help
help > CONN scripting: multiple setup.conditions for multiple sess x runs design->loss of degrees of freedom (is it a problem?)
Jan 23, 2022 12:01 PM | Panagiotis Iliopoulos
CONN scripting: multiple setup.conditions for multiple sess x runs design->loss of degrees of freedom (is it a problem?)
Hello CONN experts & Alfonso,
Alfonso informed me in a previous post that to analyze a dataset with 2 sessions x 2 runs.
--> I should split the 9 trial type/conds I have into 18 conditions at the setup.conditions field
(by differentiating each cond to _pre & _post). e.g. 'lures_pre', 'lures_post'
1) My question is whether statistically this causes a noticeable problem, given that more conds in our 1st model will consume more degrees of freedom? (by having 18 instead of 9 regressors).
Is there a way I can avoid that or it does not pose a noteworthy problem?
As far as I understand, one likely should define the 2 sessions at the 2nd level, however at the CONN batch it seems you cannot do that(?).
Since you must have defined the conditions at the setup.cond, to assign them weights at the analysis.between_conditions etc fields.
*************************************
2) Is it a solid alternative to just continue the analysis at SPM for the 2nd level? (using the 9 regressors from the 1st level computed at CONN)
or is there a workaround in CONN?
Any help would be very welcome. Many thanks :)
Alfonso informed me in a previous post that to analyze a dataset with 2 sessions x 2 runs.
--> I should split the 9 trial type/conds I have into 18 conditions at the setup.conditions field
(by differentiating each cond to _pre & _post). e.g. 'lures_pre', 'lures_post'
1) My question is whether statistically this causes a noticeable problem, given that more conds in our 1st model will consume more degrees of freedom? (by having 18 instead of 9 regressors).
Is there a way I can avoid that or it does not pose a noteworthy problem?
As far as I understand, one likely should define the 2 sessions at the 2nd level, however at the CONN batch it seems you cannot do that(?).
Since you must have defined the conditions at the setup.cond, to assign them weights at the analysis.between_conditions etc fields.
*************************************
2) Is it a solid alternative to just continue the analysis at SPM for the 2nd level? (using the 9 regressors from the 1st level computed at CONN)
or is there a workaround in CONN?
Any help would be very welcome. Many thanks :)
Threaded View
Title | Author | Date |
---|---|---|
Panagiotis Iliopoulos | Jan 23, 2022 | |
Alfonso Nieto-Castanon | Jan 25, 2022 | |