Dear Donald,
Thank you very much for your reply and detailed assistance. I have a few questions, however:
>>(1) Your conditions (C1 and C2) are difficulty and length. The Main Effect of difficulty would be the effect of C1. The main effect of length would be
the effect of C2.
Yes.
>>Design looked sensible (G1 G2 C1 C2 G1C1 G1C2 G2C1 G2C2 S)
>>C1 --> 1/2 1/2 1 0 1/2 0 1/2 0 ones(1,29)/29
>>C2 --> 1/2 1/2 0 1 0 1/2 0 1/2 ones(1,29)/29
While the design is correct, how you define the contrast weights for the ME of C1 and ME of C2 differs from the tutorial provided by Jan and Darren (2008). I am looking on page 9 (Design 2), where they describe how to define contrast weights for a two-group, repeated measures design (their example has three conditions, but it is essentially the same). There, they state the ME of condition should be:
zeros(1,ng) MEc MEc*[n1/ (n1+n2)] MEc*[n2/ (n1+n2)] which looks like
[0 0 1 -1 .5172 -.5172 .4828 -.4828 0] because n=15 for group1 and n=14 for group2 in our study. This contrast weight is accepted as valid in SPM. I was assuming I could compute a ME for each condition (rather than either condition or an average of both conditions) much like an SPSS output would give you. I surmise now that SPM doesn't do that?
>>(2) Since you have repeated measures, the main effect of group is invalid...
>>(3) Main effects don't have direction, so if you have two contrasts where one is the negative of the other, the main effects (F-tests) will be the
same.
Again, I don't understand this with regard to Jan and Darren's tutorial. They clearly state that ME of group can be computed thusly:
MEg = [1 -1]; main effect of group, Group 1>Group 2
Is there something in my study that is different from the tutorial example? I suspect I am misunderstanding the setup of F contrasts versus t-contrasts...
>>(4) C1-C2 contrast would be the interaction of difficulty and length...
This only makes sense in looking at my output, but again not in relation to understanding the guidelines in the tutorial. I've attached the pdf of my design, and the main aspect that pops out at me is in the condition effects. In the tutorial, it appears as though condition is nested in rapid alternating fashion within group (i.e., you can visualize each subject in the group first having C1 then C2. In my design, it looks like all subjects in group1 have C1 blocked, then all subjects in group1 have C2 blocked, then the same pattern repeats for group 2. I couldn't figure out how to make my design look like the example in the tutorial.
Any additional comments or critique would be greatly appreciated from Donald or another SPM expert reading...
Best regards,
Laura HF Barde, PhD
Department of Pediatrics
Stanford University School of Medicine
[log in to unmask]
|