Just my two cents:

I think the age should be mean-centered; otherwise, the meaning of the other variables need to be interpreted as age = 0 (rather than mean age = 0).

In general, I believe in qdec or mri_glmfit, the effects of Class variable are interpreted as 0 on the continuous/nuisance variables.

Best,

Daniel

Yung-Jui "Daniel" Yang, PhD

Postdoctoral Researcher

Yale Child Study Center

New Haven, CT

(203) 737-5454

On 9/11/13 1:13 PM, "Erik Lindberg" <erikoloflindberg@gmail.com> wrote:

Dear Douglas,

attached you´ll find the two y.fsgd file from the analysis run in GLM_FIT. the results are completely different (the z-score converted age produces a believable result, while the result correcting with raw age scores seems completely wrong.

best regards

Olof