help
help > RE: Entering 2nd level covariates
Jul 5, 2014 07:07 AM | Alfonso Nieto-Castanon - Boston University
RE: Entering 2nd level covariates
Hi Julia & Eileen,
I second everything that Julia said, and just to add on the last point, if your main analysis of interest is going to be a between-conditions test (e.g. paired t-test) looking at connectivity differences between your pre and post conditions, and you want to control for potential movement confounds (this is in addition to the control provided by using estimated subject movement parameters as a first-level covariate / confounding effect in CONN, which should not be skipped) then I would probably suggest to enter the FD-differences between pre and post for each subject as a single second-level covariate in your analyses, since other effects such as between-subject differences in movement are going to be already controlled by your within-subjects / paired design.
Hope this helps
Alfonso
Originally posted by Julia Landsiedel:
I second everything that Julia said, and just to add on the last point, if your main analysis of interest is going to be a between-conditions test (e.g. paired t-test) looking at connectivity differences between your pre and post conditions, and you want to control for potential movement confounds (this is in addition to the control provided by using estimated subject movement parameters as a first-level covariate / confounding effect in CONN, which should not be skipped) then I would probably suggest to enter the FD-differences between pre and post for each subject as a single second-level covariate in your analyses, since other effects such as between-subject differences in movement are going to be already controlled by your within-subjects / paired design.
Hope this helps
Alfonso
Originally posted by Julia Landsiedel:
Okay if you have one per session than it is not
as easy as that.
The problem is that the array containing the covariate needs to have the same length as your number of subjects.
So if you have an array 'All' with 10 subjects, you cannot have a covariate 'FDprepost' containing 20 values. In that case you need to create two separate covariates. Alternativel you could also think again about using the FD measures as covariates in the first level as Alfonso suggested but that is just my intuitive idea (note that I myself use the toolbox for not such a very long time).
Best,
Julia
The problem is that the array containing the covariate needs to have the same length as your number of subjects.
So if you have an array 'All' with 10 subjects, you cannot have a covariate 'FDprepost' containing 20 values. In that case you need to create two separate covariates. Alternativel you could also think again about using the FD measures as covariates in the first level as Alfonso suggested but that is just my intuitive idea (note that I myself use the toolbox for not such a very long time).
Best,
Julia
Threaded View
| Title | Author | Date |
|---|---|---|
| Eileen Persichetti | Jul 2, 2014 | |
| Alfonso Nieto-Castanon | Jul 2, 2014 | |
| Eileen Persichetti | Jul 3, 2014 | |
| Julia Landsiedel | Jul 3, 2014 | |
| Eileen Persichetti | Jul 3, 2014 | |
| Julia Landsiedel | Jul 3, 2014 | |
| Eileen Persichetti | Jul 3, 2014 | |
| Julia Landsiedel | Jul 3, 2014 | |
| Alfonso Nieto-Castanon | Jul 5, 2014 | |