Sorry, I forgot the attachments first. Dear SPM-experts I am doing structural VBM-analyses in patients with disease A or B and want to evaluate whether brain volumes are different in patients with disease factor X. I use the "VBM and PET" in SPM8/matlab 2013a. First, I made a two-sample t-test (with patients with disease A as group one and disease B as group two and with disease factor X as a covariate, in addition to age and gender. I added the t-contrast -X (0 0 -X 0 0 cf. attached image 1). Secondly, I made a Full factorial with Group (disease A and disease B as level) as first factor and disease factor X as a second factor and age and gender as covariates. I added the t-contrast -X (1 -1 1 -1 0 0 cf. attached image 2). The results from these two analyses was almost identical. Then I wanted to expand the model to also include disease factor Y. Similar to what I did above, I made a two-sample t-test with disease A and disease B as groups and disease factor Y, X and age and gender as covariates. I added the t-contrast -X (0 0 0 -1 0 0 cf. attached image 3). Finally, I made a full factorial with disease, factor Y and factor X as factors and age and gender as covariates and added the t-contrast -X (1 1 -1 -1 1 1 -1 -1 0 0 cf. attatched image 4). Here, I found larger areas with lower volume in the two-sample t-test compared to the full factorial model. I would really appreciate if anyone could tell me the statistical difference between these two models. Is the result from the full factorial analyses more reliable (I have found no interaction between the factors tested by the F-contrast 1 -1 -1 1 1 -1 -1 1 0 0). Best, Maria Boge Lauvsnes