Fig. 4

The x-axis represents the percentage of cases from the original sample included at each step. The y-axis represents the average correlations between the original network’s centrality indices and those obtained from the networks re-estimated after excluding increasing percentages of cases. A Full samples. B Boys. C Girls