Dear Patricia and others that participated to this discussion.<div><br></div><div>I had recently the same problem, using ft_freqstatistics with Montecarlo to correlate Time-frequency power values and behavioral results.</div>
<div>All my p-values were first found to be 0, resulting in a mask full of 1. </div><div>By removing the useless cfg.uvar, the script made the correct analysis and reported good correlation values.</div><div><br></div><div>
Hope this helps, </div><div><br></div><div>Rodolphe.</div><div><br><div class="gmail_quote">On Tue, Feb 1, 2011 at 11:29 AM, Michael Wibral <span dir="ltr"><<a href="mailto:michael.wibral@web.de">michael.wibral@web.de</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">Hi Patricia,<br>
<br>
it seems that things are clarified now. I just do not know what was actually wrong with your code Was it the superfluous specification of cfg.uvar??<br>
Could you let us know? Thanks.<br>
<br>
Michael<br>
-----Ursprüngliche Nachricht-----<br>
Von: "Patricia Wollstadt" <<a href="mailto:Patricia.Wollstadt@gmx.de">Patricia.Wollstadt@gmx.de</a>><br>
Gesendet: Feb 1, 2011 4:34:04 PM<br>
An: "Email discussion list for the FieldTrip project" <<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a>><br>
Betreff: Re: [FieldTrip] ft_freqstatistics and FDR-correction<br>
<br>
>Hi Michael, Hi Jan-Mathijs,<br>
><br>
>thank you very much for the advice/clarification. Everything seems to run fine now (with the corrected design).<br>
><br>
>Thank you very much,<br>
><br>
>Best<br>
><br>
>Patricia<br>
><br>
><br>
>-------- Original-Nachricht --------<br>
>> Datum: Wed, 26 Jan 2011 10:08:51 +0100 (CET)<br>
>> Von: "Michael Wibral" <<a href="mailto:michael.wibral@web.de">michael.wibral@web.de</a>><br>
>> An: "Email discussion list for the FieldTrip project" <<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a>><br>
>> Betreff: Re: [FieldTrip] ft_freqstatistics and FDR-correction<br>
><br>
>> Hi Patricia,<br>
>><br>
>> I also would have thought your design to be correct. Maybe the<br>
>> unneccessary specification of the uvar screws thinsg up??<br>
>><br>
>> However, you really do not have any repeated measures in your design -<br>
>> here the output of FT is correct. I interpret it in the following way: You<br>
>> specified cfg.uvar - indicating to FT the possibility that this might be a<br>
>> repeated measures design. FT identified the units of observation as being 80<br>
>> (correct), and then checked how many repeated measures you ran in each UO<br>
>> and found just 1 - which is also correct. I agree that naming the units of<br>
>> observation (UO) as 'levels' of the UO variable is a bit odd, should rather<br>
>> be 'units of observation'.<br>
>><br>
>> The concept of repeated measures is also very often misunderstood: in<br>
>> repated measures you do not actually repeat anything, rather you perform<br>
>> different measurements (treatments) WITHIN a unit of observation - so a repated<br>
>> measures design is a within UO design. Your subjects just have one age, so<br>
>> there are no different treatments (levels of the indep variable) per<br>
>> subjects and, hence, no 'repeated measures'.<br>
>><br>
>> Hope this clarifies terminology for further discussion.<br>
>><br>
>> Michael<br>
>><br>
>><br>
>> -----Ursprüngliche Nachricht-----<br>
>> Von: "jan-mathijs schoffelen" <<a href="mailto:jan.schoffelen@donders.ru.nl">jan.schoffelen@donders.ru.nl</a>><br>
>> Gesendet: Jan 25, 2011 8:57:03 PM<br>
>> An: "Email discussion list for the FieldTrip project"<br>
>> <<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a>><br>
>> Betreff: Re: [FieldTrip] ft_freqstatistics and FDR-correction<br>
>><br>
>> >Hi Patricia,<br>
>> ><br>
>> >I don't really understand what is going on, but could you try to<br>
