TimelyScholarFree Tools › I² Heterogeneity Calculator
Calculate I², Cochran's Q, τ² and the 95% prediction interval from your studies, with a plain-language interpretation based on the Cochrane Handbook.
FreeNo account neededYour data stays in your browser
| Study | Events (treatment) | No event (treatment) | Events (control) | No event (control) | |
|---|---|---|---|---|---|
| Pooled risk ratio | 0.37 (95% CI 0.24 to 0.58) |
|---|---|
| Test for overall effect | z = -4.35, p < 0.001 |
| I² | 82.5%represents considerable heterogeneity (75-100%) |
| Cochran's Q | 28.51 (df = 5, p < 0.001) |
| τ² (between-study variance) | 0.1981τ = 0.4450 on the log scale |
| 95% prediction interval | 0.09 to 1.49Where the true effect in a new study is likely to lie |
With fewer than 10 studies, heterogeneity statistics are imprecise; interpret I² with its context.
Ratio (odds, risk or hazard) or difference (mean difference or SMD).
2×2 counts or each estimate with its 95% confidence interval.
I², Q, τ² and the prediction interval, with interpretation.
I² describes the percentage of variability in effect estimates that is due to heterogeneity rather than chance. Cochran's Q tests whether the studies share a common effect, and τ² estimates how much the true effects vary.
The Cochrane Handbook gives overlapping, rough guides: 0% to 40% might not be important, 30% to 60% may represent moderate heterogeneity, 50% to 90% substantial, and 75% to 100% considerable. The importance of I² also depends on the size and direction of the effects.
There is no universal cut-off. Cochrane's rough guide treats 0-40% as possibly unimportant and 75-100% as considerable, but always consider the size and direction of effects.
I² is a percentage of variability due to heterogeneity; τ² is the actual variance of true effects on the analysis scale. τ² does not depend on study size in the same way I² does.
It shows the range in which the true effect of a new study is likely to fall, which is often wider than the confidence interval when heterogeneity is present.