Dear Anderson,

thanks so much for your response!

Concerning question 2: Thanks, these contrasts make sense! Just to be sure, does it mean that my EV tab would look like this in this case? We are employing the cell means model, right?

                      EV1                EV2                 EV3
                     
Input1             1                     0                       0
Input2             1                     0                       0
...
Input85           0                     1                        0
Input86           0                     1                        0
...
Input91           0                     0                        1
Input92           0                     0                        1

And why don't I select the last contrast (G2-G3) for the F-Test? Is it enough to test for these differences to be able to reject the null hypothesis because we are comparing G2-G1 = G3-G1 = 0?

Thanks again and have a great day!

Lisa

On 29 November 2016 at 09:44, Anderson M. Winkler <[log in to unmask]> wrote:

Hi Lisa,

Question 1: Your design and contrasts are correct. The two examples in the webpages are equivalent to each other and will lead to the same test statistic.

Question 2: Although your design is correct, maybe if you replace it for the other one, that is equivalent, things become more clear. Then there will be 1 EV for each group, coded as 0 or 1 depending on whether the subject belongs to the group or not. The contrasts will then be:

C1: 1 0 0 (mean of group 1)

C2: 0 1 0 (mean of group 2)

C3: 0 0 1 (mean of group 3)

C4: 1 -1 0 (diff G1-G2) -> Select this for the F-test

C5: 1 0 -1 (diff G1-G3) -> Select this for the F-test

C6: 0 1 -1 (diff G2-G3)

Question 3: Yes, although use something as "/my_path/my_results" with the option -o

All the best,

Anderson

On 28 November 2016 at 11:36, Lisa Ka <00000dcd3f2f601e-dmarc-reques[log in to unmask]> wrote:

Dear FSL Mailing List,

I hope you can help a total beginner. I am trying to run a 1x3 ANOVA (I have 3 groups of subjects and I want to check for differences in their functional connectivity for a specific ROI). The three groups are therefore the different levels of the factor "group". Each level has a different number of subjects corresponding to it (70/16/6). Now I am really confused as to how to create a GLM for it. So far, basing on the FEAT manual, I came up with this (Input1&2 representing Group1, 85 and 86 -- Group2 and 91 and 92 -- Group3):

                      EV1                EV2                 EV3
                   (mean)            (B-mn)            (C-mn)
Input1             1                     -1                     -1
Input2             1                     -1                     -1
...
Input85           1                     1                        0
Input86           1                     1                        0
...
Input91           1                     0                        1
Input92           1                     0                        1


However, I am also confused by the differences between https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Experimental_Designs_-_Between_Subject_ANOVA_Models and https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FEAT/UserGuide#ANOVA:_1-factor_4-levels (in the first example we have -1 in the rows corresponding to the reference level but in the second example we have 0 in the rows corresponding to it), so that I am not sure about it.  Question1: Is my version correct? Why is there a difference between sources?

For the contrasts and F-tests I came up with this:
            EV1           EV2           EV3
C1         0                 1                0       (F-Test selected)
C2         0                 0                1       (F-Test selected)

Question2: I don't quite understand how to obtain pure cell means (averages for all three groups) as the guide for that also differs between the manuals. How would I model that?

Question3: Would it be correct to run randomise like this:
randomise -i merged_roi.nii.gz -o /my_path/ -d designmat -t design.con -f design. fts -T   ?

Thank you so much!!