Assessing heterogeneity

Once the overall effect size is computed, the next question is how much the individual studies actually agree with it. Heterogeneity across study effect sizes can be assessed through two statistics:

Q statistic: used for establishing if there is significant heterogeneity across studies.
I²: used to quantify heterogeneity; it estimates the proportion of observed variance that reflects real differences in effect sizes. Being a proportion, it does not tell you how much the true effects vary in absolute terms, so the familiar labels for ~25%, ~50%, and ~75% should not be read as a classification (Borenstein et al., 2017).

The underlying between-study variance itself is denoted τ² (tau-squared), an absolute measure; I² expresses it relative to the total observed variance. A prediction interval shows the spread of the true effects on the scale of the effect size.

See Methods and formulas for the Q and I² formulas, and Heterogeneity in meta-analysis for how to interpret Q, τ², I², and the prediction interval.

The result of this step is a judgment on how much the true effects vary and whether that matters for interpreting the pooled effect — and, if it's non-negligible, a reason to look for explanatory moderators, covered next in Testing moderators.

References

Article Borenstein, M., Higgins, J. P. T., Hedges, L. V., & Rothstein, H. R. (2017). Basics of meta-analysis: I2 is not an absolute measure of heterogeneity. Research Synthesis Methods, 8(1), 5–18. https://doi.org/10.1002/jrsm.1230