DataAnalysisI.BasicAnalysesandtheGeneralLinearModel.ppt

FMRI Data Analysis:I. Basic Analyses and the General Linear Model;When do we not need statistical analysis?;Why use statistical analyses?;What do our analyses generate?;What are our statistics for?;;Key Concepts;Simple Hypothesis-Driven Analyses;Correlation Approaches (old-school);Fourier Analysis;12s on, 12s off;;The General Linear Model (GLM);Basic Concepts of the GLM;Signal, noise, and the General Linear Model;Form of the GLM;Design Matrices;Regressors (How much of the variance in the data does each explain?);The optimal relation between regressors depends on our research question;Suppose that we have two correlated regressors.;Let’s now make the regressors anti-correlated .;Value of R1 (at each point in time);And, we can make the regressors uncorrelated with each other through randomization.;Fixed and Random Effects Comparisons;Fixed Effects;Random Effects;Key Concepts of Random Effects;The Problem of Multiple Comparisons;A;Options for Multiple Comparisons;Statistical Corrections;;Bonferroni Correction;Gaussian Field Theory;;Cluster Analyses;How many foci of activation?;How large should clusters be?;False Discovery Rate;Genovese, et al., 2002;ROI Comparisons;;Voxel and ROI analyses are similar, in concept;Summary of Multiple Comparison Correction;Displaying Data;Never Mask!;;;;Summary of Basic Analysis Methods;Midterm Results