Dear Alfonso,

We have a repeated measures design with subjects who were scanned at different TRs. As a result, we attempted to take your advice posted on Dec 1, 2013 to Mengyu Tian in the Conn forum, which worked fine. https://www.nitrc.org/forum/message.php?...

However, rather than looking at only Time1 vs Time 2 differences across all subjects (as in the Tian circumstance), we're interested in looking at any potential difference between controls and between patients. So we changed the covariates to:

Time1_Control
Time2_Control
Time1_TBI
Time2_TBI
within1
within2
.....
within24

We have 24 pairs, that includes 5 controls and 19 patients.

When doing the Time1_Control vs. the Time2_Control comparison, we

1) selected Time1_Control, Time2_Control, and only the 5 control subjects. Then in the "between-subjects contrast" field, we entered [1 -1 zeros(1,7)]. When we clicked "enter", this came up:

[-1 1 0 0 0 0 0; 0 -1 1 0 0 0 0; 0 0 -1 1 0 0 0; 0 0 0 -1 1 0 0; 0 0 0 0 -1 1 0; 0 0 0 0 0 -1 1]

This didn't seem to make sense. So then we tried something else and

2) defined the contrast manually with:

[1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0; 1 -1 0 0 0 0 0 0 0]

The results from this seemed to have a regular pattern, although we don't know if it's really reflecting what we want. We attached a jpg of the results from the Results Explorer. We also noted that this contrast produced an F, rather than a T, test.

We also tried:

3) [1 -1 zeros(1,24)], and the resulting patterns looked irregular/not right. In case it's relevant, in the title bar of the Results Explorer gui, it said, "Time1_Control(1).Time2_Control(-1).within1(0).within2(0).within3(0)...within18/dmn.LLP/spm.mat" (We think it might have stopped at "within18" there just wasn't room to list all 24.)

Thank you for your answer,
Xiaodi and Mary