Hi there

I have data from 83 participants comprising 4 different groups. Each participant completed a working memory task with two memory load conditions.

Is it possible to run a 4 between group by 2 condition ANOVA design with NBS, in order to examine the main effect of group, condition, and the interaction between the two?

If so, how can the design matrix be set up? Having a look through the FSL wiki for examples and the help sections of NBS, I've come up with a few possibilities below, but none of them seem correct to me:

1) have different groups listed in columns 1-4, then the condition denoted by 1 and -1 in column 5. I think this method could test between groups (with contrasts [1 1 1 1 0], and between conditions (with contrasts [0 0 0 0 1]), but  not the interaction:

group 1 group 2 group 3 group 4 condition

1 1     
1 -1
1 -1
     1 1
     1 -1
    1 1
    1 -1
   1 1
   1 -1


2) A similar design matrix to the FSL wiki two group two condition ANOVA design attached. However, this method codes groups as 1 and -1, and doesn't seem to cater for more than two groups.

Kind regards,

Neil

Hi Neil,

I suggest that you try to simplify your design - I think I mentioned this before. If this is not possible, I have attached an example of a 3x2 design matrix from Miller and Haden. You should be able to extrapolate this to 4x2, but once again I suggest that you simplify your design as much as possible. Without simplification, there are many kinds of interactions that you can test, and it is also difficult to model the repeated measures aspect with a complex design.

Andrew