Ok. I will try to make this as succinct and clear as possible. I have been going at this analysis for a couple of weeks now without any success.
I have 15 subjects that participated in this study. We employed fMRI at time 1 and time 2 after a mental training regimen.
At time 1, there were two conditions (2 volumes). One is rest and the other is a cognitive task. At the first level, I compared these two conditions.
At the second level (higher level analysis), I calculated the group average (n=15) between these conditions. In the same analysis, I also assessed the brain areas associated with anxiety reductions across these conditions. That is, I correlated the demeaned percent change in anxiety associated with the cognitive task and rest. So, there were a total of four contrasts.
Then subjects participated in a mental training intervention.
At time 2, we assessed the same two conditions (however the were concatenated across 4 volumes this time because this was the main focus of the study: that is two volumes for rest and two volumes for cognitive task). At the first level, I compared these two conditions.
At the second level (higher level), I calculated the group average between these conditions and ran the change in anxiety correlations across these conditions. Basically, its the same thing that I did for time 1.
Alright. All is well until this point. I ran multiple variations of paired t tests and the data is speckled with enormous z scores. I want to compare time 1 with time 2 with taking into account the brain areas associated with anxiety change. That is I want to compare the cognitive task at time 1 and time 2 as well as the brain areas corresponding to anxiety change for each respective time.
Is it appropriate to compare the mean effect of time 1 with time 2 and their corresponding changes in anxiety? When I did this the data was beautiful, however I dont think this is justified. Any suggestions? Thanks so much.
|