Dear SPM community, 

I have couple of quenstions concerning the statistical analysis of longitudinal fMRI.

I have an experimental design where I am testing 3 reading types and suitable control tasks (such as observing symbols), before and after a training. I want to check how the training changed the process of reading. 

After having read through this email list I decided to analyze my data with flexible factorial with 3 factors: subjects, time points, reading types.

Firstly, I am wondering if in the reading type factor I should put the 1st level contrasts for reading or control conditions vs baseline (that would give me a 2x6 anova) or if I could put there already the contrasts of a reading type vs specific control (eg. reading vs observing symbols), which would result in 2x3 anova. Are there any constraints concerning the factorial analyses of more complicated contrasts than those vs baseline?

Secondly, I have already tried to set up a flexible factorial model of my data. I have three factors in my design:

F1: 9 subjects (A)

F2: 2 time points (before and after the training) (B)

F3: 3 reading types (C)

For each subject I have set the 6x3 conditions matrix, according to the order below:

A1 B1 C1 

A1 B1 C2

A1 B1 C3

A1 B2 C1 

A1 B2 C2 

A1 B2 C3

For each subject I have ordered the contrast images according to the order in the condition matrix.

When I explored my design matrix (reviev->design->explore-> files and factors), I observed that the files are not ordered in a right way. It seems like SPM have sorted them by the contrast number, disregarding the conditions matrices I typed.  Attached, I send you a picture of my design matrix and a PDF file with the files and factors it the data. I do not really know how to correct it, do you maybe have any idea? I have specified my model accoring to the tutorial of Glashel&Gitelman.

I am sorry if my questions seem trivial, I am a begginer in fMRI analysis.

I will be very grateful for your help!