Thank you Ben!! Will update

Correlation based methods show the highest behavioural prediction effect sizes across different behavioural domains, but also show the highest association with motion

We compare; correlation, partial-correlation, mutual information, coherence, partial information decomposition, and regression dynamic causal modelling based estimates of coupling.

New pre-print from @alexfornito.bsky.social, @jchrispang.bsky.social, and myself looking at the sensitivity of different methods used to estimate FC/EC in relation to residual motion artifacts and behavioural prediction
www.biorxiv.org/content/10.1...

At the hierachical level, both a unidimensional and bi-factor structure fail to meet criteria for construct validity, while at the item level, subscales poor reliability along the latent trait continuum. We caution against uncritical use of the CBCL for dimensional measurement in this sample.

New pre-print from @alexfornito.bsky.social , Toby Constable, Jeggan Tiego, and myself, looking at the psychometric properties of the CBCL in the baseline ABCD cohort. We show that the commonly used CBCL structures fail to meet standards for validity and reliability
osf.io/preprints/ps...

Definitely would be, it would be interesting to see the differences between DiCER, temporal ICA and GSR in the brain-behaviour prediction space... Maybe in the future

New pre-print from @alexfornito.bsky.social, James Pang, and myself, looking at the effects of rs-fMRI pre-processing pipelines on FC quality and behavioural predictions.
www.biorxiv.org/content/10.1...