Hello List,
I want to run an F-test on a ROI (I selected the voxels with some criterion). If I understand correctly the way to do it is by computing the F-statistic using:
F=(roi_betas'*X'*M*X*roi_betas)/(Y'*R*Y)*(J-p)/p1
where: roi_betas are the average beta values across the roi X is the design matrix M is the projection matrix to the space generated by the reduced model given by my contrast matrix Y is the average time course of the ROI R is the residual generating matrix. J-p are the total degrees of freedom, and p1 are the degrees of freedom given by my contrast matrix
I am noticing that the residuals are not whitened (the is noticeable auto-correlation in the ROI-averaged residuals that lasts ~10 timepoints).
Could someone recommend a relatively simple way to correct for this auto-correlation (and avoid reinventing the weel in matlab)?
Thanks in advance! Pablo