Dear SPM fellows,
We made a group comparison in correlation analysis with a random effect
analysis in SPM99. First we made a subject specific correlation model,
then defined one contrast per subject with a contrast weight 1, and
finally picked the resulting con*.img:s for a two-sample t-test.
We think that this tests the difference in regression slopes between the
two groups. Is this correct? Are there any other ways to make group
comparison in correlation analysis?
Best regards,
Esa Wallius
|