[FieldTrip] Problem with ft_timelockstatistics, within subject comparison of two conditions (Clara A. Scholl)
Clara A. Scholl
clara.scholl at gmail.com
Wed Oct 8 17:21:12 CEST 2014
Dear Barbara,
I think you are right, the clustering output should report total number of
variables=2 (and the total # of measurements = 185 trials, number of
independent variables=1, and number of unit variables=1).
I think your design matrix should have two rows, rather than 1. For
example, if you had 5 trials total (2 of 1 trial type, 3 of the other trial
type), you'd specify the design matrix:
1 2 3 1 2
1 1 1 2 2
Where cfg.uvar=1 and cfg.ivar=2. Row two is what you are specifying for
design currently. Row 1 counts the observations of each trials
(1:numTrialType1 1:numTrialType2). Does that help?
Respectfully,
Clara
On Wed, Oct 8, 2014 at 7:31 AM, <barbara.schorr at uni-ulm.de> wrote:
> Dear Clara,
>
> thank you for your comments.
>
> I ran it as an independent samples Test (it is true, that this is the
> correct way).
> I also changed the design matrix to
> cfg.design = [ones(1,size(targetclean_tlt.trial,1))
> 2*ones(1,size(standardclean_tlt.trial,1))]
>
> It works and there are no more error messages. However, when it displays
> the progress of the calculation there is the field: total number of
> observations, total number of variables and number of independent
> variables. It recognizes correctly that there are 185 trials in total (both
> conditions together). But it only counts 1 variable and 1 independent
> variable. Shouldn't it recognize 2 variables as I have 2 conditions?
>
> Best regards,
> Barbara
>
>
> Zitat von fieldtrip-request at science.ru.nl:
>
> Send fieldtrip mailing list submissions to
>> fieldtrip at science.ru.nl
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>> or, via email, send a message with subject or body 'help' to
>> fieldtrip-request at science.ru.nl
>>
>> You can reach the person managing the list at
>> fieldtrip-owner at science.ru.nl
>>
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of fieldtrip digest..."
>>
>>
>> Today's Topics:
>>
>> 1. Problem with ft_timelockstatistics, within subject comparison
>> of two conditions (barbara.schorr at uni-ulm.de)
>> 2. Re: Problem with ft_timelockstatistics, within subject
>> comparison of two conditions (Clara A. Scholl)
>> 3. Re: help for statistics on ERD/ERS (Tzvetan Popov)
>> 4. Re: help for statistics on ERD/ERS (Giuly)
>>
>>
>> ----------------------------------------------------------------------
>>
>> Message: 1
>> Date: Tue, 07 Oct 2014 15:42:53 +0200
>> From: barbara.schorr at uni-ulm.de
>> To: fieldtrip at science.ru.nl
>> Subject: [FieldTrip] Problem with ft_timelockstatistics, within
>> subject comparison of two conditions
>> Message-ID: <20141007154253.1acx2qfvso4k4so4 at imap.uni-ulm.de>
>> Content-Type: text/plain; charset=ISO-8859-1; DelSp="Yes";
>> format="flowed"
>>
>> Dear fieldtripers,
>>
>> I would like to compare to conditions of an auditory oddball paradigm
>> (target tones and standard tones).
>> For the identification of the electrodes which show the biggest P300,
>> I would like to use cluster based permutation test, i.e. I want to use
>> ft_timelockstatistics, to identify the clusters with a significant
>> difference between the target and the standardtones.
>> Important: I don't want to do a group statistic, but I want to compare
>> these two conditions within each subject, to identify interesting
>> clusters for each subject separately.
>>
>>
>> I wrote the script following the tutorials on the fieldtrip homepage.
