I'm not sure if you should assign a weight to the questionnaire data in your
contrast when using it as a covariate of no interest. Also, I think if you
do include it in the contrast, it should be negatively weighted. Maybe
someone else could affirm/disaffirm?
It does not matter what "order" the covariates are in the design matrix.
However, at the 2nd level of analysis, your contrast weights should sum to
zero (0), hence:
Group1 questionnaire Group2:
1 0 -1
Or , if including the questionnaire data in the contrast:
1 1 -2
Hope this helps,
Julie
Julie E. McEntee, M.A., C.C.R.P.
Senior Research Program Coordinator
Department of Psychiatry and Behavioral Sciences
Division of Psychiatric Neuroimaging
Johns Hopkins University School of Medicine
600 N. Wolfe St./ Phipps 300 (office: room 317)
Baltimore, MD 21287
Phone: 410-502-0468
Fax: 410-614-3676
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Ciara McCabe
Sent: Thursday, July 17, 2008 7:46 AM
To: [log in to unmask]
Subject: [SPM] covariates of no interest in SPM5
Dear SPMers
I would like your help just to clarify that I'm doing the right thing,
I have at the 2nd level ran multiple regression with 1st covariate as 1st
group and 2nd cov as 2nd group
then I subtract one form the other for my contrast
but now I want to check that questionnaire results for e.g. when added as a
covariate of no interest
doesn't acc for too much of the data and the results are still signif
to do this should I just then add the quest data in as a third covariate
and continue as norm ?
and if so, the design matrix ends up with the 1st group as 1st column then
the questionnaire data as second column then the second group as the third
column does this matter?
then I do 1 0 -1 for eg.
Is this right?
Thanks in advance for help
Ciara
Dr Ciara McCabe
University of Oxford
Dept of Psychiatry
Neuroscience Building
Warneford Hospital
Oxford
OX3 7JX
Telephone (01865)-223778
mob: 07966035631
[log in to unmask]
http://users.ox.ac.uk/~psyc0288/perl/ciara_mcabe.html
|