help
help > RE: Covariate of interest
Jul 13, 2018 03:07 PM | Alfonso Nieto-Castanon - Boston University
RE: Covariate of interest
Hi Julien,
If I am understanding correctly you are looking at differences in connectivity before- and after- treatment, while controlling for those differences that may be more simply explained by pain-score differences among the two sessions. If that is the case your analysis set up looks perfectly correct, and the choice between the two versions of your analyses (whether to represent pain scores differences in raw or percent units) should probably be based on the nature/interpretation of your pain-score measures and your a priori assumptions about the association between those pain-score differences and connectivity-differences (e.g. if you expect that a difference in pain scores S=1 before treatment and S=2 after treatment would be commensurate in terms of expected connectivity differences to a difference in pain scores S=50 before treatment and S=100 after treatment then probably a percent change measure makes most sense, while if you expect that a difference in pain scores S=1 before treatment and S=2 after treatment would be commensurate to a difference in pain scores S=50 before treatment and S=51 after treatment then probably a raw change measure makes most sense)
Hope this helps
Alfonso
Originally posted by julien poublanc:
If I am understanding correctly you are looking at differences in connectivity before- and after- treatment, while controlling for those differences that may be more simply explained by pain-score differences among the two sessions. If that is the case your analysis set up looks perfectly correct, and the choice between the two versions of your analyses (whether to represent pain scores differences in raw or percent units) should probably be based on the nature/interpretation of your pain-score measures and your a priori assumptions about the association between those pain-score differences and connectivity-differences (e.g. if you expect that a difference in pain scores S=1 before treatment and S=2 after treatment would be commensurate in terms of expected connectivity differences to a difference in pain scores S=50 before treatment and S=100 after treatment then probably a percent change measure makes most sense, while if you expect that a difference in pain scores S=1 before treatment and S=2 after treatment would be commensurate to a difference in pain scores S=50 before treatment and S=51 after treatment then probably a raw change measure makes most sense)
Hope this helps
Alfonso
Originally posted by julien poublanc:
Hello,
I am working on a project with fibromyalgia patients. We have a BOLD resting state on 15 patients pre and post hyperbaric treatment.
And we have pain scores (pre and post) ranging from 0 to 100.
- I did all the default pre-processing.
- Then, I input the pre-post difference of the pain scores.
- I used the ROI-to-ROI analysis
In Subject effects, I selected "all patients" and then "pain scores difference".
For Between-subjects contrast I wrote [1 0] (or simple main effect of "all subjects")
For condition:
pre
post
Between con contrast: [-1 1]
The question is:
Should I input a "percentage difference" for the pain scores or simply a difference?
Because I get difference answers whether I input one or the other.
Thank you .
Julien.
I am working on a project with fibromyalgia patients. We have a BOLD resting state on 15 patients pre and post hyperbaric treatment.
And we have pain scores (pre and post) ranging from 0 to 100.
- I did all the default pre-processing.
- Then, I input the pre-post difference of the pain scores.
- I used the ROI-to-ROI analysis
In Subject effects, I selected "all patients" and then "pain scores difference".
For Between-subjects contrast I wrote [1 0] (or simple main effect of "all subjects")
For condition:
pre
post
Between con contrast: [-1 1]
The question is:
Should I input a "percentage difference" for the pain scores or simply a difference?
Because I get difference answers whether I input one or the other.
Thank you .
Julien.
Threaded View
Title | Author | Date |
---|---|---|
julien poublanc | Jul 10, 2018 | |
Alfonso Nieto-Castanon | Jul 13, 2018 | |
julien poublanc | Jul 27, 2018 | |