>>
>> I already wrote a code but it does not work:
>> %%
>>
>> for i = 3:10 % subject number
>> mypfad = 'G:\BackUp07.07.14_DatenPrepr\Daten1\P300'
>> cd(mypfad)
>> cd Kontrollen;
>> List = dir
>> cd (List(i).name);
>> for j = 1:4 %session number
>> cd (mypfad);
>> cd Kontrollen;
>> List = dir
>> cd (List(i).name);
>> eval(['cd M' num2str(j)]);
>> load cleandata_tlt; %preprocessed timelocked data
>> standardclean_tlt and targetclean_tlt
>>
>> cfg = []
>> cfg.neighbourdist = .09;
>> cfg.layout= ulmegilay;
>> cfg.method = 'distance';
>>
>> Neighbours = ft_prepare_neighbours(cfg, standardclean_tlt)
>>
>> cfg = []
>> cfg.method = 'montecarlo' ; %significance probability
>> cfg.statistic = 'ft_statfun_depsamplesT'
>> cfg.correctm = 'cluster'
>> cfg.clusteralpha = 0.05 %alpha level of the sample specific
>> test statistic that will
>> % be used for thresholding
>> cfg.clusterstatistic = 'maxsum'
>> cfg.minnbchan = 4
>> cfg.latency = [0.25 0.8]
>> cfg.neighbours = Neighbours
>> cfg.tail = 0
>> cfg.clustertail = 0
>> cfg.alpha = 0.025
>> cfg.numrandomization = 1000
>>
>> design = [ones(size(targetclean_tlt.trial,1),1);
>> 2*ones(size(standardclean_tlt.trial,1),1)]';
>> design([1:size(targetclean_tlt.trial)
>> 1:size(standardclean_tlt.trial,1)])
>>
>>
>> cfg.design = design
>> cfg.ivar = [ones(size(targetclean_tlt.trial,1),1);
>> 2*ones(size(standardclean_tlt.trial,1),1)]'
>> cfg.uvar = [1:size(targetclean_tlt.trial)
>> 1:size(standardclean_tlt.trial,1)]
>>
>> tlt_statsnew = ft_timelockstatistics(cfg, targetclean_tlt,
>> standardclean_tlt )
>>
>> cd 'G:\BackUp07.07.14_DatenPrepr\Daten1\P300';
>> cd Kontrollen;
>> List2 = dir
>> cd (List2(i).name);
>> eval(['cd M' num2str(j)]);
>> save statnew tlt_statsnew
>>
>> end
>> end
>>
>>
>> This is what gets displayed while running the script:
>>
>> selected 245 channels
>> selected 139 time bins
>> selected 1 frequency bins
>> using "ft_statistics_montecarlo" for the statistical testing
>> using "ft_statfun_depsamplesT" for the single-sample statistics
>> constructing randomized design
>> total number of measurements = 189
>> total number of variables = 1
>> number of independent variables = 189
>> number of unit variables = 189
>> number of within-cell variables = 0
>> number of control variables = 0
>> using a permutation resampling approach
>>
>> Then, this error occurs:
>>
>> Error using resampledesign (line 168)
>> A within-units shuffling requires a at least one unit variable and at
>> least
>> one independent variable
>>
>> Error in ft_statistics_montecarlo (line 241)
>> resample = resampledesign(cfg, design);
>>
>> Error in statistics_wrapper (line 310)
>> [stat, cfg] = statmethod(cfg, dat, design);
>>
>> Error in ft_timelockstatistics (line 113)
>> [stat, cfg] = statistics_wrapper(cfg, varargin{:});
>>
>> I have no clue where I made a mistake. I have unit variables and
>> independent variables (189 if I understand it correctly).
>> Is there anything I am missing? Any additional information or is the
>> design matrix wrong?
>>
>>
>> I really appreciate your help.
>>
>> Best regards,
>> Barbara
>>
>>
>>
>>
>> Barbara Schorr, MSc
>> Clinical and Biological Psychology
>> University of Ulm
>> Albert-Einstein-Allee 47
>> 89069 Ulm
>>
>>
>>
>>
>>
>> ------------------------------
>>
>> Message: 2
>> Date: Tue, 7 Oct 2014 09:55:36 -0400
>> From: "Clara A. Scholl" <clara.scholl at gmail.com>
>> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
>> Subject: Re: [FieldTrip] Problem with ft_timelockstatistics, within
>> subject comparison of two conditions
>> Message-ID:
>> <CAAHOyr7MU6npw9U4Wqkoo7OXxthQmsTrjBroWcxLe6qTeSkC=g at mail.
