depsamplesF

Eric Maris e.maris at DONDERS.RU.NL
Tue Sep 8 11:20:31 CEST 2009

```Hi Nadia,

Statfun_depsamplesF calculates a Hotelling's T^2-statistic. This statistic
has a scaled F-distribution under the null hypothesis that the levels of the
within-subjects independent variable have identical means.

When programming this statfun, I used the book by Johnson & Wichern, Applied
Multivariate Statistical Analysis, which I can advise to everyone who is not
afraid of matrices. The statistic that is (or should be ...) implemented is
on the top of p. 218 (edition 1988 of the book).

Can you check the code?

I know that Tineke Snijders (Donders PhD student) has also used this statfun
(actually, I programmed it especially for her) and did not obtain strange
results.

Good luck,

Eric Maris

> -----Oorspronkelijk bericht-----
> Van: FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL] Namens
> MÃ¼ller
> Verzonden: vrijdag 4 september 2009 15:50
> Aan: FIELDTRIP at NIC.SURFNET.NL
> Onderwerp: [FIELDTRIP] depsamplesF
>
> Dear all,
>
> I've a question regarding the statfun_depsamplesF.
>
> I'd like to compare source data of 9 subjects x 5 conditions.
> Therefore I performed a repeated dependent samples F-statistic on
> source level. When looking at the data I was surprised by the very
> high F-values and, at the same time, very low p-values.  So we
> debugged the statfun to get an idea of how the F-statistic is
> calculated. Thereby, we came across the 'scaledstat' where the F-
> values are dramatically reduced, resulting in very low F-values and
> consequently p-values. We also got different degrees of freedom
> compared to an anova calculated elsewhere.
>
> It would be great if you could give me some background of how this F-
> statistic is performed and why for example we need the scaledstat.
>
> Thanks a lot,
>
> ----------------------------------
> The aim of this list is to facilitate the discussion between users of the
FieldTrip
> toolbox, to share experiences and to discuss new ideas for MEG and EEG
analysis.