 |
 |
On Thu, Sep 5, 2013 at 2:46 PM, fMRI <[log in to unmask]> wrote: > Hello all, > > I have not used one way anova before but I believe it is used for more > than two groups. > A a one-way between group ANOVA would be for 2 or more groups. A repeated measures ANOVA would be for 2 or more conditions. A mixed 2-way ANOVA would be for 2 or more conditions and 2 or more groups. > > My experiment involves five conditions, A, B, C, D, and E. > You want a repeated measures ANOVA. > > My aim is to see the relation between the five conditions and whether this > increases the activations. > > I have specified each condition in a separate regerssor for each subject. > Then I contrast each condition versus rest, for example 1 0 0 0 0 0. So I > created five contrasts in the first level. > This is correct > > Then, I took the contrast images of the first level from each condition > and each subject and entered them into a one way anova. So I had five cells > representing the five conditions. Here, I added an F tests like this: > 1 > 0 1 > 0 0 1 > 0 0 0 1 > 0 0 0 0 1 > Your ANOVA is missing a subject term and the contrast is invalid. When you have a repeated-measures ANOVA (in SPM/FSL/etc.), you must compare the levels against each other. You cannot compare a level against 0. The reason for this is that the error term for the model is the within-subject error term and comparing an individual condition (level) against 0 is a between-subject effect and requires the between-subject error term, which is not available in SPM/FSL GLMs. > > This allows me to investigate the contrast estimate for all of the five > conditions. > No. It only allows comparisons of conditions and only if you add a subject factor. > > > Questions: > > 1)is this correct? > > See comments above for the changes needed. > 2) if I would like to investigate a linear or non linear relations between > the five conditions, for example acceleration rate - very slow, slow, mid, > fast, very fast, and the contrast estimate at a voxel, can test the > correlation outside spm and report it OR is there anyway of adding these > five condition into spm, for example representing them by 1 2 3 4 5, and > then see the relation. > Linear(T-test): -2 -1 0 1 2 Non-linear: custom defined change over level Non-linear (F-test): 1 -1 0 0 0; 0 1 -1 0 0; 0 0 1 -1 0; 0 0 0 1 -1 > > > > I hope I have explained it clearly:-). > > Thank you in advance, > , > > Aser