Dear fsler,
I have two questions about my 2x2 ANOVA interaction. The 2x2 design included two main effects: diagnosis group (patient and control ) and age (young and old). For example, there are 2 subjects in each subgroup which makes 8 subjects in total. The design matrix of ANOVA looks like following
age diagnosis interaction mean
1 1 1 1
1 1 1 1
1 -1 -1 1
1 -1 -1 1
-1 1 -1 1
-1 1 -1 1
-1 -1 1 1
-1 -1 1 1
contrast:
F1 F2 F3
C1 1 0 0 0 X
C2 0 1 0 0 X
C3 0 0 1 0 X
C4 -1 0 0 0
C5 0 -1 0 0
C6 0 0 -1 0
The significant brain activities was found on all three F contrasts which means significant brain activity between age, diagnosis and age by diagnosis interaction. Since the interaction between two main effect showed significant in my model, the post hoc statistic will be explored by examining one main effect according to each level of another main effect in normal statistic analysis. My first question is how could I achieve this in fsl? Do I need to run four separate One way ANOVA in fsl?
My second question is: when I checked the t-test results: C1-C6, there are only C4 showed significant brain activities. According to Jeanette Mumford, C3 or C6 should be significant as well which is not true in my case. Dose anyone know how to explain the result?
bw,
Joyce