>> gmail.com>
>> Content-Type: text/plain; charset="utf-8"
>>
>> Dear Barbara,
>>
>> Just three thoughts:
>>
>> 1) Should cfg.ivar and cfg.uvar both be single values, i.e. the dimension
>> of the design matrix that contains the independent and unit variables?
>> Then you'd have cfg.ivar=1 and cfg.uvar=2 (or vise versa, depending on the
>> design matrix).
>>
>> 2) I think the design matrix itself might be off, right now it just seems
>> to have dimensions of 1 x (# targets+standards). It should have a second
>> dimension which counts the # of trials, something like:
>>
>> design(1,:) = [ones(size(targetclean_tlt.trial,1),1);
>> 2*ones(size(standardclean_tlt.trial,1),1)]';
>> design(2,:)= ([1:size(targetclean_tlt.trial)
>> 1:size(standardclean_tlt.trial
>> ,1)])
>>
>> 3) Since this is a design where each trial is an observation, I think you
>> should be using indepsamplesT for cfg.statistic (the trials aren't paired
>> and there aren't necessarily identical #'s of trials for each condition.
>>
>> What do you think?
>> Respectfully,
>> Clara
>>
>>
>> On Tue, Oct 7, 2014 at 9:42 AM, <barbara.schorr at uni-ulm.de> wrote:
>>
>> Dear fieldtripers,
>>>
>>> I would like to compare to conditions of an auditory oddball paradigm
>>> (target tones and standard tones).
>>> For the identification of the electrodes which show the biggest P300, I
>>> would like to use cluster based permutation test, i.e. I want to use
>>> ft_timelockstatistics, to identify the clusters with a significant
>>> difference between the target and the standardtones.
>>> Important: I don't want to do a group statistic, but I want to compare
>>> these two conditions within each subject, to identify interesting
>>> clusters
>>> for each subject separately.
>>>
>>>
>>> I wrote the script following the tutorials on the fieldtrip homepage.
>>>
>>> I already wrote a code but it does not work:
>>> %%
>>>
>>> for i = 3:10 % subject number
>>> mypfad = 'G:\BackUp07.07.14_DatenPrepr\Daten1\P300'
>>> cd(mypfad)
>>> cd Kontrollen;
>>> List = dir
>>> cd (List(i).name);
>>> for j = 1:4 %session number
>>> cd (mypfad);
>>> cd Kontrollen;
>>> List = dir
>>> cd (List(i).name);
>>> eval(['cd M' num2str(j)]);
>>> load cleandata_tlt; %preprocessed timelocked data
>>> standardclean_tlt and targetclean_tlt
>>>
>>> cfg = []
>>> cfg.neighbourdist = .09;
>>> cfg.layout= ulmegilay;
>>> cfg.method = 'distance';
>>>
>>> Neighbours = ft_prepare_neighbours(cfg, standardclean_tlt)
>>>
>>> cfg = []
>>> cfg.method = 'montecarlo' ; %significance probability
>>> cfg.statistic = 'ft_statfun_depsamplesT'
>>> cfg.correctm = 'cluster'
>>> cfg.clusteralpha = 0.05 %alpha level of the sample specific test
>>> statistic that will
>>> % be used for thresholding
>>> cfg.clusterstatistic = 'maxsum'
>>> cfg.minnbchan = 4
>>> cfg.latency = [0.25 0.8]
>>> cfg.neighbours = Neighbours
>>> cfg.tail = 0
>>> cfg.clustertail = 0
>>> cfg.alpha = 0.025
>>> cfg.numrandomization = 1000
>>>
>>> design = [ones(size(targetclean_tlt.trial,1),1);
>>> 2*ones(size(standardclean_tlt.trial,1),1)]';
>>> design([1:size(targetclean_tlt.trial) 1:size(standardclean_tlt.
