[FieldTrip] Quick question about depsamplesregrT

[Eelke Spaak]

Dear Kyoungeun,

If your code is exactly as you posted, specifically these lines:

design(2,1:subj) = 1; design(2,subj+1:2*subj)= 2; design(2,2*subj+1:3*subj)= 3;
...
stat = ft_freqstatistics(cfg, High, Low, No);

then yes, indeed, you are correct (and my last sentence was wrong,
sorry). I'm assuming you are dealing with an ordinal variable (or
discrete numeric) coded as No/Low/High. It might help if you swap
around the order of variables like so:

stat = ft_freqstatistics(cfg, No, Low, High);

which would result in perhaps more easily interpretable directions of
effects (avoiding "double negatives").

Another brief word of caution (which you may already be aware of, but
just to be sure): even if you find a significant positive/negative
cluster, this of course does not necessarily mean that high > low AND
low > no (or vice versa, or even one of those). It just means there is
a positive/negative trend, which could be caused by high > no and low
== high or low == no, or something like that. To make the more
specific tests of high > low etc.,would require pairwise comparisons
(e.g. depsamplesT).

Best,
Eelke

On Wed, 8 Jul 2020 at 00:06, Lee, Kyoung Eun <klee773 at gatech.edu> wrote:
>
> Dear Eelke,
>
> Thank you so much for your kind reply.
> If I understood your comment correctly, then a positive cluster means high(1)<low(2)<no(3) and a negative cluster means high(1)>low(2)>no(3), is it right?
> Thank you again!
>
> Best,
> Kyoungeun
>
> -----Original Message-----
> From: fieldtrip <fieldtrip-bounces at science.ru.nl> On Behalf Of Eelke Spaak
> Sent: Thursday, July 2, 2020 11:34 AM
> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
> Subject: Re: [FieldTrip] Quick question about depsamplesregrT
>
> Dear Lee Kyoung Eun,
>
> ft_statfun_depsamplesregrT assigns no special meaning to the values of your regressor, i.e. it uses them exactly as you put in. So, a positive t-statistic reflects a positive regression coefficient, which means the dependent variable is positively associated with the regressor. In your case, if high=1, low=2, no=3, then a negative cluster means no>low>high.
>
> Best,
> Eelke
>
> On Wed, 1 Jul 2020 at 12:08, Lee, Kyoung Eun <klee773 at gatech.edu> wrote:
> >
> > Dear FieldTrip users,
> >
> >
> >
> > I have a simple quick question.
> >
> > Those are my parameters for the clustering permutation test.
> >
> >
> >
> > cfg.latency          = 'all';
> >
> > cfg.channel          = 'all';
> >
> > cfg.frequency        = 'all';
> >
> > cfg.avgovertime      = 'no';
> >
> > cfg.avgoverchan      = 'no';
> >
> > cfg.avgoverfreq      = 'no';
> >
> > cfg.method           = 'montecarlo';
> >
> > cfg.statistic        = 'depsamplesregrT';
> >
> > cfg.correctm         = 'cluster';
> >
> > cfg.clusteralpha     = 0.05;
> >
> > cfg.clusterstatistic = 'maxsum';
> >
> > cfg.minnbchan        = 1;
> >
> > cfg.tail             = 0;
> >
> > cfg.clustertail      = 0;
> >
> > cfg.alpha            = 0.025;
> >
> > cfg.numrandomization = 1000;
> >
> > cfg.parameter        = 'm';
> >
> >
> >
> > And this is my design matrix. I only have a within-subject factor (high, low, no).
> >
> >
> >
> > subj = 42;
> >
> > design(1,1:subj) = 1:subj; design(1,subj+1:2*subj)= 1:subj;
> > design(1,2*subj+1:3*subj)= 1:subj;
> >
> > design(2,1:subj) = 1; design(2,subj+1:2*subj)= 2;
> > design(2,2*subj+1:3*subj)= 3;
> >
> >
> >
> > cfg.design = design;
> >
> > cfg.ivar = 2;
> >
> > cfg.uvar = 1;
> >
> >
> >
> > Then, I ran the freqstatistics like this.
> >
> >
> >
> > stat = ft_freqstatistics(cfg, High, Low, No);
> >
> >
> >
> >
> >
> > If I found a significant negative cluster, does it mean there is a linear relationship of “high > low > no” or “no> low>high”?
> >
> > I always confused about the direction of the effect.
> >
> >
> >
> > Thank you so much!
> >
> > _______________________________________________
> > fieldtrip mailing list
> > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > https://doi.org/10.1371/journal.pcbi.1002202
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202