Dear Yannis,
For A) and B) I would actually use the same design (one-way ANOVA with 8
levels) and F-contrast, the difference being the input images:
Active+Placebo for A) and Active-Placebo for B).
For C) this would be a paired t-test for each time point, or
equivalently a one-sample t-test of the difference.
Best regards,
Guillaume.
On 20/07/18 11:38, Yannis Paloyelis wrote:
> Dear Guillaume, list,
>
> I wonder if you could help with another question on a different repeated
> measures design.
> N=16
> Factor 1: Treatment (active, placebo)
> Factor 2: Time (8 time points) (perfusion maps)
>
> This can be specified easily in a flexible factorial.
>
> But if we were to avoid the non-sphericity issues in a flexible, would
> it make sense to create contrast images Active-Placebo for each time
> point (hence 8 images)? How would we then ask for these effects:
> A. Main of treatment
> B. Treatment x Time
> C. An F test, testing for an effect of treatment at any time point?
>
> _My attempts to an answer: _
>
> B. Treatment x Time: conduct a one factor repeated measures ANOVA on
> (Active-placebo) CBF maps and use this contrast:
> 1 -1 0 0 0 0 0 0
> 0 1 -1 0 0 0 0 0
> 0 0 1 -1 0 0 0 0
> 0 0 0 1 -1 0 0 0
> 0 0 0 0 1 -1 0 0
> 0 0 0 0 0 1 -1 0
> 0 0 0 0 0 0 1 -1
>
> A. Main effect of time: repeated measures ANOVA will not accept
> one(1,8); should we do a one sample t-test on the average of the 8 time
> points?
>
> C. An F test, testing for an effect of treatment at any time point?
> repeated measures ANOVA will not accept eye(1,8); any ideas (others than
> 8 x one sample t-tests?)
>
> Many thanks for your advice!
>
> Best wishes,
> Yannis
>
>
>
>
>
> On Wed, 6 Jun 2018 at 11:05, Guillaume Flandin <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
> Dear Yannis,
>
> A comparison of the two approaches is available in this book chapter
> from Rik Henson:
>
> http://www.mrc-cbu.cam.ac.uk/wp-content/uploads/2015/03/Henson_EN_15_ANOVA.pdf
> A key advantage of the partitioned error model here, with one sample
> t-tests, is that you completely eschew the non-sphericity problem.
>
> Best regards,
> Guillaume.
>
>
> On 06/06/18 09:15, Yannis Paloyelis wrote:
> > Dear Guillame,
> >
> > With respect to this 2x2 repeated measures design, I wonder if you
> could
> > explain: is there any advantage (and what is it) of using a
> partitioned
> > errors model with a series of appropriately constructed contrasts and
> > one sample t-tests, versus a flexible factorial (and pool error
> variance)?
> >
> > Many thanks for your advice
> >
> > Yannis
> >
> >
> >
> >
> > On 7 May 2018 at 15:56, Guillaume Flandin <[log in to unmask]
> <mailto:[log in to unmask]>
> > <mailto:[log in to unmask] <mailto:[log in to unmask]>>> wrote:
> >
> > Dear Sophie,
> >
> > You have two within-subjects factors with two levels each here
> so I
> > would recommend to compute a contrast image for each main
> effect and
> > interaction for each subject and enter these in a one-sample
> t-test at
> > the second level, ie at first level, compute:
> > main effect of drug: [1 1 -1 -1]
> > main effect of cardiac: [1 -1 1 -1]
> > drug x cardiac interaction: [1 -1 -1 1]
> >
> > Best regards,
> > Guillaume.
> >
> >
> > On 03/05/18 09:31, Sophie Betka wrote:
> > > Dear Spm users,
> > >
> > >
> > >
> > > Could you let me know if what I did is correct? Thanks a lot
> > >
> > >
> > >
> > > So I am using a 2*2 flexible design (within-subjects design with
> > > Subjects, Drug:2 Levels ; CardiacTiming:2 Levels).
> > >
> > >
> > >
> > >
> > >
> > > 1) I attached a screen shot of my batch. The variance of
> Drug and
> > > Cardiac should be unequal right?
> > >
> > > I am interested in main effect of Drug, main
> effect of
> > > Cardiac and the interaction.
> > >
> > >
> > >
> > > 2) I attached a screen shot of the design matrix.
> > >
> > > Following the Glascher’s paper, I created contrasts.
> > >
> > >
> > >
> > > Could confirm/correct my contrasts:
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > Drug1
> > >
> > >
> > >
> > > Durg2
> > >
> > >
> > >
> > > Card1
> > >
> > >
> > >
> > > Card2
> > >
> > >
> > >
> > > D1C1
> > >
> > >
> > >
> > > D1C2
> > >
> > >
> > >
> > > D2C1
> > >
> > >
> > >
> > > D2C2
> > >
> > > Main effect of Drug
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > -1
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > > Main effect of Cardiac
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > -1
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > > Interaction Drug*Cardiac
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > >
> > >
> > > -0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > > Test single repressor Drug1
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > 0.5
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > > Test single repressor D2C2
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 0
> > >
> > >
> > >
> > > 1
> > >
> > >
> > >
> > > The main effect of cardiac gives me a massive cluster; this
> a bit odd
> > > (enclosed).
> > >
> > >
> > >
> > >
> > >
> > > Thank you for your help,
> > >
> > > S.
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > *Sophie BETKA*
> > > *Ph.D Student*
> > > *Neuropsychologist*
> > >
> > > *Brighton and Sussex Medical School*
> > >
> > > Neurosciences Department - CritchleyLab
> > >
> > > /Trafford Centre, 202 (Brighton, UK)/
> > >
> > > University of Sussex, Falmer, BN1 9RR
> > > +44 (0)1273 873132
> > >
> > > *SCALab - Cognitive and Affective Sciences Laboratory*
> > > /DEEP Team - Emotional Dynamics and Pathology (Lille, FR)/
> > > Faculté de Medecine, Pôle Recherche
> > > 1, Place de Verdun, 59045 Lille cedex
> > >
> > > +33 (0)3.20.62.68.13
> > >
> > > _Twitter @SophieBetka <https://twitter.com/SophieBetka
> > <https://twitter.com/SophieBetka>>___
> > >
> > > _ResearchGate <https://www.researchgate.net/profile/Sophie_Betka
> > <https://www.researchgate.net/profile/Sophie_Betka>>___
> > >
> > > _Linkedin <https://www.linkedin.com/in/sophie-betka-54b149155/
> > <https://www.linkedin.com/in/sophie-betka-54b149155/>>___
> > >
> > >
> > >
> > >
> > >
> > >
> > > ___________________________________________________________
> > > This email has been scanned by MessageLabs' Email Security
> System
> > > on behalf of the Brighton & Sussex Medical School. For more
> > information see:
> > > https://staff.brighton.ac.uk/is/computing/Pages/Email/spam.aspx
> > <https://staff.brighton.ac.uk/is/computing/Pages/Email/spam.aspx>
> >
> > --
> > Guillaume Flandin, PhD
> > Wellcome Centre for Human Neuroimaging
> > University College London
> > 12 Queen Square
> > London WC1N 3BG
> >
> >
>
> --
> Guillaume Flandin, PhD
> Wellcome Centre for Human Neuroimaging
> University College London
> 12 Queen Square
> London WC1N 3BG
>
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|