Dear FSL-users,
Currently, I am working on my fMRI (FEAT) analysis and after asking several people, there are still some parts I am not sure about. Therefore, I was hoping you could help me out.
In my study, a N-back (working memory) task was administered in 5 groups:
* Patients with Dissociative Identity Disorder (who have multiple identity states) (DID-G)
* Actors who simulated the disorder (DID-S)
* Imitative patients (who believe they have the disorder but do not meet the official criteria) (DID-I)
* Patients with Post-traumatic Stress Disorder (PTSD)
* Healthy controls (HC)
The thing is, that patients with DID-G, DID-S and DID-I were scanned twice; once as a neutral identity state (NIS) and once as a trauma-related identity state (TIS). In the control groups, the HC represents a NIS and PTSD respresents a TIS.
On the first level, I made 5 copes (mean, 1back, 2back, 3back, linear load), separately for NIS and TIS.
In the higher level analysis, I would like to compare the NIS and TIS differences from DID-G with the NIS-TIS differences in DID-S and DID-I (and also between DID-S and DID-I); a 3 way interaction, I believe (?). I figured out a way to do this but, first of all, I am not sure if this is a valid way of testing. Furthermore, there are some other questions that remain.
A picture of the design is attached (only cope 5; linear load). In this design, I entered 13 DID-G patients (first all NIS, then all TIS), 15 DID-S (first NIS, then TIS) and 16 controls (HC (as NIS) and PTSD (representing TIS)). The DID-G and DID-S are paired since NIS and TIS belong to the same person (hence G1-G13 and S1-S15), but controls are unpaired, since the were not actually the same subject (so I made no C1-C16).
* Is it aloud to have 2 paired and one unpaired group within one design? And is it problematic that the groups do not have equal sizes?
With the 6 contrasts, I try to figure out within groups, where NIS shows more activation than TIS and the other way around.
* Is the way I defined the contrasts a valid one for that question? And what would it mean when Io define C1 as 1 -1 and C2 as -1 1?
And then, at last, I would like to investigate the NIS and TIS differences from DID-G with the NIS-TIS differences in DID-S and DID-I (and also between DID-S and DID-I).
* Is a valid way of doing so, with a F-test? And, if yes, how should I define the F-tests?
I really hope you can help me out with this. Thanks a lot for your help!
Kind Regards,
Eline
|