Overview
- Group-level analysis refers to analyses that seek to model associations between nonimaging individual differences with measures of the brain derived from MRI (or another imaging modality)
- For univariate data, this often would correspond to a mixed model or other multi-level analysis model
- Because of computational demand, in Neuroimaging these are fit in multiple stages
Section outline
- Modeling and estimation
- Statistical inference
Why are neuroimaging data multilevel?
- Introduces two challenges
- Correlation among measurements
- Heteroskedasticity (variance differs between participants)
Example: Group differences in
Multiple measurements on the same participant
