[log in to unmask]" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite">

Hi

On 24 Nov 2010, at 10:14, Gabriel Gonzalez Escamilla wrote:

Thanks Stephen

just to be sure, I've run a one-way ANOVA for this design with 3 contrasts and one F:
c1: 1 0 0 [and zeros for covs]
c2: 0 1 0 [and zeros for covs]
c3: 0 0 1 [and zeros for covs]
and F: 1 1 1

I think this is ok - though also see the example at

http://fsl.fmrib.ox.ac.uk/fsl/feat5/detail.html#ANOVA1factor4levels

is it correct? for meassuring any difference on the 3 groups???
and if a would perform paired differences, would be:
1 -1 0 [and zeros for covs]
1  0 -1 [and zeros for covs]
0 1 -1 [and zeros for covs]

Sure - though if you really mean *paired* differences then presumably you also need one EV per subject to model out the subject means - again see the various FEAT examples….

Cheers.

El 24/11/10, Stephen Smith <[log in to unmask]">[log in to unmask]> escribió:

[log in to unmask]" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0pt;" type="cite">

Hi - in general you should demean the additional covariates, including the sex.

Cheers.

On 23 Nov 2010, at 15:44, Gabriel Gonzalez Escamilla wrote:

I've introduce a regressor of no interest de-meaned but i have another one the sex wich has not been de-meaned

Group  EV1     EV2     EV3     EV4     EV5     EV6     EV7     EV8     EV9

1          1          0          0          1.5       0          0          1          0          0

1          1          0          0          -6.4     0          0          1          0          0

1          1          0          0          2.8       0          0          2          0          0

1          1          0          0          -6.4     0          0          2          0          0         

2          0          1          0          0          1.4       0          0          2          0         

2          0          1          0          0          1.3       0          0          1          0         

2          0          1          0          0          -1.7     0          0          1          0         

2          0          1          0          0          1.4       0          0          2          0

3          0          0          1          0          0          -6.4     0          0          1         

3          0          0          1          0          0          1.2       0          0          2         

3          0          0          1          0          0          -1.6     0          0          1         

3          0          0          1          0          0          -2.4     0          0          2

This is an example of the design matrix, I have 3 groups and I'd insert the age(z-scored) and sex per group.
 

Bests

Gabriel

El 23/11/10, David Flitney <[log in to unmask]">[log in to unmask]> escribió:

[log in to unmask]" class="iwcQuote" style="border-left: 1px solid rgb(0, 0, 255); padding-left: 13px; margin-left: 0px;" type="cite">

Hi

This means that one, or more, of your regressors are already modelling the mean. Please check and if you're in any doubt then post your design so we can double check it.

Dave

On 23 Nov 2010, at 12:02, Gabriel Gonzalez Escamilla wrote:

Dear FSL users

I'm running randomise, and when i add the -D flag warning message appears:

"Warning: You have demeaned your data, but at least one design column has non-zero mean"

Does this message tells me that is something wrong in my design, or it just appears?

Regards

Gabriel

-- 

Dave Flitney, IT Manager

Oxford Centre for Functional MRI of the Brain

E:[log in to unmask]">[log in to unmask] W:+44-1865-222713 F:+44-1865-222717

URL: http://www.fmrib.ox.ac.uk/~flitney

--

Gabriel González Escamilla
Laboratory of Functional Neuroscience
Department of Physiology, Anatomy, and Cell Biology
University Pablo de Olavide
Ctra. de Utrera, Km.1
41013 - Sevilla
- Spain -

Email: [log in to unmask]">[log in to unmask]
http://www.upo.es/neuroaging/es/

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]">[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------

--
Gabriel González Escamilla
Laboratory of Functional Neuroscience
Department of Physiology, Anatomy, and Cell Biology
University Pablo de Olavide
Ctra. de Utrera, Km.1
41013 - Sevilla
- Spain -

Email: [log in to unmask]">[log in to unmask]
http://www.upo.es/neuroaging/es/

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]">[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------