A comparison of permutation and parametric testing for between group effective connectivity differences using DCM

Effective connectivity is becoming an increasingly popular technique for obtaining additional information from functional magnetic resonance imaging (fMRI) cognitive activation studies. It is potentially important for investigating psychiatric illnesses, which are thought to depend on disrupted connections and in observing the action of psychoactive drugs used to treat these disorders. If researchers are to apply these techniques confidently then it is important to establish the level of power that is available in an experiment. This study compares the level of power available when applying effective connectivity to test for differences between groups using parametric tests and permutation testing. Permutation testing has previously been shown to have superior sensitivity to parametric tests in fMRI studies. As an illustrative example, both the parametric t-test and equivalent permutation test were applied to a comparison between healthy controls and remitted depressed volunteers performing an emotional face processing task. Permutation testing was found to provide superior power compared with the nonparametric equivalent.