Hello! I am trying to use scanner site as a covariate in a regression model. There are multiple sites, and when I tried using a matrix, it would not let me do that. I'm guessing you can only have 1 number per subject. So is the solution to have a covariate for each site and use 0/1 coding? I.e. if subject one is site 1, subject two is site 2, and subject 3 is site 3, you would have 3 covariates that looked like this: Covariate 1 1 0 0 Covariate 2 0 1 0 Covariate 3 0 0 1 Also, generally in a regression model, you use one of the variables as a reference so do you need to do that or does SPM do that for you?