 |
 |
I have another wondering about demean: Will –D only demean for those items included in the contrasts or will it demean for all items in the matrix? For instance, assuming there are 4 groups in the model with age as covariate. Let’s say I want to make an ANOVA testing contrast only including 3 of the groups with age as covatiate. Is it correct to then assume the –D will in the model demean age for only those 3 groups together or will it demean age including the 4th group as well, although it is not part of this particular model? The latter then givin an erroneous results, implying that the matrix should only include those groups which the contrasts should test? Best nick Niklas Lenfeldt, PhD, MSc. E.P. Clinical Neuroscience, Umeå University, Sweden Norrlands University Hospital Email: [log in to unmask]<mailto:[log in to unmask]> Phone: +46 70 6096687 From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Anderson M. Winkler Sent: den 2 november 2016 08:43 To: [log in to unmask] Subject: Re: [FSL] radomize demean Hi Nick, The -D demeans both data and design. Whether demeaning should be used or not depends on the research question encoded by the contrast and whether the intercept is already represented in the model. If not already, please see Jeanette Mumford's excellent guide on this topic: http://mumford.fmripower.org/mean_centering/ All the best, Anderson On 1 November 2016 at 17:50, Niklas Lenfeldt <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hi I am wondering about demeaning using randomize, Is it correct that for FSL5 or higher, using the –D option will demean the covariates and one does not have to enter demeaned values in the matrix? Best nick