Fig. 2. Replication analyses in the Dunedin Study of the original ICA-derived cerebellar morphology associations with p factor scores from Moberget et al. [19].

a The nine independent components resulting from data-driven decomposition of cerebellar gray matter maps projected onto flat-maps of the cerebellar cortex [36]. b Distributions of correlations between predicted and actual p factor scores across 10,000 iterations of the tenfold cross-validated model using the average of the nine independent components from a compared with the empirical null distribution. The black dotted lines represent the mean for each distribution and the gray dotted line represents the one-tailed 0.05 threshold. The nine ICA-derived components predicted p factor scores beyond chance on average, but the difference from the empirical null distribution was p = 0.53, suggesting nonsignificant replication of Moberget et al. [19].