help
help > RE: Paired T-test with covariates
Feb 25, 2015 02:02 AM | Alfonso Nieto-Castanon - Boston University
RE: Paired T-test with covariates
Hi Olaia,
You cannot explicitly define "condition-specific" second-level covariates in CONN. That said, for most analysis designs you should be able to simply enter all desired covariates in CONN (e.g. MedicationPre, MedicationPost, MedicationDiff, etc., each as a different covariate) and then in your second-level analysis select the appropriate combination of covariates for your planned analysis. For example, if you are interested in a paired t-test looking at post vs. pre differences in connectivity correcting for potential differences in medication between the pre- and post- conditions, in the Second-level analysis tab you would select the 'ExperimentalGroup' and 'MedicationDiff' between-subject effects (with a between-subjects contrast [1 0]) and the 'Pre' and 'Post' conditions (with a between-conditions contrast [-1 1]). This will compute the association between connectivity differences and medication differences between conditions and return the average post-pre connectivity differences after discounting the observed medication effects. Let me know if this clarifies
Hope this helps
Alfonso
EDIT: corrected typo in between-subjects contrat vector: [0 1] will test medication effects while [1 0] will test main pre/post condition effects after correcting for medication differences.
Originally posted by Olaia Lucas:
You cannot explicitly define "condition-specific" second-level covariates in CONN. That said, for most analysis designs you should be able to simply enter all desired covariates in CONN (e.g. MedicationPre, MedicationPost, MedicationDiff, etc., each as a different covariate) and then in your second-level analysis select the appropriate combination of covariates for your planned analysis. For example, if you are interested in a paired t-test looking at post vs. pre differences in connectivity correcting for potential differences in medication between the pre- and post- conditions, in the Second-level analysis tab you would select the 'ExperimentalGroup' and 'MedicationDiff' between-subject effects (with a between-subjects contrast [1 0]) and the 'Pre' and 'Post' conditions (with a between-conditions contrast [-1 1]). This will compute the association between connectivity differences and medication differences between conditions and return the average post-pre connectivity differences after discounting the observed medication effects. Let me know if this clarifies
Hope this helps
Alfonso
EDIT: corrected typo in between-subjects contrat vector: [0 1] will test medication effects while [1 0] will test main pre/post condition effects after correcting for medication differences.
Originally posted by Olaia Lucas:
Hi Alfonso,
I am trying to analyse a simple paired t-test.
I have 31 subjects (experimentals and controls), with 2 sessions (pre and post). In the condition setup, I have associated each session with 1 condition, and in the covariates setup I have introduce one covariate for experimental group and another for controls. But my question is, how could I add a covariate of medication (different values in each session) for analysis in the second-level results?
It would be possible to do only one covariate or one per condition?
Thank you in advance,
Bests,
Olaia
I am trying to analyse a simple paired t-test.
I have 31 subjects (experimentals and controls), with 2 sessions (pre and post). In the condition setup, I have associated each session with 1 condition, and in the covariates setup I have introduce one covariate for experimental group and another for controls. But my question is, how could I add a covariate of medication (different values in each session) for analysis in the second-level results?
It would be possible to do only one covariate or one per condition?
Thank you in advance,
Bests,
Olaia
Threaded View
| Title | Author | Date |
|---|---|---|
| Olaia Lucas | Feb 19, 2015 | |
| Alfonso Nieto-Castanon | Feb 25, 2015 | |
| Olaia Lucas | Feb 25, 2015 | |