[FieldTrip] freqstatistics with sbj_chan_freq (Vit?ria Piai)

[Vitória Piai]

Thanks a lot, Haiteng, but that was not really the issue. I have a 
specific interval I'm interested in and by using dpss over that 
interval, my frequency resolution is better than if I use TFRs and then 
average.

The problem really was in my cfg for the stats and I've just found out 
what it was. Don't know why (stupid me!!), but I set cfg.tail = 1 and 
that of course will only give me the right tail (posclusters!!)

Thanks anyways, Vitória



On 4/13/2014 11:08 PM, Haiteng Jiang wrote:
> Hi Viotira,
>    If you don't care about time information , you can average  it by 
> specifying cfg.avgovertime = 'yes' on the TFR , then the cluster will 
> be formed in channel-frequency format.  If you use  mtmfft , you 
> should pre-select effect interval .   Otherwise,   it could be not 
> consistent .
>            Best,
>            Haiteng
>
>
>
>
>     Message: 1
>     Date: Sun, 13 Apr 2014 18:53:10 -0700
>     From: Vit?ria Piai <v.piai.research at gmail.com
>     <mailto:v.piai.research at gmail.com>>
>     To: FieldTrip discussion list <fieldtrip at science.ru.nl
>     <mailto:fieldtrip at science.ru.nl>>
>     Subject: [FieldTrip] freqstatistics with sbj_chan_freq
>     Message-ID: <534B3F86.9070904 at gmail.com
>     <mailto:534B3F86.9070904 at gmail.com>>
>     Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
>
>     Hi all,
>
>     I'm kinda stuck not being able to figure out what's going wrong
>     with my
>     freqstats, so I thought I could share it here and see if anyone
>     can see
>     where I'm going wrong.
>     When I run freqstatistics on TFRs, I get a negative cluster (p = .001)
>     between 8 and 25 Hz with the topography very similar to the one in the
>     figure. But I just want non-time-resolved spectra.
>     So I have power spectra calculated with mtmfft (dpss tapers) between 1
>     and 50 Hz (5 Hz smoothing) and this is the one for which I've been
>     trying to run freqstatistics, unsuccessfully.
>     I'm using a pretty standard cfg for freqstats (basically, the same
>     I use
>     for the TFRs, which give me the negative cluster with p =.001) and the
>     data in the comparison are structured as sbj_chan_freq. With this
>     time-unresolved data, only posclusters are detected, with ps > .4!
>     If I
>     plot the stat output, it has the exact same negative cluster as I get
>     for the TFRs (see Figure), and as far as I can see, no real positive
>     cluster, but the function doesn't even detect any negative clusters.
>     I must be doing something wrong because the time-resolved and
>     -unresolved effects are pretty much the same in the descriptives. I
>     remember running stats on time-unresolved data before and getting a
>     reasonable output, so I guess my ft_freqstatistics isn't working well
>     right now. I just can't figure out why.
>     Does anyone have any tips for how I could go about to find out why the
>     negative clusters are being disregarded?
>
>     Thanks a lot, Vit?ria
>
> -- 
> Haiteng Jiang
> PhD candidate
> Neuronal Oscillations Group
> Donders Institute for Brain, Cognition and Behaviour
> Centre for Cognitive Neuroimaging
> Radboud University Nijmegen
>
> Visiting address
> Room 2.32
> Donders Centre for Cognitive Neuroimaging
> Kapittelweg 29
> 6525 EN  Nijmegen
> the Netherlands
>
>  Tel.: +31 (0)243668291
>  Web: https://sites.google.com/site/haitengjiang/
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140414/029ea2ea/attachment-0002.html>