David, If you are using a F contrast you also need to include the design.fsf to tell randomise about the F contrast. Let us know if it still isn't flying. -Tom On Fri, Feb 29, 2008 at 12:27 AM, David Kideckel <[log in to unmask]> wrote: > Hi. Thanks for the reply. I have read those posts and now I am unclear as > to which files I should be including in the randomise TFCE > option. I have included only 1 F contrast with all 3 groups (I assume this > is what I want based on Tom's post). I have included the design.mat and > design.con files as well. Is this correct, or should I be including > different files? > Thanks. > David > > On 2/28/08, Gwenaëlle DOUAUD <[log in to unmask]> wrote: > > > > Hi David, > > > > yes the number of rows should equal the number of > > subjects you have in your 3 groups. For the F test > > using randomise, please look at Tom emails from the > > 22nd of Februrary on this exact subject. > > > > Hope this helps, > > Gwenaëlle > > > > > > --- David Kideckel <[log in to unmask]> a écrit > > : > > > > > HI Tim, > > > > > > One more thing. Will the number of inputs be equal > > > to the number of subjects > > > I have? > > > David > > > > > > On 2/28/08, David Kideckel > > > <[log in to unmask]> wrote: > > > > > > > > Hi Tim. I am trying to set up the design matrix > > > and am getting errors. I > > > > am trying to compare 3 groups (as described > > > above), and this is the info I > > > > have included in the design matrix: > > > > 3 EV'S with 6 > > > > inputs: the first 2 inputs are 1 0 0, the second > > > two are 0 1 0, and the final 2 are 0 0 1 (as per > > > your email above). > > > > > > > > The contrasts include 3 contrsats and 1 F-test. > > > The first contrast is 1 -1 0, the second contrast is > > > 1 0 -1, and the third is 0 1 -1 (as per your email > > > above). All 3 C's are checked with yellow in the > > > Glm_gui, but the F's are not. > > > > Can you please tell me why it isn't working. The > > > error message when I > > > > click View Design > > > > is: "problem with processing the model: F-test 1 > > > isn't valid - each included contrast cannot be a > > > linear combination of the others. child process > > > exited abnormally" > > > > > > > > I look forward to hearing from you. > > > > > > > > David > > > > > > > > Right. Can I set up each TBSS script (ie TBSS1... > > > > > > > > On Jan 30, 2008 12:39 PM, Tim Behrens > > > <[log in to unmask]> wrote: > > > > > > > > > Hi > > > > > On 30 Jan 2008, at 17:32, David Kideckel wrote: > > > > > > > > > > Thank you for the response. How exactly do I do > > > this. Do I have to run > > > > > all TBSS preprocessing steps before this though, > > > and place all subjects' FA > > > > > maps into the original file folder? > > > > > > > > > > > > > > > Yes > > > > > > > > > > > > > > > Once I get to the design matrix, can I set up a > > > script that will allow > > > > > me to do this 3 group comparison with a cluster > > > threshold of 3? > > > > > Cheers. > > > > > > > > > > > > > > > What question do you want to ask of your three > > > groups? > > > > > The simplest one you can ask is where are there > > > any differences between > > > > > any of the groups. > > > > > For this, you design matrix looks like > > > > > 1 0 0 > > > > > 1 0 0 > > > > > ... > > > > > 0 1 0 > > > > > 0 1 0 > > > > > ... > > > > > 0 0 1 > > > > > 0 0 1 > > > > > > > > > > your contrast looks like > > > > > > > > > > 1 -1 0 > > > > > 1 0 -1 > > > > > 0 1 -1 > > > > > > > > > > and then you compute an F-statistic across all > > > three contrasts. > > > > > > > > > > You can then use randomise with whatever > > > F-threshold you want using -F > > > > > > > > > > Cheers > > > > > > > > > > T > > > > > > > > > > > > > > > > > > > > David > > > > > > > > > > On Jan 30, 2008 12:26 PM, Tim Behrens > > > <[log in to unmask]> wrote: > > > > > > > > > > > You can set up a GLM with an EV for each > > > group, and test any > > > > > > comparison you like with the contrast matrix - > > > all in one TBSS run. > > > > > > > > > > > > > > > > > > T > > > > > > On 30 Jan 2008, at 17:21, David Kideckel > > > wrote: > > > > > > > > > > > > > Dear FSLers, > > > > > > > > > > > > > > Can one compare 3 groups in TBSS all at > > > once, or is it a > > > > > > > requirement of TBSS to make three separate > > > comparisons, essentially > > > > > > > running three separate TBSS analyses to > > > obtain all possible > > > > > > > comparisons, i.e., CON vs EXP1, CON vs EXP2, > > > and EXP1 vs EXP2. > > > > > > > > > > > > > > Thank you. > > > > > > > > > > > > > > David Kideckel > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > _____________________________________________________________________________ > > Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! > > Mail http://mail.yahoo.fr > > > > -- ____________________________________________ Thomas Nichols, PhD Director, Modelling & Genetics GlaxoSmithKline Clinical Imaging Centre Senior Research Fellow Oxford University FMRIB Centre