The design is fine - although for designs where the mean is not present in the model, all EVs need to be manually demeaned in conjunction with the -D option ( e.g. see https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1102&L=FSL&P=R31431&1=FSL&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4 for more information ).

Hi,

I wanna do regression analysis between FA map and behavioral data with age and sex as covariaters using randomise. I have two questions to ask.

First, I have designed matrix using GLM.

Group EV1(behav) EV2(sex) EV3(age)

1            3.2                   0             15

1            5.3                   0             18

1            4.2                   1             18

1            5.2                   1             17

1            4.9                   1             17

And the contrast:

EV1   EV2 EV3

1          0        0

I want to know the matrix is correct or not.

Second, does is it need to demean my data using the parameter ‘–D’ when runing randomise.

Thanks a lot,

Haitao

-------------------------------------------------------

Haitao Ge, Ph.D.

Center for Sectional and Imaging Anatomy
Shandong University School of Medicine
44#, West Wenhua Rd.
Jinan 250012, Shandong
P.R.China

Tel: +86-15965644699
Email: [log in to unmask]