I just wish to correct an error of terminology I made in my
just-previous message:
I wrote:
> What I am not sure about in SPM is which would be the best module to
> use for the 2nd level analysis if one wishes to relax the
> sphericity assumption. I think some of what Stephen has been saying
> is relevant to this issue. But if one assumes sphericity vis a vis
> groups (i.e., homoscedasticity across groups), then Basic Models
> would be correct to use.
Above I incorrectly used the term "sphericity" to refer to homogeneity
of variance across groups. Sphericity properly refers to the
assumption that the variances of all differences between treatment
levels in a repeated measures design are identical. This has no
bearing on my answer to Thomas's question.
Eric
|