 |
 |
Hi On 6 Oct 2010, at 15:38, Rodrigo Perea wrote: > Hello, > I am new in this field and I am very skeptical with the different matrix designs I've tried when I wanted to analyze a linear relationship case. I want to check the correlation between FA and age in one time point in individuals of the same group with sex as a covariate. > > For example, lets suppose we have 10 individuals, my guesses were: > > > 1. I demeaned the age and set up two contrasts. > > Design.mat: In the first column it shows that all belong to Group1 Not quite - in the actual design matrix the first EV as shown below is modelling the overall group mean. > , 2nd colum is the demeaned age and the 3rd column my covaraite sex. You would normally demean the 3rd EV as well - but in this case it won't matter. > > /NumWaves 3 > /NumPoints 47 > /PPheights 1.000000e+00 2.558000e+01 1.000000e+00 > > /Matrix > 1 -2.4 1 > 1 -2.8 1 > 1 5.7 0 > 1 -4.8 1 > 1 3.4 0 > > Desing.con: Since the 2nd EV is the one I want either a positive or negative relation, I ended doing this... > > /Matrix > 0 1 0 > 0 -1 0 Correct > > > Or my second guess > 2. I didnt set up the group values since all will be assigned to the same group. Thus, 1st column is the demeaned age and 2nd the covariate sex. No - the first example is correct - note that the part of the Glm GUI where you enter group membership may be used in the modelling, but is NOT part of the design matrix. Cheers. > > Design.mat: > /Matrix > -2.4 1 > -2.8 1 > 5.7 0 > -4.8 1 > 3.4 0 > > Design.con > > /Matrix > 1 0 > -1 0 > > Thanks in advance! > > > > Rodrigo Dennis Perea > Neuro-Imaging Research Assistant > [log in to unmask] > Alzheimer's & Aging Program > The University of Kansas - School of Medicine > --------------------------------------------------------------------------- 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 ---------------------------------------------------------------------------