Dear Claus,
> 1) What is the difference between multiple regression and multiple
> regression with constant? Which one should I use when I want to asses
> the effects of two predictors (scores of questionnaires) on contrast images
> (results of fixed effects analysis)
I would recommend the multiple regression with a constant term, since this is the most general way to do a regression.
It's according to y = b1*x1 + b2*x2 + constant , which is the usual equation for a linear regression analysis.
> is this because I don't have a constant term, so the significant voxels
> are simply reflecting the mean activity in the whole sample?
Yes, I guess this happened in your case. If one variable has almost no correlation with the observed data, it could mainly capture the mean, i.e. the offset in your regression and you did basically only something like y = constant + b2*x2 , what was not what you intended to do.
Good luck,
Karsten
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
& National Competence Centre for functional MRI
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
[log in to unmask]
http://fmri.uib.no/
|