Hi SPMers,
We are running a multiple regression analysis with one covariate of interest and a nuisance variable. We first included the covariate of interest under matlabbatch{1}.spm.stats.factorial_design.des.mreg.mcov and the nuisance variable under matlabbatch{1}.spm.stats.factorial_design.cov. This gives a design matrix with the nuisance variable in the first column. If we run the analysis with variables specified as matlabbatch{1}.spm.stats.factorial_design.des.mreg.mcov(1) and mcov(2), the design matrix has the variables in the first and second columns as we expect. The results are the same for a contrast of 1 for our variable of interest regardless of the order of the regressors . Is SPM orthogonalizing the regressors in any way? If it does, how? is there any difference in how spm treats the covariates under the matlabbatch{1}.spm.stats.factorial_design.des.mreg.mcov or the matlabbatch{1}.spm.stats.factorial_design.cov? We have mean centered both variables.
|