it means the change would occur by measurement error less than 5% of the time. Some applications use a one-tailed 1.645 when only improvement is of interest, or a stricter value for higher confidence, but 1.96 is the convention Jacobson and Truax used and remains standard for comparability.
What is the difference between reliable change and clinical significance?
They answer different questions. The RCI asks whether the amount of change exceeds measurement error — is the change real? Clinical significance asks whether the client has moved from the dysfunctional range into the functional or normal range — does the change matter? Jacobson and Truax require both: a client is 'recovered' only if they show reliable change and end below the clinical cutoff.
Which reliability coefficient should I plug in?
Conceptually the RCI concerns the stability of scores over the test-retest interval, so a test-retest reliability coefficient is the most defensible choice. Many published applications substitute internal-consistency reliability (Cronbach's alpha) because it is more readily available, but this can overstate stability and should be reported transparently. The chosen coefficient should come from a sample comparable to your clients.
Sources
- 1.Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59(1), 12–19.
- 2.Christensen, L., & Mendoza, J. L. (1986). A method of assessing change in a single subject: An alteration of the RC index. Behavior Therapy, 17(3), 305–308.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Reliable Change Index. ScholarGate. https://scholargate.app/social-work/reliable-change-index