Fig. 6
AUC results on simulated data across different sample sizes using Pearson Correlations. Each row reports the results obtained for a given sample size (5, 20 and 50 samples in each study, respectively). For each row the performances of each method are reported for level of systematic bias Ï„ equal to 0.1, 0.5 and 1. All results are averaged over five different synthetic networks and different numbers of studies included in each collection (5, 10 and 50 studies). Standard deviations are indicated by the whiskers at the top of each plot. SD stands for Single-Dataset, while FEM, REM and FREM stand for Fixed, Random and FR-Effects method, respectively