Paul,
The model of your 3dttest++ analysis seems to look like this:
inflammation ~ cytokine + motion + VBM_Jacobians
> Is this the proper way to proceed? I think so but I wanted to check.
You specified the above model, so the results would be interpreted as being you intended to achieve: you want the effect of cytokine on inflammation while controlling motion and VBM Jacobians (at their respective average values?).
> I found a significant relation between my two variables of interest (cytokines and neural inflammation,
> accounting for the noise covariate effects) in the resulting statistical map AND I wanted to show this
> relation in a scatterplot accompanying the statistical map.
The motion and VBM Jacobians vary across subjects, and that's why you wanted to control their effects at their mean values. On the other hand, you do not have the inflammation values corresponding to the observed cytokine when motion and VBM Jacobians are at their respective average values. In other words, if you just directly do the scatter plot with the observed inflammation vs. cytokine, you won't be able to control motion and Jacobians. Ideally you would need a 3D plot, but that would be not intuitive. Are you thinking about scatterplot at the ROI level?
Gang
Edited 1 time(s). Last edit at 10/01/2015 04:17PM by Gang.