>> >specify your design only in a single row (design = design(2,:) in your<br>
>> >case), specify cfg.ivar = 1; and cfg=rmfield(cfg,'uvar'), and see how<br>
>> >it runs? As such the uvar row is not relevant in an independent<br>
>> >samples design.<br>
>> ><br>
>> >Best,<br>
>> ><br>
>> >Jan-Mathijs<br>
>> ><br>
>> ><br>
>> ><br>
>> >On Jan 25, 2011, at 6:08 PM, Patricia Wollstadt wrote:<br>
>> ><br>
>> >> Dear all,<br>
>> >><br>
>> >> I'm still trying to identify the underlying reason for the issue<br>
>> >> described below. I ran the analysis in debug mode (I'm using<br>
>> >> Matlab2008b and the Fieldtrip-release from the 16th of January 2011)<br>
>> >> and found that the matrix 'resample', that is generated in<br>
>> >> 'statistics_montecarlo', line 187:<br>
>> >><br>
>> >> resample = resampledesign(cfg, design);<br>
>> >><br>
>> >> doesn't contain actual permutations of my initial design, but rather<br>
>> >> sorted rowvectors from 1 to 80 (no of subjects). So (since there is<br>
>> >> no permutation of the units of observation) the 'randstat' that is<br>
>> >> subsequently calculated for each row of the resample-matrix is<br>
>> >> always the same.<br>
>> >> Is this maybe due to my design and my units of observations<br>
>> >> respectively? I am looking at age-dependent variations in resting-<br>
>> >> state data, so my UO are 80 subjects, divided into four age groups.<br>
>> >> The commandline-output of the 'resampledesign'-function is as follows:<br>
>> >><br>
>> >> total number of measurements = 80<br>
>> >> total number of variables = 2<br>
>> >> number of independent variables = 1<br>
>> >> number of unit variables = 1<br>
>> >> number of within-cell variables = 0<br>
>> >> number of control variables = 0<br>
>> >> using a permutation resampling approach<br>
>> >> repeated measurement in variable 1 over 80 levels<br>
>> >> number of repeated measurements in each level is 1 1 1 1 1 1 1 1 1 1<br>
>> >> 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1<br>
>> >> 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1<br>
>> >> 1 1<br>
>> >><br>
>> >><br>
>> >> Is this correct? Shouldn't this read 'repeated measurement in<br>
>> >> variable 2 over 4 levels' with the number of subjects per group as<br>
>> >> repeated measurements within a condition? Still I am pretty sure<br>
>> >> about my design matrix, which contains the indices 1 to 80 (for all<br>
>> >> subjects) in the first row (this is my cfg.uvar) with the<br>
>> >> corresponding age groups (1 to 4) in the second row (cfg.ivar, see<br>
>> >> my previous email below). I use an 'indepsamplesF'-statistic, since<br>
>> >> I am comparing between four groups of different subjects (averaged<br>
>> >> over trials).<br>
>> >><br>
>> >> I hope, I got the settings and my design right. Any suggestions,<br>
>> >> whether this is a problem due to wrong configurations in my script<br>
>> >> are very welcome.<br>
>> >><br>
>> >> Right now I work around this issue by replacing line 279 in<br>
>> >> 'statistics_montecarlo':<br>
>> >><br>
>> >> tmpdesign = design(:,resample(i,:));<br>
>> >><br>
>> >> with:<br>
>> >><br>
>> >> tmpdesign = design(:,randperm(size(design,2)));<br>
>> >><br>
>> >> creating permutations of my design 'by hand'. Is this acceptable or<br>
>> >> is there anything I'm maybe missing.<br>
>> >><br>
>> >> Thank you very much, kind regards,<br>
>> >><br>
>> >> Patricia<br>
>> >><br>
>> >><br>
>> >><br>
>> >><br>
>> >> -------- Original-Nachricht --------<br>
>> >>> Datum: Mon, 24 Jan 2011 17:47:26 +0100<br>
>> >>> Von: "Patricia Wollstadt"<br>
>> >>> An: <a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
>> >>> Betreff: [FieldTrip] ft_freqstatistics and FDR-correction<br>
>> >><br>
>> >>> Dear Fieldtrip-users,<br>
>> >>><br>
>> >>> I'm currently trying to calculate freqstatistics for four<br>
