Hi,
I have 3 groups of subjects (CON, ALC, and METH) and I want to determine:
1) whether there are positive or negative correlations between brain activation on a task
in the scanner and scores obtained on another task (called DS) outside the scanner.
2) whether these correlations are significant within each group
3) whether there is a significant group by DS score interaction (i.e. whether the
correlation between brain activation and DS score is the same across all groups)
I have 6 EVs: CON, METH, ALC, CON/DS(scores on DS demeaned from CON group mean),
METH/DS(scores on DS demeaned from METH group mean) and ALC/DS(scores on DS
demeaned from ALC group mean) and this is the set up, assuming 3 subjects per group:
CON METH ALC CON/DS METH/DS ALC/DS
1 0 0 DS-c1 0 0
1 0 0 DS-c2 0 0
1 0 0 DS-c3 0 0
0 1 0 0 DS-m1 0
0 1 0 0 DS-m2 0
0 1 0 0 DS-m3 0
0 0 1 0 0 DS-a1
0 0 1 0 0 DS-a2
0 0 1 0 0 DS-a3
Based on discussions with colleagues, I used the following contrast and f-tests:
CON METH ALC CON/DS METH/DS ALC/DS F1 F2
DS Positive Interaction 0 0 0 .33 .33 .33
DS Negative Interaction 0 0 0 -.33 -.33 -.33
CON Positive Correlation 0 0 0 1 0 0 X
METH Positive Correlation 0 0 0 0 1 0 X
ALC Positive Correlation 0 0 0 0 0 1 X
CON Negative Correlation 0 0 0 -1 0 0 X
METH Negative Correlation 0 0 0 0 -1 0 X
ALC Negative Correlation 0 0 0 0 0 -1 X
I'm having difficulties with the interpretations of my results. In particular, what do my
first 2 contrasts tell me? How is it different from what my 2 f-tests tell me? Is this matrix
the best way to answer my questions of interest?
Thank you so much for any help.
|