Statistical approaches used in functional magnetic resonance imaging (fMRI) to study cognitive development are varied and evolving. Two approaches have generally been used. These are between-group end-point analysis of variance (ANOVA) and age-related regression. Differences in these 2 approaches could produce different results when applied to a single data set. Event-related fMRI data from a group of typically developing participants (n = 95; age range = 7-35 years) performing controlled lexical processing tasks were analyzed using both methods. Results from the 2 approaches showed significant overlap, but also noteworthy differences. The results suggest that for regions showing age-related changes, correlation was relatively more sensitive to more linear changes whereas ANOVA was relatively more sensitive to less-linear changes. These findings suggest that full characterization of developmental dynamics will require converging methodologies.
Bibliographical noteFunding Information:
This work was supported, in part, by the Washington University Chancellor’s Fellowship to Damien Fair and by National Institutes of Health Neurological Sciences Academic Development Award (to Bradley L. Schlaggar), NS32979 (to Steven E. Petersen), NS41255 (to Steven E. Petersen), NS46424 (to Steven E. Petersen), The McDonnell Center for Higher Brain function (to Steven E. Petersen and Bradley L. Schlaggar), and The Charles A. Dana Foundation (to Bradley L. Schlaggar). We thank all the participants in this study, Mark McAvoy and Avi Snyder for neuroimaging application development, and David Van Essen and his colleagues for the use of the Computerized Anatomical Reconstruction and Editing Toolkit for figures.