help
help > Clarification on contrasts in CONN 2nd-level multivariate analysis
Oct 4, 2018 09:10 AM | Martyn McFarquhar
Clarification on contrasts in CONN 2nd-level multivariate analysis
Dear Alfonso,
I am currently working with some colleagues on a data analysis using CONN. They have sent me the individual subject component maps from an MVPA analysis, and have reported 2nd-level results using these components in CONN, but I have been unable to replicate their results using my own MANOVA implementation. I was just wondering whether you could clarify exactly what CONN is doing so that I can hopefully replicate this.
For context, the analysis is based on the 2 x 2 x 2 mixed ANOVA example on page 26 of the CONN manual. There are pre- and post-treatment scans and so each subject has a set of components for the pre and a set for the post. Based on this I was assuming that CONN was running some sort of omnibus test across components, so the contrasts would be something like (using 3 components as an example):
C = [1 -1 -1 1]
M = [1 -1 0 0 0 0
0 0 1 -1 0 0
0 0 0 0 1 -1]
such that you would get a multivariate test across components on the difference between pre and post, however, this does not seem to be producing the same results in my own implementation (using Wilks Lambda as the test statistic).
Could you please clarify how the omnibus test from across components is formulated in CONN and whether I have misunderstood anything about the CONN implementation.
Best wishes,
Martyn McFarquhar
I am currently working with some colleagues on a data analysis using CONN. They have sent me the individual subject component maps from an MVPA analysis, and have reported 2nd-level results using these components in CONN, but I have been unable to replicate their results using my own MANOVA implementation. I was just wondering whether you could clarify exactly what CONN is doing so that I can hopefully replicate this.
For context, the analysis is based on the 2 x 2 x 2 mixed ANOVA example on page 26 of the CONN manual. There are pre- and post-treatment scans and so each subject has a set of components for the pre and a set for the post. Based on this I was assuming that CONN was running some sort of omnibus test across components, so the contrasts would be something like (using 3 components as an example):
C = [1 -1 -1 1]
M = [1 -1 0 0 0 0
0 0 1 -1 0 0
0 0 0 0 1 -1]
such that you would get a multivariate test across components on the difference between pre and post, however, this does not seem to be producing the same results in my own implementation (using Wilks Lambda as the test statistic).
Could you please clarify how the omnibus test from across components is formulated in CONN and whether I have misunderstood anything about the CONN implementation.
Best wishes,
Martyn McFarquhar
Threaded View
Title | Author | Date |
---|---|---|
Martyn McFarquhar | Oct 4, 2018 | |
Alfonso Nieto-Castanon | Oct 5, 2018 | |
Martyn McFarquhar | Oct 5, 2018 | |
Alfonso Nieto-Castanon | Oct 5, 2018 | |
Martyn McFarquhar | Oct 8, 2018 | |
Alfonso Nieto-Castanon | Oct 8, 2018 | |
Martyn McFarquhar | Oct 9, 2018 | |
Alfonso Nieto-Castanon | Oct 9, 2018 | |
Ali Amad | Oct 11, 2018 | |
Alfonso Nieto-Castanon | Oct 12, 2018 | |
Ali Amad | Oct 19, 2018 | |
Martyn McFarquhar | Oct 11, 2018 | |
Alfonso Nieto-Castanon | Oct 12, 2018 | |
Martyn McFarquhar | Oct 12, 2018 | |
Alfonso Nieto-Castanon | Oct 12, 2018 | |
Martyn McFarquhar | Oct 15, 2018 | |
Martyn McFarquhar | Oct 5, 2018 | |