help
help > RE: How to consider effects of baseline resting-state in ANOVA anaylsis
Nov 10, 2020 09:11 PM | Alfonso Nieto-Castanon - Boston University
RE: How to consider effects of baseline resting-state in ANOVA anaylsis
Dear Jun,
Yes, in these analyses a 2x2 ANOVA design is perfectly appropriate and it is not necessary to include the baseline condition. More specifically, introducing the baseline condition as a common reference/control condition in those comparisons would not change anything at all, since the same baseline is used for the two conditions of interest that you wish to compare, so the three suggested between-condition contrasts:
option 1) Between-condition contrast (P1, P2): [-1 1]
option 2) Between-condition contrast (B, P1, P2): [0 -1 1] option 3) Between-condition contrast (P1-B, P2-B): [-1 1]
would simply produce exactly the same results, since all of them are effectively looking at the P2-P1 differences in connectivity (e.g. [B, P1, P2] * [0 -1 1] = 0*B -1*P1 +1*P2 = P2 - P1; and similarly, [P1-B, P2-B] * [-1 1] = -1*(P1-B) +1*(P2-B) = -P1+B +P2 -B = P2 - P1)
Hope this helps
Alfonso
Originally posted by jun Kim:
Yes, in these analyses a 2x2 ANOVA design is perfectly appropriate and it is not necessary to include the baseline condition. More specifically, introducing the baseline condition as a common reference/control condition in those comparisons would not change anything at all, since the same baseline is used for the two conditions of interest that you wish to compare, so the three suggested between-condition contrasts:
option 1) Between-condition contrast (P1, P2): [-1 1]
option 2) Between-condition contrast (B, P1, P2): [0 -1 1] option 3) Between-condition contrast (P1-B, P2-B): [-1 1]
would simply produce exactly the same results, since all of them are effectively looking at the P2-P1 differences in connectivity (e.g. [B, P1, P2] * [0 -1 1] = 0*B -1*P1 +1*P2 = P2 - P1; and similarly, [P1-B, P2-B] * [-1 1] = -1*(P1-B) +1*(P2-B) = -P1+B +P2 -B = P2 - P1)
Hope this helps
Alfonso
Originally posted by jun Kim:
Dear conn users and Dear Alfonso
I have a question about my design setup.
I have 2 groups (individuals who underwent different interventions, Ma and Mb), and each group underwent cognitive tasks before and after the intervention. Two resting-state fMRI was acquired after cognitive tasks.
So, my study design comprises of 5 conditions (baseline resting-state, cognitive task2 [C1], resting-state[P1], intervention[M], cognitive task2 [C2], resting-state[P2]).
I want to test the hypothesis that different interventions (Ma vs. Mb) may differently alter functional connectivity during cognitive tasks (C1 vs. C2) and resting-state FC after cognitive tasks (P1 vs. P2).
For this, I did two 2x2 ANOVA anaylses.
ANOVA1: interaction between groups (Ma and Mb) and conditions (resting state after cognitive tasks, P1 and P2)
Between-subject contrast: [-1 1]
Between-condition contrast: [-1 1]
ANOVA2: interaction between groups (Ma and Mb) and conditions (during cognitive tasks, C1 and C2)
Between-subject contrast: [-1 1]
Between-condition contrast: [-1 1]
My first question is whether I should include baseline resting-state in between-condition contrast and how to include.
In my opinion, I would be able to include baseline condition by addressing baseline condition as a covariate as;
Between-condition contrast (B, P1, P2): [0 -1 1]
or, subtracting baseline condition from other condition conditions as;
Between-condition contrast (P1-B, P2-B): [-1 1]
Suppose I would be better to include the baseline condition, which of the two ways is more appropriate. Or is there any other better way?
Additionally, I also ask how to make a second analysis, "Between-condition contrast (P1-B, P2-B): [-1 1]".
Best
Jun
I have a question about my design setup.
I have 2 groups (individuals who underwent different interventions, Ma and Mb), and each group underwent cognitive tasks before and after the intervention. Two resting-state fMRI was acquired after cognitive tasks.
So, my study design comprises of 5 conditions (baseline resting-state, cognitive task2 [C1], resting-state[P1], intervention[M], cognitive task2 [C2], resting-state[P2]).
I want to test the hypothesis that different interventions (Ma vs. Mb) may differently alter functional connectivity during cognitive tasks (C1 vs. C2) and resting-state FC after cognitive tasks (P1 vs. P2).
For this, I did two 2x2 ANOVA anaylses.
ANOVA1: interaction between groups (Ma and Mb) and conditions (resting state after cognitive tasks, P1 and P2)
Between-subject contrast: [-1 1]
Between-condition contrast: [-1 1]
ANOVA2: interaction between groups (Ma and Mb) and conditions (during cognitive tasks, C1 and C2)
Between-subject contrast: [-1 1]
Between-condition contrast: [-1 1]
My first question is whether I should include baseline resting-state in between-condition contrast and how to include.
In my opinion, I would be able to include baseline condition by addressing baseline condition as a covariate as;
Between-condition contrast (B, P1, P2): [0 -1 1]
or, subtracting baseline condition from other condition conditions as;
Between-condition contrast (P1-B, P2-B): [-1 1]
Suppose I would be better to include the baseline condition, which of the two ways is more appropriate. Or is there any other better way?
Additionally, I also ask how to make a second analysis, "Between-condition contrast (P1-B, P2-B): [-1 1]".
Best
Jun
Threaded View
| Title | Author | Date |
|---|---|---|
| jun Kim | Nov 5, 2020 | |
| Alfonso Nieto-Castanon | Nov 10, 2020 | |
| jun Kim | Nov 11, 2020 | |