Yes, you should account for scanner differences. You can do a factorial design for this if you'd like. Then your contrast would be [1 1 0] or whatever is appropriate for what you're testing.
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Patrick Floyd [[log in to unmask]]
Sent: Tuesday, November 30, 2010 8:38 AM
To: [log in to unmask]
Subject: [SPM] one group, two magnets
Hello SPMers,
I have a fMRI dataset of healthy subjects collected on two 3T magnets. Id like to combine these data into a single group, controlling for any between-magnet effects (which according to recent literture should be minimal compared to within subject variability).
Using SPM2, Ive run a multiple regression with constant, using a single binary covariate denoting magnet, and run the contrast [0 1] - ie weighing the mean column and treating the magnet regressor as nuisance. The results do differ from the one-sample t-test across both magnets suggesting some adjustment has been made to the group mean in the multiple regression model (though it could be due to the differences in df?). Was this the correct model/method to use in this instance, or is there a more appropriate procedure to control for magnet (or similar nuisance variable) effects within group?
Any feedback would be much appreciated.
Pat Floyd
|