if you have coded your design matrix to contain the 4 regressors
'old correct', 'old incorrect', 'new correct', 'new incorrect', you
would assess possible positive and negative interaction effects with
these contrast weightings:
'1 -1 -1 1' and '-1 1 1 -1'. You can then take the resulting contrast
images from this 1st-level analyses up to the 2nd (group) level, for
example doing a 1-sample t-test. Main effects at the 1st level would
be assessed with '1 1 -1 -1' (and for the opposite direction: '-1-1 1
1') and '1 -1 1 -1' (and for the opposite direction: '-1 1 -1 1').
On 4/18/07, Imran Siddiqui <[log in to unmask]> wrote:
> We are trying to conduct a 2X2 factorial design at the first level of an fMRI study, but are having some difficulty. We are analyzing a memory task where the participant memorizes a list of words, and is then presented a new list where he or she must recall whether the words on that list are either new (not on the first list) or old (from the first list).
> We are currently setting up the first level analysis with two factors: word type and correctness. Each factor has two levels: old and new for word type, and correct and incorrect for correctness.
> When we run the analysis we end up with a matrix with only one contrast, the average activation. Is there something we are doing wrong that is causing us not to get the other contrasts? The SPM5 manual has an example of using a factorial design at the first level, but the directions are vague. Maybe someone can give us more case specific directions as to how to run the analysis.
Tobias Egner, Ph.D.
Cognitive Neurology and Alzheimer's Disease Center
Feinberg School of Medicine
320 East Superior, Searle 11
Chicago, IL 60611
Ph: (+1) 312 503 1749
Fax: (+1) 312 908 8789