I have a design with one group variable (patient vs. controls) and one
continuous variable (a personality measure). I am using SPM2 and have set up
the second-level analysis using the Multiple Regression option with two
regressors, specifying the first column as the patient personality scores
and entering zeros for the rows pertaining to the control subjects and the
second column as the control personality scores and entering zeros for the
rows pertaining to the patients. Thus, I have a design matrix with two
columns, one column with patient personality information and the other with
control subject personality information. Then, to test the group (patient >
control) X personality variable interaction, should I use a [1 -1] t
contrast or will this give me areas of activation that result from the
positive correlation with personality variable scores for patients vs. the
negative correlation with personality scores for controls? Does anyone have
a better suggestion for how to set up this contrast to effectively test the
group (categorical variable) X personality measure (continuous variable)
interaction? I ultimately want to know where brain activation is associated
with increasing personality variable more in patients than controls. Thanks.
|