The general linear model (GLM) is the statistical method of choice used in brain morphometric analyses because of its ability to incorporate a multitude of effects. This chapter starts by presenting the theory, focusing on modeling, and then goes on discussing multiple comparisons issues specific to voxel-based approaches. The end of the chapter discusses practicalities: variable selection and covariates of no interest. Researchers have often a multitude of demographic and behavioral measures they wish to use, and methods to select such variables are presented. We end with a note of caution as the GLM can only reveal covariations between the brain and behavior, and prediction and causation mandate specific designs and analyses.
|Title of host publication||Brain Morphometry: Methods and Clinical Applications|
|Publication status||Published - 28 Feb 2018|