Hello,
I am adapting the design matrix for the repeated measures ANOVA described in the randomize manual (http://www.fmrib.ox.ac.uk/fsl/randomise/index.html) for use in an experiment with 14 subjects, three conditions, and various possible covariates.
In the example, there are two subjects and four conditions, and the matrix looks like this:
1 0 1 0 0
1 0 0 1 0
1 0 0 0 1
1 0 0 0 0
0 1 1 0 0
0 1 0 1 0
0 1 0 0 1
0 1 0 0 0
The first two columns specify the two subjects and the remaining three specify the four conditions.
My adaptation would add one column for each of the 14 subjects (3 shown), and two columns for the conditions.
1 0 0... 1 0
1 0 0... 0 1
1 0 0... 0 0
0 1 0... 1 0
0 1 0... 0 1
0 1 0... 0 0
0 0 1... 1 0
0 0 1... 0 1
0 0 1... 0 0
However, I have some outstanding questions:
1) In the example, the fourth (and in our case third) condition is modeled by zeros. In the FEAT example I had been using previously, I used:
1 1
0 -1
-1 0
to define the three conditions, which allowed me to do any sort of contrast afterwards, for example: [-2 -1] to compare the first with the second condition. Is there some reason, advantage, or limitation to using one or the other? We hypothesized that in the example given the fourth condition was the baseline and only basic comparisons with baseline were intended for the contrasts. Would it be most appropriate to model the conditions using my previous method if I want to be able to look at any contrast between conditions?
2) It has been mentioned that when using this matrix, subject-related covariates (e.g. age, gender) would lead to rank-deficient matrix. Is this the case, and if so, does that mean that only covariates for which we have three values (one per time point) can be included? Can anyone suggest an alternative model or analysis that would allow us to look at both repeated measures and subject-related covariates?
3) The order of the files you wish to process should be in the same order as the rows of the matrix; but why does it say in the randomize manual "Note the different arrangement of rows relative to the FEAT example" - was there some computational or software-related reason for using a different file order?
Insights much appreciated!
Emily
|