This study also calculated the small-world properties of these graphs, and found that throughout development from childhood to young adulthood, network clustering coefficient values were near those of lattices, and network path length values were near those of random graphs, indicating that the graphs were small-world networks at all ages examined. This suggested that despite the large differences in community structure across development, that child, adolescent, and adult networks were all organized in a manner that facilitated simultaneous efficiency on local and global scales. Edge density was not held constant in this analysis for small-world comparisons, but subsequent studies have replicated this finding while controlling for edge density. Supekar et al. reported small-world structure in both child and adult whole-brain graphs, but found no significant differences between them (Supekar et al., 2009). Similarly, Fransson et al. reported small-world architectures in both infant and adult voxelwise graphs, though direct comparisons between graphs were not possible (Fransson et al., 2010).
