 |
 |
Ben, See inline responses below. On Tue, Dec 27, 2011 at 5:10 AM, Ben Becker <[log in to unmask]> wrote: > Dear SPMers, > > I`m using a 2-sample t-test to comapare fmri data from 2 groups. However, > the groups differ in mean age. To control for this confounding factor I > want to include age as a covariate in the spm5 two-sample t-test. > 1. Is it correct to specify one covariate (age) > It depends on what you want to do. See http://mumford.fmripower.org/mean_centering/ for a full description of the issues. One covariate means that both groups will have the same slope, which can help account for the group difference in age. However, the age difference will also weaken the group effects because the age covariate will be collinear with the group terms. Also, you could have no relationship with age and your measurement and incorrectly attribute group differences to the age difference. It's a really tough problem to deal with and there is no ideal solution. > 2. How would the corresponding t-contrast (controlling for age) look like? > Is 1 -1 (or -1 1) correct? > Yes. These contrasts would test for group1>group2 and group2>group1, respectively. > > Thanks in advance & best regards > > ben >