 |
 |
Dear Donald, Thank you for the prompt reply. Let me specify that my 2X2 anova has a between-subjects factor (Group) and a whitin-subjects factor (Condition). In Group 1 I have 17 people, whereas in Group 2 I have 14 people. Each subject saw 2 different categories of visual stimuli. Basically, I should use a flexible-factorial design where I can specify the 'subjects' (any advice on where can I look for a useful tutorial on this?). It is still unclear to me whether I should use the F-contrast or the t-contrast for the designs you indicated. Thanks again and apologies for these basic questions, Davide From: MCLAREN, Donald [mailto:[log in to unmask]] Sent: Wednesday, 7 March 2012 12:05 AM To: Rivolta, Davide Cc: [log in to unmask] Subject: Re: [SPM] 2x2 ANOVA MEG On Tue, Mar 6, 2012 at 10:51 AM, Rivolta, Davide <[log in to unmask]> wrote: Dear all, I am performing a 2X2 anova (Group: controls & patients; Category: 1 & 2) on MEG data (visual task). As a beginner with SPM I wish to receive a feedback about my design matrix. I chose a full factorial design: Controls1, Controls2, Patients1, Patients2. I calculated the: a) Main effect of Group: [1 1 0 0; 0 0 1 1] The MEG should be [1 -1 0 0; 0 1 -1 0; 0 0 1 -1] If Controls1 are the same as Controls2 (meaning the same subject has 2 observations, then the MEG would be: [1 1 -1 -1]. However, in the case of 2 observations per subject, the contrast is not valid. Also if the you have 2 observations per category, then you will need a flexible factorial design with a subject term in the model. b) Main effect of Category: [1 0 1 0; 0 1 0 1] The MEC is [1 -1 1 -1]. This will be the case if all observations are from different subjects or if they are repeated. If they are repeated, then you need a subject term. c) Group x Category interaction: [1 -1 0 0 ; 0 0 1 -1] The interaction would be [1 -1 -1 1]. This will be the case if all observations are from different subjects or if they are repeated. If they are repeated, then you need a subject term. Is it correct? See the corrections above. The contrast you had specified were testing if any group or category were different than 0 and whether the category had an effect in either group. You were not testing if the groups or categories were different. I then started with the t-tests for the post-hoc comparisons. Any advice would be great, Thanks, Davide