Dear SPMers,
I have data from 16 participants in a repeated measures design (three sessions: placebo, drug A, drug B). On each session they listened to music and scrambled sound in a blocked fMRI design, and so my first-level contrast is music>scrambled.
At the second level my main aim is to compare across the three sessions. I have done this using a one-way within-subjects ANOVA and t contrasts across each of the conditions. However I would also like to look at the overall effect of the task (i.e. music>scrambled) across each of the three sessions. I have been advised to use a one-sample t-test on all 48 scans, but this seems intuitively wrong to me as it doesn't account for between/within-subjects variance.
Could anyone advise me whether a one-sample t-test is suitable for my question? If not, does anyone know of suitable contrast codes for a within-subject ANOVA or an alternative model?
Thank you,
Tom
|