Dear Experts,
I am running into trouble with a flexible factorial design wherein the condition effects seemed to be merged together. I would greatly appreciate any feedback from the watchful eye of those more experienced than I!
The study had two groups (patient, control - unequal groups of 15 and 14, respectively) and two repeated measures conditions ("difficulty:" contrast image of high-low difficulty, and "length:" contrast image of long-short sentences).
I followed Jan and Dan's conweights tutorial. I specified three factors: subject (independent, equal variance), group (independent, unequal variance), and condition (dependent, equal variance). There were two main effects (factor 2-group, and factor 3-condition) and one interaction (factors 2 and 3).
In subjects I entered all the scans and the condition array was ordered thusly:
1 1; 1 2; 1 1; 1 2; ... (for 15 subjects in group 1)...2 1; 2 2; 2 1; 2 2; ... (for 14 subjects in group 2). This gave me a 58 x 2 condition array and 58 contrast image scans.
Design looked sensible (G1 G2 C1 C2 G1C1 G1C2 G2C1 G2C2 S)
Contrast weights on the results were:
ME Group (Patients>Controls): [1 -1 0 0 .5 .5 -.5 -.5] (all contrasts had a 0 on the end for subjects)
ME Group (Controls>Patients): [-1 1 0 0 -.5 -.5 .5 .5]
ME Condition (difficulty): [0 0 1 -1 .5172 -.5172 .4828 -.4828]
ME Condition (length): [0 0 -1 1 -.5172 .5172 -.4828 .4828]
Interaction group X condition: [0 0 0 0 1 -1 -1 1]
When I analyzed for ME of group, there were none, which was consistent with an earlier preliminary two-way ANOVA. However, the ME of condition is identical for both difficulty and length. What am I doing wrong?
I also tried entering the subjects' scans and specifying the condition array as:
1 1; 1 1; 1 1; ...(all group 1 subject scans for condition 1)...1 2; 1 2; 1 2; ...(all group 1 subject scans for condition 2)...2 1; 2 1; 2 1; (all group 2 subject scans for condition 1)...2 2; 2 2; 2 2;...(all group 2 subject scans for condition 2). But this had the same effect as what I described above.
Thank you in advance for any advice.
Best regards,
Laura Barde
Department of Pediatrics
Stanford University School of Medicine
|