Dear CONN experts,
We are trying to perform two-way repeated measures ANOVA on ROI to ROI connectivity in CONN, testing the effects of Placebo (A) vs. Medication (B) as the first repeated measures factor (2 levels), and Reward (1) vs. Non-Reward (2) as the second repeated measures factor (2 levels). We set up the data so that each subject has 4 variables: A1, A2, B1, B2.
To test the interaction effect, we tried two ways which gave us the same results: 1 -1 0 0; 1 0 -1 0; 0 1 0 -1; 0 0 1 -1 or 1 -1 0 0; 0 1 -1 0; 0 0 1 -1. The latter is the contrast defined by ‘Any difference’ in the dropdown menu for the condition effect.
We are wondering when using the above contrasts if we are getting ‘any' effect (main and/or interaction) rather than interaction effect only? If so, how do we correctly specify the interaction effect (A-B*1-2)?
When using the above contrast, for each ROI-ROI, we get degrees of freedom of F (3, 15). We have 18 subjects. When we enter contrasts for what we think testing for a main effect, e.g., 1-1 0 0; 0 0 1 -1, we get degrees of freedom of F (2, 16).
Because we are running the repeated measures with 2 factors with 2 levels in each, shouldn't the degrees of freedom be F(1, 17) for both main and interaction effects? df (medication [2 levels]) = 2 -1 = 1, df (reward [2 levels]) = 2 -1 = 1, df medicationXreward = 1 x 1 = 1, df subject (18) = 18-1 = 17, df medicationX subject = 1 x 17 = 17, df rewardXsubject = 1 x 17 = 17, df medicationXrewardXsubject = 1 x 1 x 17 = 17
Does the CONN set up the F test in a different way? If so, could you enlighten us as to how it results in the degrees of freedom we get?
Any insight would be appreciated!
Thank you!
Best,
Patricia
|