Bram -
> Dear Dr. Henson,
>
> I would like to use your Matlab function for mixed design ANOVAs
> (batch_spm_anova.m). I was wondering whether the F contrasts for main
> effects and interaction effects outlined in the section
> "Generalization to M-way ANOVAs" of your chapter "Analysis of
> Variance" in the SPM book are valid for mixed-designs or whether they
> only hold for within-subject designs? If they are not valid, could you
> point me to a reference were I can find how to construct the right
> F-contrasts for mixed-design ANOVAs?
Yes, the contrast weights are equally valid for within vs
between-subject factors.
> I already tried the SPM full factorial design, which generates F
> contrasts automatically. Unfortunately, the results (ie. F-scores of
> main effects and interaction effects for a single voxel) I obtained do
> not agree with the results I obtained from a repeated measures GLM
> with a between-subject factor (fed with the same data) in SPSS. In
> contrast, when I average my contrast images accross conditions
> within-subjects (so that I have only image per subject) and then run a
> one-way between-subject ANOVA, the results from SPM and SPSS
> (univariate GLM with between-subject factor) do agree.
Yes, that is because SPSS (by default) partitions the error for each
within-subject ANOVA effect. You can emulate this in SPM by the
averaging process for each contrast that you describe (and that is the
point of the "Analysis of Variance" chapter you cite). The alternative
is to include all conditions in the same GLM and use a pooled error.
This will give different answers to partitioning the error (eg as in
SPSS), as you found. I am not sure that either approach is right or
wrong (they just make different assumptions), but there has been some
debate about this on the SPM helpline recently (cc'ed here). But
basically you can do either in SPM, so it's up to you and reviewers to
decide (as long as you explain clearly in any report which you did).
Rik
>
> Yours sincerely,
>
> Bram Zandbelt
> PhD-student
> -------------------------------------------------------
> Bram Zandbelt
> Rudolf Magnus Institute of Neuroscience
> University Medical Center Utrecht
> Room A.01.126
> P.O. Box 85500
> 3508 GA Utrecht
> The Netherlands
>
> E: [log in to unmask]
> T: +31-88-7557121
> F: +31-88-7555443
>
>
>
>
>
> ------------------------------------------------------------------------------
>
>
> De informatie opgenomen in dit bericht kan vertrouwelijk zijn en is
> uitsluitend bestemd voor de geadresseerde. Indien u dit bericht onterecht
> ontvangt, wordt u verzocht de inhoud niet te gebruiken en de afzender
> direct
> te informeren door het bericht te retourneren. Het Universitair Medisch
> Centrum Utrecht is een publiekrechtelijke rechtspersoon in de zin van
> de W.H.W.
> (Wet Hoger Onderwijs en Wetenschappelijk Onderzoek) en staat
> geregistreerd bij
> de Kamer van Koophandel voor Midden-Nederland onder nr. 30244197.
>
> Denk s.v.p aan het milieu voor u deze e-mail afdrukt.
>
> ------------------------------------------------------------------------------
>
>
> This message may contain confidential information and is intended
> exclusively
> for the addressee. If you receive this message unintentionally, please
> do not
> use the contents but notify the sender immediately by return e-mail.
> University
> Medical Center Utrecht is a legal person by public law and is
> registered at
> the Chamber of Commerce for Midden-Nederland under no. 30244197.
>
> Please consider the environment before printing this e-mail.
--
-------------------------------------------------------
Dr Richard Henson
MRC Cognition & Brain Sciences Unit
15 Chaucer Road
Cambridge
CB2 7EF, UK
Office: +44 (0)1223 355 294 x522
Mob: +44 (0)794 1377 345
Fax: +44 (0)1223 359 062
http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
-------------------------------------------------------
|