Thanks, Stephen! (My first question solved itself - it wrote the zstats for pill, cycle and their interaction under the four cope directories - angry face etc. - from my first-level analysis.) Klara On Wed, Apr 21, 2010 at 4:56 AM, Stephen Smith <[log in to unmask]> wrote: > Hi > > On 19 Apr 2010, at 20:13, Klara Mareckova wrote: > > Hi, > > could anybody please help me setting up my design for a voxelwise analysis > of 2by2 ANOVA? I'm trying to follow the directions from > http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html, but I'm not really sure > if I'm setting up the EVs correctly. > > My study has 40 subjects divided into 4 groups: on pill & menstruation, on > pill & ovulation, free & menstruation, free & ovulation. > In the first level analysis, I had 5 conditions (EVs) where participants > were were watching angry faces, angry hands, neutral faces, neutral hands or > control condition. In the higher level analysis, I would like to explore the > main effect of pill, main effect of cycle phase and their interaction. > > Setting up the contrasts & F test seems pretty easy then: > pill 100 0 on, off, off > cycle 010 0 off, on, off > pill * cycle 001 0 off, off, on > > What is more of a puzzle for me are the EVs: > 1. Am I not working with my 5 conditions from 1st level analysis (angry, > neutral,..) at all? > > > I don't quite understand - presumably at first level you will make a > contrast of interest (e.g. angry vs neutral) and then feed that up for all > sessions/subjects into the higher level modelling? You could then repeat > that for each first-level contrast of interest? > > 2. If so, would the EV set up look like this? > > group cycle group interaction grandmean > 1 1 1 1 1 > 1 1 1 1 1 > 1 1 1 1 1 > 1 1 1 1 1 > 1 1 1 1 1 > 1 -1 1 -1 1 > 1 -1 1 -1 1 > 1 -1 1 -1 1 > 1 -1 1 -1 1 > 1 -1 1 -1 1 > 2 1 -1 -1 1 > 2 1 -1 -1 1 > 2 1 -1 -1 1 > 2 1 -1 -1 1 > 2 1 -1 -1 1 > 2 -1 -1 1 1 > 2 -1 -1 1 1 > 2 -1 -1 1 1 > 2 -1 -1 1 1 > 2 -1 -1 1 1 > > > Yes > > 3. With the set up above, I'm getting a warning: "design matrix uses > different groups but these do not separable EVs for the different groups" > > > yes - this design is not "separable" so just set all the "group > memberships" to 1. > > Cheers. > > > > > Do you happen to have any suggestion what am I supposed to change to get > the right design for my study? > > Thanks a lot, > > Klara > > > > > > > --------------------------------------------------------------------------- > Stephen M. Smith, Professor of Biomedical Engineering > Associate Director, Oxford University FMRIB Centre > > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK > +44 (0) 1865 222726 (fax 222717) > [log in to unmask] http://www.fmrib.ox.ac.uk/~steve > --------------------------------------------------------------------------- > > > >