Dear Sally,

Again, thanks very much for your answer. While I stilll hope to receive a reply from Doug, maybe I could add another question concerning a different fMRI study of ours in which, again, an SPSS-like Repeated Measures Design would be required for optimal statistical analysis of the data.

We have scanned 30 patients on two separate occasions (Measurement 1 (Baseline; 30 patients) and Measurement 2 (medication (15 patients) or placebo (15 patients)) and we would like to extract the medication-effect (corrected for baseline brain activation) on brain activation patterns during task-performance. Again, ANOVA does not take into account that some measurements were repeated in a single subject (i.e.: are paired).

The problem is therefore that we have two different groups in one study: one placebo, one medication, which all undergo the same procedure. A Repeated Measures Design would allow us to 'disentangle' these two, by first specifying a within-subject factor 'Treatment' with two levels (1 and 2), yielding an effect of treatment with either medication or placebo.

Then, a between-subject factor 'Group' with two levels (Medication and Placebo) can be specified, by which the study-population is separated into a group receiving placebo and one receiving medication. This will extract the effect that 'being a member' of one of those groups has on brain aktivation, and hence the medication effect. Additionally, it will account for possible pre-existent differences in brain activation at baseline between these groups that are due to small sample size and may mask or enhance medication effects.

Therefore in short, would current AFNI-code allow such a design? If so, add on every happy person to this world.

Rutger Goekoop