System | Ï„2 | I2 | H2 | R2 | AIC | AICc |
|---|
White-3 | 1.3 | 97.1 | 34.0 | 33.6% | 707.7 | 707.9 |
Lover | 1.6 | 97.5 | 40.8 | 19.4% | 728.4 | 729.0 |
White-5 | 0.9 | 95.4 | 21.8 | 57.3% | 623.6 | 624.0 |
Macdonald | 0.8 | 95.0 | 20.0 | 59.9% | 612.7 | 614.1 |
Modified Macdonald | 0.8 | 95.0 | 19.9 | 60.1% | 612.9 | 614.0 |
- Mixed-effects meta-analysis has been performed for 214 different studies or study arms by using the R package metafor [68]. The statistics are: Ï„2, amount of residual heterogeneity; I2, residual heterogeneity/unaccounted variability; H2, unaccounted variability/sampling variability; R2, calculated from the residual heterogeneity (Ï„2) and the residual heterogeneity of an empty model as suggested by Raudenbush [70]; AIC, Akaike information criterion; and AICc, corrected AIC. The values of AIC, BIC and AICc are based on the restricted maximum likelihood, as this corresponds to the REML (restricted maximum likelihood) estimator recommended by experts [67].
