help > RE: Contrast for Condition * Second Level Covariate Contrast
Dec 10, 2019  08:12 AM | Alfonso Nieto-Castanon - Boston University
RE: Contrast for Condition * Second Level Covariate Contrast
Hi Aaron,

If I am understanding correctly you have selected 'AllSubjects' and 'pers' in the 'subject effects' list, and entered a [0 1] between-subjects contrast, and have selected 'treatment' and 'placebo' in the 'conditions' list, and entered there a [1 -1] contrast. Is this correct? If so, then yes, that is the correct specification to evaluate potential differences in the association between connectivity and personality variables when comparing the 'treatment' vs. 'placebo' conditions (or equivalently, evaluate whether the size of the connectivity differences when comparing treatment vs. placebo conditions within each subject is associated with the personality variable).


Originally posted by Aaron Smith:
Hello all,

I was hoping someone could verify my thinking on this setup for seed to voxel analysis. I currently have two (first-level, within-subject) conditions (treatment and placebo) and have a personality (second-level) covariate. I'm currently trying to model an interaction where the association between my personality variable (pers) and connectivity is different in treatment relative to placebo. That is, the connectivity varying as a function of pers is different between treatment and placebo.

I have run the contrast [0 1 1 -1] for all subjects, pers, treatment, and placebo, respectively. Is this a correct contrast? I have tried to verify this by extracting the results of this contrast (which returned a significantly positive beta) as an ROI and doing an additional ROI to ROI analysis between the seed and result. The ROI to ROI analysis showed that pers was positively associated with connectivity in these two regions during treatment, and negative during placebo. My current interpretation is therefore a change in slope between connectivity and personality as a function of my conditions.

Any guidance is greatly appreciated.

Threaded View

Aaron Smith Dec 6, 2019
RE: Contrast for Condition * Second Level Covariate Contrast
Alfonso Nieto-Castanon Dec 10, 2019
Aaron Smith Dec 13, 2019