Hi,
1) The order doesn't matter as long as the order of the entries in the EVs match with the order of the input files. So what you've done is OK.
2) You never _need_ to use the group column. It is an option to allow you to model separate variances for the different groups, and we would normally only recommend using it if you have a reasonably large number of subjects in each group, as otherwise the variance is poorly estimated and this makes it better to only have a single group and a better estimate of the variance.
3) Randomise swaps rows in the design matrix, which is equivalent to swapping subjects between groups.
All the best,
Mark
On 20 Nov 2012, at 09:24, Gabor Perlaki <[log in to unmask]> wrote:
> Dear all,
>
> I've two groups, 8 controls and 8 patients. I've an EV for each group ( EV1 for controls and EV2 for patients).
> My questions are:
>
> 1, Is it important that the first 8 subjects shoould be controls and the second 8 subjects should be patients, or the order doesn't matter? (I think order doesn't matter)
>
> for example the order is: contr1 contr2 pat1 pat2 contr3 pat3 contr4 contr5 contr6 contr7 contr8 pat4 pat5 pat6 pat7 pat8
>
> and the EVs
>
> EV1: EV2:
> 1 0
> 1 0
> 0 1
> 0 1
> 1 0
> 0 1
> 1 0
> 1 0
> 1 0
> 1 0
> 1 0
> 0 1
> 0 1
> 0 1
> 0 1
> 0 1
>
> Contrast file:
> 1 -1
> -1 1
>
>
> Is it OK?
>
> 2, Do I need to use *.grp in this case? (I think I shouldn't)
>
> 3, How randomise works in this case? Does it interchange one or more subjects between the two groups, so interchange zero and one in one ore more rows randomly, or how to imagine?
>
> Thanks a lot,
> Gabor
>
|