Hi Carole,
I initially read your question incorrectly and answered as if you were
going to include these as a covariate. I do not know how to account for
them if you do not include them as a covariate. But here is my possible
solution below if you were going to include them as covariates.
So if you are going to include plasma level and performance as EVs
(please, make sure that your values are demeaned) In the contrasts tab of
your GLM you will create 4 new contrasts: Positive effect of plasma level,
negative effect of plasma level, positive effect of performance, negative
effect of performance.
Lets say that your EV6 is your demeaned plasma level values and your EV7
is your demeaned performance level values
EV1 EV2 EV3 EV4 EV5 EV6 EV7
your other contrasts here
C# Positive effect of plasma level 0 0 0 0 0 1 0
C# negative effect of plasma level 0 0 0 0 0 -1 0
C# positive effect of performance 0 0 0 0 0 0 1
C# negative effect of performance 0 0 0 0 0 0 -1
This GLM should be included in your DR analysis command.
See "Two-Group Difference Adjusted for CovariateČ in
https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM
I think this should work for a paired test as well but please correct me
if I am wrong Anderson or anyone else.
I hope that someone else will still answer your correct question as well.
Cheers,
Noelle Dalin
On 2016-11-30, 7:37 AM, "FSL - FMRIB's Software Library on behalf of
Carole Guedj" <[log in to unmask] on behalf of
[log in to unmask]> wrote:
>Thank you for your answer!!
>
>So if I decide to not include them in the model as covariate, once I have
>obtained the result of the Dual Regression and Randomise, how can I test
>whether these differences in conditions are related to the plasma level
>of the drug or the level of performance of the subject?
>
>Thank you in advance for any help!
>
>Carole
|