Random-Effects Analysis

Neuroimaging data from multiple subjects can be analysed using fixedeffects (FFX) or random-effects (RFX) analysis. FFX analysis is used for reporting case studies and RFX is used to make inferences about the population from which subjects are drawn. We demonstrate both methods on data from a multiple subject PET study and describe the mathematical basis of the two approaches. Finally, in neuroimaging, RFX is implemented using the computationally efficient ‘summary-statistic’ approach. We show that this is mathematically equivalent to the more computationally demanding maximum likelihood procedure.