Dear Lilla,
it is easier to include both parameters as covariates in your model and use the contrast to covariate age. Assume you have a design with one group and and your cognitive score and age. Then you have 3 columns in the design matrix:
sample effect
cognitive score
age
you can set the contrasts to
0 1 0 positive correlation with cognitive score
0 -1 0 negative correlation with cognitive score
It is the same example such as in the CAT12 manual for Multiple Regression (Linear), except that the 2nd column is here the confound and the 3rd the parameter of interest:
https://neuro-jena.github.io/cat12-help/#contrasts
Best,
Christian
On Mon, 6 Nov 2023 13:13:20 +0000, Lilia Jorge <[log in to unmask]> wrote:
>Hi SPM members
>
>I’m trying to use a multiple regression model in CAT12. My main purpose is to check whether there are associations (positive/negative) between cortical thickness measures and cognitive scores. At first, I entered the cognitive scores as covariates in the regression models, but now I’m interested in controlling for the effects of age (enter age as a nuisance variable).
>
>After some research in the SPM Archives concerning this matter, I have set up the model as follows:
>(1) Tested for the positive and negative age*cognitive_scores interactions; saved the map; excluded the significant areas from the next step
>(2) Tested for the effect of cognitive scores (-1 and 1); saved the map.
>
>Can you please check if I have done it correctly, please?
|