>>> trial,1)])
>>>
>>>
>>> cfg.design = design
>>> cfg.ivar = [ones(size(targetclean_tlt.trial,1),1);
>>> 2*ones(size(standardclean_tlt.trial,1),1)]'
>>> cfg.uvar = [1:size(targetclean_tlt.trial)
>>> 1:size(standardclean_tlt.
>>> trial,1)]
>>>
>>> tlt_statsnew = ft_timelockstatistics(cfg, targetclean_tlt,
>>> standardclean_tlt )
>>>
>>> cd 'G:\BackUp07.07.14_DatenPrepr\Daten1\P300';
>>> cd Kontrollen;
>>> List2 = dir
>>> cd (List2(i).name);
>>> eval(['cd M' num2str(j)]);
>>> save statnew tlt_statsnew
>>>
>>> end
>>> end
>>>
>>>
>>> This is what gets displayed while running the script:
>>>
>>> selected 245 channels
>>> selected 139 time bins
>>> selected 1 frequency bins
>>> using "ft_statistics_montecarlo" for the statistical testing
>>> using "ft_statfun_depsamplesT" for the single-sample statistics
>>> constructing randomized design
>>> total number of measurements = 189
>>> total number of variables = 1
>>> number of independent variables = 189
>>> number of unit variables = 189
>>> number of within-cell variables = 0
>>> number of control variables = 0
>>> using a permutation resampling approach
>>>
>>> Then, this error occurs:
>>>
>>> Error using resampledesign (line 168)
>>> A within-units shuffling requires a at least one unit variable and at
>>> least
>>> one independent variable
>>>
>>> Error in ft_statistics_montecarlo (line 241)
>>> resample = resampledesign(cfg, design);
>>>
>>> Error in statistics_wrapper (line 310)
>>> [stat, cfg] = statmethod(cfg, dat, design);
>>>
>>> Error in ft_timelockstatistics (line 113)
>>> [stat, cfg] = statistics_wrapper(cfg, varargin{:});
>>>
>>> I have no clue where I made a mistake. I have unit variables and
>>> independent variables (189 if I understand it correctly).
>>> Is there anything I am missing? Any additional information or is the
>>> design matrix wrong?
>>>
>>>
>>> I really appreciate your help.
>>>
>>> Best regards,
>>> Barbara
>>>
>>>
>>>
>>>
>>> Barbara Schorr, MSc
>>> Clinical and Biological Psychology
>>> University of Ulm
>>> Albert-Einstein-Allee 47
>>> 89069 Ulm
>>>
>>>
>>>
>>> _______________________________________________
>>> 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/20141007/64be64b4/attachment-0001.html>
>>
>> ------------------------------
>>
>> Message: 3
>> Date: Tue, 7 Oct 2014 21:07:37 +0200
>> From: Tzvetan Popov <tzvetan.popov at uni-konstanz.de>
>> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
>> Subject: Re: [FieldTrip] help for statistics on ERD/ERS
>> Message-ID: <D25CE622-A81C-493D-8905-8536BB51023C at uni-konstanz.de>
>> Content-Type: text/plain; charset=windows-1252
>>
>> Hi Giulia,
>>
>>
>> Dear Fieldtrippers
>>> I'm writing to ask you some help on data in a single subject study.
>>> I did a time frequency decomposition and I would like to do a statistic
>>> to state whether the average TFR has significant ERD/ERS or not.
>>> I'm not sure how to proceed
>>>
>> What you are referring to is a ?between trial experiment? which is
>> described in this tutorial:
>> http://fieldtrip.fcdonders.nl/tutorial/cluster_permutation_freq
>> One possible way to go is to compute the TFR?s with cfg.keeptrials =
>> ?yes?. Next you?d treat the baseline and the task interval as two
>> conditions by separating them using ft_selectdata. After this you can
>> compute the statistical comparison in a similar fashion as described in
>> the tutorial.
