Which statistic is commonly used to quantify the proportion of total variation across studies due to heterogeneity?

• A. Cochran's Q statistic
• B. I-squared (I^2)
• C. Tau-squared
• D. Pooled effect size

Answer: (B)

In meta-analysis, observed differences in study results come from two sources: sampling error within studies and real differences in effect sizes between studies, known as heterogeneity. The statistic that expresses what portion of the total variation is due to that heterogeneity is I-squared (I²). It is reported as a percentage from 0 to 100 and is derived from Cochran’s Q statistic, indicating the proportion of observed variance that cannot be attributed to chance alone. Interpreting I² helps you gauge consistency across studies: values near 0% suggest little heterogeneity, while higher values indicate more substantial differences among study effects.

Note how this differs from related metrics: Cochran’s Q tests whether heterogeneity is present but doesn’t quantify its magnitude as a proportion. Tau-squared estimates the between-study variance in the same units as the effect size, giving an absolute amount of dispersion rather than a proportion. Pooled effect size is the overall average effect across studies, not a measure of heterogeneity.