>> >>> experimental<br>
>> >>> groups, using the following options:<br>
>> >>><br>
>> >>> groupSize=[length(subjects{1}) length(subjects{2})<br>
>> >>> length(subjects{3})<br>
>> >>> length(subjects{4})];<br>
>> >>><br>
>> >>> design = [1:groupSize(1)+groupSize(2)+groupSize(3)+groupSize(4)];<br>
>> >>> % 1 to N subjects<br>
>> >>> design(2,:) = [ones(1,groupSize(1)) 2*ones(1,groupSize(2))<br>
>> >>> 3*ones(1,groupSize(3)) 4*ones(1,groupSize(4))]; % indep var -><br>
>> >>> group membership<br>
>> >>><br>
>> >>><br>
>> >>> cfg=[];<br>
>> >>> cfg.design=design;<br>
>> >>> cfg.uvar = 1;<br>
>> >>> cfg.ivar = 2;<br>
>> >>> cfg.method = 'montecarlo';<br>
>> >>> cfg.statistic = 'indepsamplesF';<br>
>> >>> cfg.correctm = 'fdr';<br>
>> >>> cfg.numrandomization = 180000;<br>
>> >>> cfg.alpha = 0.01/7;<br>
>> >>> cfg.tail = 1;<br>
>> >>> cfg.design = design;<br>
>> >>> cfg.channels = myChannels;<br>
>> >>> cfg.avgovertime = 'yes';<br>
>> >>> cfg.avgoverfreq = 'yes';<br>
>> >>> cfg.frequency = [1 3];<br>
>> >>><br>
>> >>> stat = ft_freqstatistics(cfg, group1avg, group2avg, group3avg,<br>
>> >>> group4avg);<br>
>> >>><br>
>> >>> With these settings, the coputation itself runs fine, but the<br>
>> >>> resulting<br>
>> >>> 'prob'-matrix contains only zeros (accordingly the 'mask'-matrix<br>
>> >>> consists of<br>
>> >>> only ones). This would mean, that I found significant effects for all<br>
>> >>> channels. Is this possible or rather due to an error somewhere in<br>
>> >>> my script? Is<br>
>> >>> it likely, that these are 'real' zeroes or rather very small<br>
>> >>> values, close<br>
>> >>> to zero?<br>
>> >>> Any hints and suggestions are very welcome, thank you in advance,<br>
>> >>><br>
>> >>> Kind regards<br>
>> >>><br>
>> >>> Patricia Wollstadt<br>
>> >>><br>
>> >>><br>
>> >>> --<br>
>> >>> GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit<br>
>> >>> gratis Handy-Flat! <a href="http://portal.gmx.net/de/go/dsl" target="_blank">http://portal.gmx.net/de/go/dsl</a><br>
>> >>> _______________________________________________<br>
>> >>> fieldtrip mailing list<br>
>> >>> <a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
>> >>> <a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
>> >><br>
>> >> --<br>
>> >> Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir<br>
>> >> belohnen Sie mit bis zu 50,- Euro! <a href="https://freundschaftswerbung.gmx.de" target="_blank">https://freundschaftswerbung.gmx.de</a><br>
>> >> _______________________________________________<br>
>> >> fieldtrip mailing list<br>
>> >> <a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
>> >> <a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
>> ><br>
>> >Dr. J.M. (Jan-Mathijs) Schoffelen<br>
>> >Donders Institute for Brain, Cognition and Behaviour,<br>
>> >Centre for Cognitive Neuroimaging,<br>
>> >Radboud University Nijmegen, The Netherlands<br>
>> ><a href="mailto:J.Schoffelen@donders.ru.nl">J.Schoffelen@donders.ru.nl</a><br>
>> >Telephone: 0031-24-3614793<br>
>> ><br>
>> >_______________________________________________<br>
>> >fieldtrip mailing list<br>
>> ><a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
>> ><a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
><br>
>--<br>
>GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit<br>
>gratis Handy-Flat! <a href="http://portal.gmx.net/de/go/dsl" target="_blank">http://portal.gmx.net/de/go/dsl</a><br>
>_______________________________________________<br>
>fieldtrip mailing list<br>
><a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
><a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>_______________________________________________<br>
fieldtrip mailing list<br>
<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
<a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br></blockquote></div><br></div>