Hi SPM Experts!
I was trying to figure out, how to orthogonalize linear,quadratic and cubic age expansions as
been used by Good et al based on a framework developed by Buechel et al.
As first appoach I tried just to calculate pol = [age age.^2 age.^3]
and orthogonalized this matrix in Matlab with [Q,R] = qr(pol,0).
I ran in the same problems as been discussed by Buechel a long time ago:
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0004&L=spm&P=R10294&I=-1
He suggested a mean correction...
Therefore, I tried to do, what I understood under a mean correction:
A1 = age - mean(age)
pol = [ A1 A1.^2 A1.^3]
However, after doing the mean correction values are centered around zero and values of e.g.
-10 and +10 become 100 for the quaddratic part and +/-1000 after the cubic part.
This did not make much sense to me... and it would mean that in the case of a mean age of
30, 20 year and 40 year old subjects have the same degree of qudratic expansion...
Well, I have probably missed something....
If anyone knows the answer, please let me know....
Thanks,
Lukas
|