Recommendations for individual difference research using dynamic functional connectivity
| Area . | Keyword . | Recommendation . |
|---|---|---|
| Methodological | Parcellations | Test different parcellation schemes and atlases |
| Test different node resolutions to explore stability of effects (currently only for Schaefer atlas possible) | ||
| Preprocessing, denoising | Optimize preprocessing and denoising strategies to different types of functional connectivity indices | |
| Test influence of different denoising pipelines to identify possible relationship between motion and measure of interest | ||
| Exclude high motion subjects (rather strict than lenient if amount of data allows for) | ||
| Sliding window technique, dynamic functional connectivity | Test different windowing schemes (e.g., size of windows, amount of overlap) | |
| Use sufficient amount of data, if possible, e.g., via multiband fMRI (but take into account that acceleration decreases signal-to-noise ratio) or longer measurements | ||
| Multilayer modularity | Test different parameter settings | |
| Psychological | Construct of interest | Incorporate different measures for (e.g., two resilience scales) or aim for a complete characterization (i.e., all possible metrics) for the construct of interest, if possible |
| Brain: construct of interest relationship/confounding variables | Motivate in/exclusion of covariates (test both if applicable) | |
| Provide descriptive statistics for measures of interest | ||
| Provide reliability measures (if applicable) | ||
| Replications | Method section | Provide enough detail to allow for replication attempts |
| Area . | Keyword . | Recommendation . |
|---|---|---|
| Methodological | Parcellations | Test different parcellation schemes and atlases |
| Test different node resolutions to explore stability of effects (currently only for Schaefer atlas possible) | ||
| Preprocessing, denoising | Optimize preprocessing and denoising strategies to different types of functional connectivity indices | |
| Test influence of different denoising pipelines to identify possible relationship between motion and measure of interest | ||
| Exclude high motion subjects (rather strict than lenient if amount of data allows for) | ||
| Sliding window technique, dynamic functional connectivity | Test different windowing schemes (e.g., size of windows, amount of overlap) | |
| Use sufficient amount of data, if possible, e.g., via multiband fMRI (but take into account that acceleration decreases signal-to-noise ratio) or longer measurements | ||
| Multilayer modularity | Test different parameter settings | |
| Psychological | Construct of interest | Incorporate different measures for (e.g., two resilience scales) or aim for a complete characterization (i.e., all possible metrics) for the construct of interest, if possible |
| Brain: construct of interest relationship/confounding variables | Motivate in/exclusion of covariates (test both if applicable) | |
| Provide descriptive statistics for measures of interest | ||
| Provide reliability measures (if applicable) | ||
| Replications | Method section | Provide enough detail to allow for replication attempts |
Note: This brief list of recommendations does not claim to be complete, but rather advocates to always aim at incorporating the most recent findings and empirical evidence from methodological studies.