Effect sizes and clinical significance
What the numbers mean, and how they are made to mean more than they do.
An effect size expresses a difference in standard deviation units, which makes results comparable across measures and incomparable across designs. The number alone tells you almost nothing without the comparator, the population and the outcome.
Reliable change and clinical significance
Jacobson and Truax’s two criteria remain the standard. Reliable change: the difference exceeds what measurement error would produce, calculated from the instrument’s reliability. Clinical significance: the person has moved from the clinical population’s range into the functional population’s range. A patient can show reliable change and remain clinically depressed; the two criteria answer different questions.
| Term | Question it answers |
|---|---|
| Statistical significance | Is the difference unlikely to be chance? |
| Effect size | How big is the difference, in standardised units? |
| Reliable change | Is the change larger than measurement error, for this individual? |
| Clinical significance | Has this person crossed into the functional range? |
| Number needed to treat | How many people must be treated for one additional good outcome? |
Sources
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
article