[FieldTrip] Permutation test design matrix for grand-averaged ERPs

[Eelke Spaak]

Hi Kate,

You can't do permutation statistics if you've already consumed the
degrees of freedom you want to do statistics across (i.e., subjects in
this case).

Did you use cfg.keepindividual = 'yes' in the call to
ft_timelockgrandaverage? If so, then the design matrix should be
specified exactly as in the tutorial; just as if you had been using
the individual subjects' data structures. If not, then indeed you only
have a grand-average left, and no permutation stats can be performed.

Cheers,
Eelke
On Tue, 23 Oct 2018 at 14:04, Kate Stone <katemsto at gmail.com> wrote:
>
> Hi all,
>
> I've been following the tutorial here to do a time-locked perm test on ERP data: www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock
>
> I have computed grand averages for my two conditions and am using indepsamplesT with ft_timelockstatistics(cfg, GA_1, GA_2).
>
> But what should the design matrix be? The example in the tutorial doesn't apply as I am using grand averages and the reference for ft_timelockstatistics doesn't mention the design matrix, although it is definitely required. I have tried design = [1,2], but this doesn't seem to give sensible results.
>
> Thanks in advance,
> Kate
>
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202