Borenstein, M., Higgins, J., Hedges, L., & Rothstein, H. (2017) argue that the $ I^2 $ statistic in meta-analysis is not an absolute measure of heterogeneity. They explain that $ I^2 $ represents the proportion of variance in observed effects due to true heterogeneity rather than sampling error. However, it does not provide information about the actual range of true effects. Instead, they advocate for reporting prediction intervals, which give the absolute range of true effects across populations. The authors emphasize that $ I^2 $ is often misinterpreted as an absolute measure of heterogeneity, but it is actually a relative measure. They provide examples showing that $ I^2 $ values do not directly translate to the range of true effects. The paper highlights the importance of using prediction intervals to convey the true variability of effects in a meta-analysis, as they provide a more accurate and interpretable measure of heterogeneity. The authors conclude that while $ I^2 $ has a role in meta-analysis, it should not be used as a substitute for prediction intervals when reporting the range of true effects.Borenstein, M., Higgins, J., Hedges, L., & Rothstein, H. (2017) argue that the $ I^2 $ statistic in meta-analysis is not an absolute measure of heterogeneity. They explain that $ I^2 $ represents the proportion of variance in observed effects due to true heterogeneity rather than sampling error. However, it does not provide information about the actual range of true effects. Instead, they advocate for reporting prediction intervals, which give the absolute range of true effects across populations. The authors emphasize that $ I^2 $ is often misinterpreted as an absolute measure of heterogeneity, but it is actually a relative measure. They provide examples showing that $ I^2 $ values do not directly translate to the range of true effects. The paper highlights the importance of using prediction intervals to convey the true variability of effects in a meta-analysis, as they provide a more accurate and interpretable measure of heterogeneity. The authors conclude that while $ I^2 $ has a role in meta-analysis, it should not be used as a substitute for prediction intervals when reporting the range of true effects.