Hi Patricia,

You can include demeaned covariates (like you've done, demeaning over all groups) and that's fine.
It is fine to have more than one - just demean each one separately.

I would not include -D though as your main EVs are modelling the mean.
You should include the -f if you want to see your F-contrast results (which I assume you do!)

The warning message is only about the group column, which you are not using, not you can ignore this.
In general I wouldn't bother setting the group column if you are using randomise, unless you need to use it to specify exchange groups (and you don't have to do that here).

All the best,
Mark



On 24 Jan 2013, at 15:34, Patricia Pires <[log in to unmask]> wrote:

Dear Mark,

is it possible to include a covariable (e.g. age demeaned, gender or both) to my F-tests analysis?

I have done now the following F-test design matrix (attached) as you recommended and I included a new EV (EV5) with age demeaned values. However a warning message appeared (also attached). I don't know if i have to include 4 EV (one for each group) instead of 1 EV and if my contrast are ok or not.

After that i wanted to run randomise: randomise -i all_FA_skeletonised.nii -m mean_FA_skeleton_mask.nii -o resultados_TBSS -d matriz.mat -t matriz.con -f matriz.fts -n 5000 --T2 -V -D

Is this design ok?

What if i want to include both age and gender covariables in the design matrix?

Kind regards,

Patricia.

2013/1/23 Patricia Pires <[log in to unmask]>
Ok, Mark!

Thanks a lot for your help!


Patricia.

2013/1/21 Mark Jenkinson <[log in to unmask]>
Hi,

That looks right - any pairwise difference will be detected with this F-test and so it is looking for any departure from the mean, as an ANOVA would.  The randomise call looks correct to me, and will generate results for both the t-contrasts and the f-contrast.  

All the best,
Mark


On 21 Jan 2013, at 14:02, Patricia Pires <[log in to unmask]> wrote:

Dear Mark,

following your advice, i have designed the following matrix (attached).

Now should i have to run randomise. Is this command correct correct?:

randomise -i all_FA_skeletonised.nii -m mean_FA_skeleton_mask.nii -o resultados_TBSS -d matriz.mat -t matriz.con -f matriz.fts -n 5000 --T2 -V

Then, if there is any result, would it be from an ANOVA with post-hoc, right?

Thank you very much for your help!

Patricia.

2013/1/21 Mark Jenkinson <[log in to unmask]>
Dear Patricia,

Just include 1, 3 and 5 only.
There are other combinations that would be equivalent, but you must restrict it to linearly independent contrasts, and 1, 3 and 5 are the simplest set of those.  Any other contrast you added to them could be generated by a linear combination of these three.  And so that is why I chose these.

All the best,
Mark




On 21 Jan 2013, at 10:00, Patricia Pires <[log in to unmask]>
 wrote:

Hello Mark,

thank you very much for your answer. I am not very sure what do you mean when you say "include one F-test that consisted of t-contrasts 1, 3 and 5." Does it mean that my contrast are allright but i have to include option F-test=1 and mark t contrast 1, 3 and 5? Why only these contrast?

On the other way, should my randomise command be different than:

randomise -i all_FA_skeletonised.nii -m mean_FA_skeleton_mask.nii -o resultados_TBSS -d matriz.mat -t matriz.con -n 5000 --T2 -V

-t option in randomise should now be -f because of the F-test analysis?

Thank you very much,

Patricia.


2013/1/19 Mark Jenkinson <[log in to unmask]>
Hi,

If you want a single F-test that checks for any differences then it would be sufficient to use your existing t-contrasts and include one F-test that consisted of t-contrasts 1, 3 and 5.

All the best,
        Mark



On 17 Jan 2013, at 11:42, SUBSCRIBE FSL Patricia Pires <[log in to unmask]> wrote:

> Dear fsl list,
>
> i have 4 different groups to analize FA index. I performed a Glm matrix:
>
> /Matrix
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 1.000000e+00  0.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  1.000000e+00    0.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    1.000000e+00    0.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
> 0.000000e+00  0.000000e+00    0.000000e+00    1.000000e+00
>
> And their contrasts:
>
> /Matrix
> 1.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00
> -1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00 0.000000e+00 -1.000000e+00
> -1.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00
> 1.000000e+00 -1.000000e+00 0.000000e+00 0.000000e+00
> -1.000000e+00 1.000000e+00 0.000000e+00 0.000000e+00
> 0.000000e+00 0.000000e+00 1.000000e+00 -1.000000e+00
> 0.000000e+00 0.000000e+00 -1.000000e+00 1.000000e+00
> 0.000000e+00 -1.000000e+00 1.000000e+00 0.000000e+00
> 0.000000e+00 1.000000e+00 -1.000000e+00 0.000000e+00
> 0.000000e+00 -1.000000e+00 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00 0.000000e+00 -1.000000e+00
>
>
> There are two contrast with significant results when i run randomise:
>
> randomise -i all_FA_skeletonised.nii -m mean_FA_skeleton_mask.nii -o resultados_TBSS -d matriz.mat -t matriz.con -n 5000 --T2 -V
>
> However this procedure has been criticized because there is not a previous overall of results and i am wondering if i first have to do a F-tests:
>
> http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FEAT/UserGuide#F-Tests
>
> I have to do a F-test first? In that case, what will indicate these results:
>
> - C1: Group A        1  0  0  0
>
> (if i found results here, e.g. cluster --in=resultados_FA_mean_FA_skeleton_mask_tfce_corrp_tstat1.nii.gz --thresh=0.95), would this indicating Group A differ from other groups and then i could finally do my first procedure? Or it is not necessary this first F-test analysis?
>
> Thanks in advance,
>
> Patricia.



<GlmANOVAMark.png>



<WarningANOVACovariable.png><WarningANOVACovariavle2.png>