help
help > RE: 2-by-3 repeated measures ANOVA
May 13, 2022 03:05 AM | Andrew Zalesky
RE: 2-by-3 repeated measures ANOVA
Hi Konstantinos,
yes - this looks correct me.
For the second case, you may need to ask whether averaging over all the tasks is sensible, given that particular tasks may be associated with distinct task-specific connectivity patterns. There may be certain connections that are universally engaged by all the tasks (and your second analysis would be sensitive to these connections), but it might not provide sensitivity to connections that are uniquely activated by only one of the tasks.
In summary, both of the designs appear to be statistically correct, but you may need to determine which design is most suited to your research question.
Andrew
Originally posted by Konstantinos Tsilimparis:
yes - this looks correct me.
For the second case, you may need to ask whether averaging over all the tasks is sensible, given that particular tasks may be associated with distinct task-specific connectivity patterns. There may be certain connections that are universally engaged by all the tasks (and your second analysis would be sensitive to these connections), but it might not provide sensitivity to connections that are uniquely activated by only one of the tasks.
In summary, both of the designs appear to be statistically correct, but you may need to determine which design is most suited to your research question.
Andrew
Originally posted by Konstantinos Tsilimparis:
Dear dr. Zalesky,
Thank you for your previous recommendations.
To test the group effect I followed two different ways:
1. I did a two-sample unpaired t-test for each emotion separately. For example, to test if ASD>Control on the Happy emotion I used the Design Matrix and Contrast (I suppose that each group has 2 subjects):
1 0
1 0
0 1
0 1
1st column: ASD
2nd column: Control
with contrast [1, -1]
2. I averaged across all emotions for each subject. For example for subject 1 I averaged together the adjacency matrices for Angry, Happy and Neutral emotion. Similarly for subject 2, 3 and 4. The Design matrix and Contrast for ASD>Control is the same as before:
1 0
1 0
0 1
0 1
1st column: ASD
2nd column: Control
with contrast [1, -1]
Are these methodologies correct? What methodology gives a more robust statistical result?
I really appreciate your help.
Kind regards,
Konstantinos Tsilimparis
Thank you for your previous recommendations.
To test the group effect I followed two different ways:
1. I did a two-sample unpaired t-test for each emotion separately. For example, to test if ASD>Control on the Happy emotion I used the Design Matrix and Contrast (I suppose that each group has 2 subjects):
1 0
1 0
0 1
0 1
1st column: ASD
2nd column: Control
with contrast [1, -1]
2. I averaged across all emotions for each subject. For example for subject 1 I averaged together the adjacency matrices for Angry, Happy and Neutral emotion. Similarly for subject 2, 3 and 4. The Design matrix and Contrast for ASD>Control is the same as before:
1 0
1 0
0 1
0 1
1st column: ASD
2nd column: Control
with contrast [1, -1]
Are these methodologies correct? What methodology gives a more robust statistical result?
I really appreciate your help.
Kind regards,
Konstantinos Tsilimparis
Threaded View
| Title | Author | Date |
|---|---|---|
| Konstantinos Tsilimparis | Mar 2, 2022 | |
| Konstantinos Tsilimparis | May 13, 2022 | |
| Konstantinos Tsilimparis | May 11, 2022 | |
| Andrew Zalesky | May 13, 2022 | |
| Konstantinos Tsilimparis | May 13, 2022 | |
| Andrew Zalesky | May 13, 2022 | |
| Andrew Zalesky | Mar 3, 2022 | |
| Konstantinos Tsilimparis | Mar 28, 2022 | |
| Andrew Zalesky | Mar 29, 2022 | |