>> Keep in mind that you have to cut the data into equal lengths. Please
>> have a look at this post :
>> http://mailman.science.ru.nl/pipermail/fieldtrip/2007-July/001303.html
>>
>>> , as a first step I compared the single trial baseline with the
>>> respective event, with a t-test for each condition and band of interest
>>> but I'm not really sure if this is enough and anyway if it is the correct
>>> way.
>>>
>> Also please type ?actvsblT? in the search bar and then click on ?Search
>> the FieldTrip mailing list?. There you?ll find some info about this.
>>
>> best
>> tzvetan
>>
>>
>>
>>
>> ------------------------------
>>
>> Message: 4
>> Date: Wed, 08 Oct 2014 10:45:23 +0200
>> From: Giuly <giulia.rizza at tiscali.it>
>> To: FieldTrip discussion list <fieldtrip at science.ru.nl>
>> Subject: Re: [FieldTrip] help for statistics on ERD/ERS
>> Message-ID: <eeebb8a2bf90ed8099aea3a3e21a7a38 at tiscali.it>
>> Content-Type: text/plain; charset="utf-8"
>>
>> Hi Tzvetan
>> Thank you so much for your reply
>> I will try these
>> functions keeping in mind your explanation
>> Best regards
>> Giulia
>>
>> Il
>> 07.10.2014 21:07 Tzvetan Popov ha scritto:
>>
>> Hi Giulia,
>>>
>>> Dear
>>>>
>>> Fieldtrippers I'm writing to ask you some help on data in a single
>> subject study. I did a time frequency decomposition and I would like to
>> do a statistic to state whether the average TFR has significant ERD/ERS
>> or not. I'm not sure how to proceed
>>
>>>
>>> What you are referring to is a
>>>
>> "between trial experiment" which is described in this tutorial:
>>
>>>
>>> http://fieldtrip.fcdonders.nl/tutorial/cluster_permutation_freq [1]
>>
>>>
>>> One possible way to go is to compute the TFR's with cfg.keeptrials =
>> 'yes'. Next you'd treat the baseline and the task interval as two
>> conditions by separating them using ft_selectdata. After this you can
>> compute the statistical comparison in a similar fashion as described in
>> the tutorial.
>>
>>> Keep in mind that you have to cut the data into equal
>>>
>> lengths. Please have a look at this post :
>>
>>>
>>> http://mailman.science.ru.nl/pipermail/fieldtrip/2007-July/001303.html
>> [2]
>>
>>>
>>> , as a first step I compared the single trial baseline with
>>>>
>>> the respective event, with a t-test for each condition and band of
>> interest but I'm not really sure if this is enough and anyway if it is
>> the correct way.
>>
>>>
>>> Also please type "actvsblT" in the search bar and
>>>
>> then click on "Search the FieldTrip mailing list". There you'll find
>> some info about this.
>>
>>>
>>> best
>>> tzvetan
>>>
>>>
>>> _______________________________________________
>>
>>> fieldtrip mailing
>>>
>> list
>>
>>> fieldtrip at donders.ru.nl [3]
>>>
>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip [4]
>>
>>
>>
>>
>> Scopri istella, il nuovo motore per il web italiano.
>> Istella garantisce risultati di qualit? e la possibilit? di condividere,
>> in modo semplice e veloce, documenti, immagini, audio e video.
>> Usa istella, vai su http://www.istella.it?wtk=amc138614816829636
>>
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/
>> attachments/20141008/ff2ad283/attachment-0001.html>
>>
>> ------------------------------
>>
>> _______________________________________________
>> fieldtrip mailing list
>> fieldtrip at donders.ru.nl
>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>
>> End of fieldtrip Digest, Vol 47, Issue 6
>> ****************************************
>>
>>
>
>
> Barbara Schorr, MSc
> Clinical and Biological Psychology
> University of Ulm
> Albert-Einstein-Allee 47
> 89069 Ulm
>
>
>
> _______________________________________________
> 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/20141008/89c3a1ce/attachment-0002.html>
More information about the fieldtrip
mailing list