Hi FSL Experts,
I have a question regarding a two group (G1 and G2), two measures design. I am interested in group differences between post (T2) and pre-treatment (T1) and set up a design similar to a previous post (https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;12b7234c.0902). I am interested in contrasts comparing groups and T2 > T1, thus my two contrast of interest will be G1 > G2, T2 > T1 and G2 > G1, T2 > T1. However, based on the GLM wiki page, I now think my design should be 3-level with a second level contrasts T2 > T1 for each participant and then passing the COPEs to a 3rd level group analysis. My initial 2nd level design looks similar to:
EV1 EV2 EV3 EV4 EV5 EV6
Input 1 (ID1:T1) -1 0 1 0 0 0
Input 2 (ID2:T1) -1 0 0 1 0 0
Input 3 (ID3:T1) 0 -1 0 0 1 0
Input 4 (ID4:T1) 0 -1 0 0 0 1
Input 5 (ID1:T2) 1 0 1 0 0 0
Input 6 (ID2:T2) 1 0 0 1 0 0
Input 7 (ID3:T2) 0 1 0 0 1 0
Input 8 (ID4:T2) 0 1 0 0 0 1
Contrasts:
G1>G2,T2>T1 1 -1 0 0 0 0
G2>G1,T2>T1 -1 1 0 0 0 0
I wanted to confirm if this an appropriate model or if I should instead use a 3-level design. In addition, I wish to calculate the effect size of these two contrasts. Based on a recent post (https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;944ae6da.1902) I originally applied d = t * sqrt( 1/N1 + 1/N2 ), however this does not seem appropriate for my design since it is a repeat measure (and that my effect sizes seem very large for N = 20) but neither does the F-test recommendation. Advice is greatly appreciated.
Regards,
Ryan
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|