lost, as the calculation implies that all edges affiliated with the same network contribute to network connectivity equally. Therefore, we also used the connectome-based predictive modeling (Shen et al., 2017), which is an edge-level analysis that can address the issues of network-level analysis. That is, CPM extends the analysis to non-canonical networks and only focuses on important edges that significantly associated with binge drinking. Lastly, to better understand the behavioral correlates of connectivity variables, correlations between change in cognitive performance and change in connectivity were calculated for connectivity variables that were significantly associated with standard/extreme bingeing.
